Matthew B.
Prendergast
a,
Benjamin B.
Kirk
b,
John D.
Savee
c,
David L.
Osborn
c,
Craig A.
Taatjes
c,
Kye-Simeon
Masters
d,
Stephen J.
Blanksby
e,
Gabriel
da Silva
f and
Adam J.
Trevitt
*a
aSchool of Chemistry, University of Wollongong, Wollongong, NSW 2522, Australia. E-mail: adamt@uow.edu.au
bLawrence Berkeley National Laboratory, Berkeley, CA 94720, USA
cCombustion Research Facility, Sandia National Laboratories, Livermore, CA 94551-0969, USA
dSchool of Chemistry, Physics and Mechanical Engineering, Faculty of Science and Engineering, Queensland University of Technology, Brisbane, QLD 4001, Australia
eCentral Analytical Research Facility, Queensland University of Technology, Brisbane, QLD 4001, Australia
fDepartment of Chemical and Biomolecular Engineering, The University of Melbourne, Melbourne, VIC 3010, Australia
First published on 19th October 2015
Gas-phase product detection studies of o-hydroxyphenyl radical and O2 are reported at 373, 500, and 600 K, at 4 Torr (533.3 Pa), using VUV time-resolved synchrotron photoionisation mass spectrometry. The dominant products are assigned as o-benzoquinone (C6H4O2, m/z 108) and cyclopentadienone (C5H4O, m/z 80). It is concluded that cyclopentadienone forms as a secondary product from prompt decomposition of o-benzoquinone (and dissociative ionization of o-benzoquinone may contribute to the m/z 80 signal at photon energies ≳9.8 eV). Ion-trap reactions of the distonic o-hydroxyphenyl analogue, the 5-ammonium-2-hydroxyphenyl radical cation, with O2 are also reported and concur with the assignment of o-benzoquinone as the dominant product. The ion-trap study also provides support for a mechanism where cyclopentadienone is produced by decarbonylation of o-benzoquinone. Kinetic studies compare oxidation of the ammonium-tagged o-hydroxyphenyl and o-methylphenyl radical cations along with trimethylammonium-tagged analogues. Reaction efficiencies are found to be ca. 5% for both charge-tagged o-hydroxyphenyl and o-methylphenyl radicals irrespective of the charged substituent. G3X-K quantum chemical calculations are deployed to rationalise experimental results for o-hydroxyphenyl + O2 and its charge-tagged counterpart. The prevailing reaction mechanism, after O2 addition, involves a facile 1,5-H shift in the peroxyl radical and subsequent elimination of OH to yield o-benzoquinone that is reminiscent of the Waddington mechanism for β-hydroxyperoxyl radicals. These results suggest o-hydroxyphenyl + O2 and decarbonylation of o-benzoquinone serve as plausible OH and CO sources in combustion.
The pyrolysis of phenol proceeds with H-migration and CO elimination to produce cyclopentadiene or, at higher temperatures, H-loss to produce the phenoxyl radical.11 Investigations into the phenol + OH reaction report the H-abstraction product as the phenoxyl radical.12,13 However, at >390 K, H-abstraction from the phenyl ring and OH addition reactions are also expected with the former process resulting in hydroxyphenyl radicals.14 The o-hydroxyphenyl radical is an intermediate in the pyrolysis reaction reported for dimethoxybenzene (a model compound for the β-04 aryl ether unit within G-type lignin).15 The addition of O2 to the o-hydroxyphenyl radical site will produce the o-hydroxyphenylperoxyl radical, with its hydroxy H-atom within close proximity to the peroxyl radical substituent. As is the case for the o-methylphenylperoxyl radical,16,17o-hydroxyphenylperoxyl is expected to isomerise and eliminate OH via a phenoxyl QOOH intermediate to produce o-benzoquinone (o-BQ), a known precursor to cyclopentadienone (CPO) + CO.18–21 This mechanism was reported for the oxidation of protonated tyrosinyl radicals22 and has some similarities to the Waddington mechanism for β-hydroxyperoxyl radicals.23–25 Yet, to date, no direct experimental results have validated this mechanism for the o-hydroxyphenyl + O2 reaction system.
In this work, we report reactions of gas-phase o-hydroxyphenyl with O2 using two approaches: synchrotron-based time-resolved photoionisation mass spectrometry and distonic-ion mass spectrometry. The synchrotron-based method couples a slow-flow kinetic reactor to a time-of-flight mass spectrometer and VUV photoionisation that allows detection of reaction products with kinetic and isomeric details. The distonic ion approach exploits charge-tagged derivatives of neutral radical species to study radical kinetics by ion-trap mass spectrometry.26 These distonic ion oxidation experiments build on a framework provided by previous studies of distonic phenyl27 and o-methylphenyl radical oxidation.17 In combination, we show that OH elimination follows the reaction of o-hydroxyphenyl radicals with O2 to form o-BQ. The stability of this nascent o-BQ is also investigated.
The heatable quartz reactor flow tube is 62 cm long with a 1.05 cm inner diameter maintained at 4 Torr (533.3 Pa). Gas continuously escapes the reactor into a differentially pumped vacuum chamber through a 650 μm pinhole situated 37 cm along the flow tube. In the experiments reported here, o-bromophenol, O2 gas, and He gas are supplied to the reactor through separate mass-flow controllers at the overall rate of 202 sccm. The o-bromophenol was entrained in He gas using a fritted bubbler with the liquid sample maintained at 291 K (18 °C) and ∼573 Torr (76.4 kPa). The vapour pressure of o-bromophenol is roughly approximated at 291 K to be 0.17 Torr using Antoine parameters known for phenol.31 At 373 K and 4 Torr, number densities within the reactor are ca. 1.7 × 1012 molecule cm−3 for o-bromophenol, 7.7 × 1015 molecule cm−3 for O2 gas, and a total of 9.6 × 1016 molecule cm−3 for He gas. Reactions were conducted with the reactor temperature maintained at 373 K, 500 K and 600 K. The temperature profile of the reactor is such that the length ca. 20 cm above the pinhole is maintained at the set temperature. Gas flow velocities are as follows: 10.1 m s−1 at 373 K, 13.5 m s−1 at 500 K, and 16.2 m s−1 at 600 K. Total gas flow densities were: 1.0 × 1017 molecule cm−3 at 373 K, 7.7 × 1016 molecule cm−3 at 500 K, and 6.4 × 1016 molecule cm−3 at 600 K.
The gas that escapes through the 650 μm pinhole is sampled by a skimmer to create a near-effusive molecular beam that is intersected by quasi-continuous vacuum-ultraviolet (VUV) synchrotron light. Ions produced by photoionisation are detected using a 50 kHz pulsed orthogonal-acceleration time-of-flight mass spectrometer. The photoionisation energy was typically scanned from 9 to 10 eV with 0.025 eV steps. Mass spectra are compiled into three-dimensional arrays of mass-to-change (m/z), reaction time, and photoionisation energy. All data are normalised for variations in the ALS photocurrent using a NIST-calibrated photodiode (SXUV-100, International Radiation Detectors Inc.). Background subtraction is achieved by subtracting the average signal during the 20 ms prior to the photolysis pulse from the dataset. The resulting photoionisation spectra and kinetic traces are normalised by the area under the curve and averaged together for each temperature. The error bars provided at a given photoionisation energy represent two standard deviations (2σ) for a mean of at least three measurements at 373 and 500 K, and two measurements at 600 K.
The O2 concentration (molecule cm−3) within the ion-trap region was determined using the measured pseudo-first order rate coefficient for 3-carboxylatoadamantyl + O2 and its known second-order rate coefficient of 8.5 ± 0.4 × 10−11 cm3 molecule−1 s−1 with the O2 concentration determined for each experiment.37 The background O2 concentration within the ion trap is typically 6.4 × 109 molecule cm−3 and the increased O2 concentrations ranged from 1.6–2.2 × 1011 molecule cm−3 with an O2-doped bath gas. The effective temperature of ions stored within a linear quadrupole ion trap has been estimated at 318 ± 23 K,38 consistent with an earlier estimate of 307 K.37
The kinetic plots that show ion signal decay with increasing reaction time were produced by integrating the ion signal intensity over a selected mass-to-charge range and normalising it to the integrated total ion signal intensity. The normalised integrated ion signal intensity is then averaged for at least 10 scans and plotted against reaction time (0.030–10000 ms) to track changes in ion signal intensity due to reactions with O2. Measured pseudo-first order rate coefficients (k1st) were obtained by fitting eqn (1) to the average normalised integrated peak intensity against reaction time, for a select mass-to-charge range, using the Levenberg–Marquardt algorithm. Satisfactory fits with eqn (1) are consistent with pseudo-first order kinetic behaviour. Thus, allowing the second-order rate coefficient (k2nd) to be calculated using eqn (2) with a measured [O2]. The residual plots accompanying kinetic curves in Fig. 5, Fig. S7, and S8 (ESI†) show the difference between the average normalised integrated ion signal intensity and the expected value from eqn (1), i.e. the residuals, plotted as a function of reaction time.
y = A0exp(−k1stt) + constant | (1) |
(2) |
Statistical uncertainty from fitting pseudo-first order rate coefficients (k1st) to experimental decay curves was typically 2σ ≤ 10%. Systematic uncertainty in the ion-trap pressure and O2 concentration, including the generation of neutrals and charged species with mass-to-charge less than the low mass cut-off (50 Th) result in an upper limit of 50% uncertainty in the O2 concentration that is accumulated in reported second-order rate coefficients and reaction efficiencies.
Fig. 1b is a product mass spectrum from photolysis of o-bromophenol in the presence of 7.7 × 1015 molecule cm−3 O2. The new product peaks at m/z 80, 108 and the minor peak at m/z 110 are consistent with C5H4O, C6H4O2, and C6H6O2 and are attributed to the o-hydroxyphenyl + O2 reaction. The PI spectra for m/z 80 and 108, integrated 0 to 20 ms after photolysis at 373 K, are provided in Fig. 2a and b. PI spectra at 500 K and 600 K are provided in the ESI† (Fig. S1 and S2, respectively). The PI onsets for m/z 80 at 9.4 eV and m/z 108 at 9.2 eV are in agreement with reference spectra for cyclopentadienone (CPO, m/z 80) and o-benzoquinone (o-BQ, m/z 108),53–55 and consistent with AIEs provided in Table 1. Ionization onsets for CPO and o-BQ were recently reported by Ormond et al.53 and compared within the inset of Fig. 2. The p-benzoquinone isomer can be excluded as a m/z 108 product contributor as its AIE is 9.96 eV with a sharp photoionisation onset,55,56 and there is no such feature in the PI spectrum up to 10 eV. The m/z 109 signal present in mass spectra obtained at 373, 500 and 600 K (ESI,† Fig. S3) could result, in part, from decomposition of o-hydroxyphenylperoxyl to o-hydroxyphenoxyl + O(3P). The hydroxyphenoxyl cation is expected at m/z 109, however unequivocal assignment of the m/z 109 species is confounded by the 13C isotope peak of the dominant m/z 108 product. In unpublished studies, we have observed phenoxyl radical decay that is kinetically matched to the growth of a +1 Da ion signal intensity. A m/z 110 product ion is present in Fig. S3 (ESI†) and kinetic traces in Fig. S5 (ESI†) show that the appearance of m/z 110 ions is delayed relative to m/z 108 ions (a primary product kinetic reference). Therefore, the delayed appearance of m/z 110 ions could be explained via H-abstraction by the o-hydroxyphenoxyl radical to produce o-catechol (C6H4OHOH, m/z 110).
Fig. 2 Photoionisation spectra integrated 0–20 ms after photolysis for (a) m/z 80 and (b) m/z 108 from o-hydroxyphenyl + O2 at 373 K. Each spectrum is an average of three PI spectra and the 2σ statistical uncertainty is represented by vertical error bars. Figures inset within (a) and (b) compare the experimental PI spectra near the onset to reference spectra for CPO and o-BQ from ref. 53 (1000 K). Reference PI spectra are also provided in (a) for CPO from ref. 54 and 55 (873 K). |
Species | Measured (eV) | Calculated AIE (eV) | Literature values |
---|---|---|---|
Cyclopentadienone (CPO, m/z 80) | 9.4 | 9.41 | 9.41 ± 0.01 (AIE)53 |
o-Benzoquinone (o-BQ, m/z 108) | 9.2 | 9.18 | 9.3 ± 0.1 (AIE)53 |
o-Hydroxyphenoxyl radical (m/z 109) | 8.14 | ||
o-Hydroxyphenol (catechol, m/z 110) | 8.14 | 8.56 (VIE)48,57 | |
p-Benzoquinone (p-BQ) | 9.89 | 9.96 ± 0.01 (AIE)48,56 |
The detection of o-BQ (m/z 108) is rationalised by O2 addition to the o-hydroxyphenyl radical, followed by isomerisation of the hydroxyphenylperoxyl intermediate to hydroperoxyphenoxyl and subsequent OH loss to form o-BQ (Scheme 1). This pathway is analogous to the O2 addition and subsequent OH loss mechanism that operates in the o-methylphenyl + O2 reaction16,17 and OH loss in the Waddington mechanism for β-hydroxyperoxyl radicals.23,24Scheme 1 also includes pathways from o-BQ that lead to CPO and the CPO radical cation that will now be discussed.
Included in Fig. 2a are reference PI spectra for CPO from Yang et al.54 and Parker et al.55 The close agreement between the m/z 80 and reference PI spectra shown in Fig. 2a from 9 to 9.8 eV support our assignments of m/z 80 as CPO. It is evident that at PI energies ≥9.8 eV all m/z 80 PI spectra diverge with the reference spectra under-predicting the current experimental data. Additional PI spectra acquired at 500 and 600 K (Fig. S1 and S2, ESI†) also diverge from the reference spectra at PI energies ≥9.8 eV.
The possibility of other C5H4O isomers contributing to the m/z 80 ion signal was ruled out by calculating AIEs for closed-shell linear C5H4O isomers listed in Table S1 (ESI†). Isomers were excluded on the basis of having: an AIE <9.2 eV, or an AIE >10.0 eV and a relatively high formation enthalpy. As it stands, CPO is the only plausible isomer contributing to the m/z 80 PI spectra, however, the source of neutral CPO and the cause of the disparity around 9.8 eV in the m/z 80 PI spectra (Fig. 2a) require further examination.
The systematic differences between the m/z 80 signal and CPO reference spectra in Fig. 2a at photoionisation energies ≥9.8 eV could arise from dissociative ionisation of higher mass species, where o-BQ is a likely candidate. It is known that dissociative ionisation of 1,2-naphthoquinone and 9,10-phenanthrenequinone result in CO loss (both contain the o-BQ substructure).58 Fig. S3 (ESI†) shows product mass spectra at 373, 500 and 600 K integrated over two energy ranges; 9.40–9.75 eV (Fig. S3a–c, ESI†) and 9.85–10.00 eV (Fig. S3d–f, ESI†). These mass spectra reveal some variation in the product ratios but no additional product signals. Comparing the kinetic traces for m/z 80 and 108 at 500 K integrated over 9.40–9.75 eV (Fig. S4a, ESI†) shows that the kinetic traces are clearly different and consistent with the dominant fraction of each ion population arising from photoionisation of different neutrals. However, at higher energies (9.85–10.00 eV, Fig. S4b, ESI†), the m/z 80 and 108 kinetic traces appear more similar – this is consistent with a portion of C6H4O2 (108 Da) undergoing dissociative ionisation to yield product ions with m/z 80. Furthermore, the potential energy scheme for CO loss from the o-BQ radical cation (m/z 108) provided in Fig. 3 shows the cation dissociation barrier to be 9.7 eV relative to neutral o-BQ. These results support the proposition that at photoionisation energies ≥9.8 eV some of the m/z 80 signal arises from the dissociative ionisation of o-BQ. This contribution is in addition to the ionisation of CPO produced within the reactive flow.
Analogous to product pathways of the phenylperoxyl radical in phenyl + O2 reactions,59–62 CPO could be produced after decomposition of hydroxyl-substituted oxepinoxyl radicals. Unimolecular reaction pathways leading directly to CPO are discounted on the basis of experiments in Section 3.2 and prohibitively high energy pathways reported in Section 3.3.3. Ultimately, we propose that the m/z 80 and 108 products are generated according to processes summarised in Scheme 1: the o-hydroxyphenyl radical undergoes O2 addition to form the hydroxyphenylperoxyl radical and subsequent OH loss to produce o-BQ. And, a portion of the nascent vibrationally-excited o-BQ population then decomposes via decarbonylation to produce CPO.18–20 In addition, dissociative ionisation of o-BQ possibly contributes to the measured m/z 80 signal at energies ≥9.8 eV.
To further establish connections between the reaction products of o-hydroxyphenyl + O2 (cf.Scheme 1), charge-tagged derivatives of o-hydroxyphenyl radicals were prepared within an ion-trap mass spectrometer (at the University of Wollongong). The study of distonic radical ions can provide useful insight into the reactions of their neutral radical counterparts. The presence of a relatively unreactive charged substituent enables isolation and manipulation of reactive intermediates using ion-trap mass spectrometry, while products arise from reactions with the spatially separated radical moiety.17,63,64 Quantum chemical calculations were also conducted, and discussed later in Section 3.3, to rationalise experimental results for both the neutral and charge-tagged systems.
Fig. 5 Kinetic curves for m/z 109 (solid blue circles), from PD of 3-bromo-4-hydroxybenzaminium cations, in reactions with (a) background O2 (6.4 × 109 molecules cm−3) and (b) increased O2 (1.9 × 1011 molecules cm−3). Residual plots from the fitting of eqn (1) are provided above. The m/z 124 product data are shown (red diamonds) and track with a rate coefficient in agreement with the m/z 109 decay (within 2σ). Error bars are 1σ. |
The mass spectrum in Fig. 4c, from isolation and subsequent collision-induced dissociation (CID) of m/z 124 product ions, shows major signals at m/z 96 and 107 and minor signals at m/z 79 and 81. The product ion at m/z 96 (−28 Da) is consistent with decarbonylation of ammonium-tagged o-BQ to yield ammonium-tagged CPO + CO. Fragment ions at m/z 107, 79, and 81 are assigned to loss of NH3 (−17 Da), NH3 + CO (−45 Da) and NC2H5 or C2H3O (−43 Da) from m/z 124, respectively. To verify these assignments, 18O2 was introduced into the ion trap and reacted with m/z 109 radical cations. The m/z 126 ions produced are consistent with 18O2 addition and 18OH loss (−19 Da, Fig. 4d) and exclude any contribution of NH3 loss (−17 Da). Isolation and subsequent CID of the m/z 126 ions resulted in fragments at m/z 96 and 98 consistent with loss of C18O and C16O from 18O-labelled o-BQ to yield CPO. Taken together, these data demonstrate a connection between the o-BQ intermediate (m/z 124) and the CPO structure (m/z 96) via processes summarised in Scheme 2. These data do not provide evidence for a “phenyl-like” oxidation mechanism for the direct formation of CPO via phenoxyl and oxepinoxyl radicals.59–62 Other fragment ions at m/z 79, 81, 83, and 109 are assigned to loss of NH3 + C18O (−47 Da), NH3 + C16O or C2H318O (−45 Da), NC2H5 or C2H316O (−43 Da), and NH3 (−17 Da) from m/z 126, respectively.
Potential energy schemes for formation of o-BQ and CPO are compared and discussed for both neutral and distonic cases in Section 3.3. Reactions of the PD generated 5-ammonium-2-hydroxyphenyl radical cation (m/z 109) with O2 were characterised further by kinetic measurements.
A single exponential decay (eqn (1)) was satisfactorily fitted to the experimental data, in accord with pseudo-first order kinetic behaviour. Representative kinetic curves for m/z 109 and 124 ions are provided in Fig. 5 with fitted data and residuals from eqn (1) for m/z 109 signal decay. The k1st values for m/z 109 signal decay and m/z 124 signal growth are in agreement (e.g., in Fig. 5a, 1.9 ± 0.2 s−1 compared to 1.8 ± 0.1 s−1 within 2σ) and the m/z 124 intensity is well matched to the m/z 109 signal decay. This indicates that m/z 124 ions are the main reaction product from depletion of m/z 109 ions. As shown in Fig. 5b, at increased O2 concentrations the m/z 109 ion signal intensity ultimately approaches a constant value of ca. 10% at 1000 ms and remains constant up to a reaction time limit of 10000 ms. This indicates the presence of an unreactive isomer (or isomers) and is accounted for by the constant offset included in eqn (1).
Additional experiments that compare the oxidation kinetics of the ammonium-tagged o-hydroxyphenyl and o-methylphenyl radical cations along with trimethylammonium-tagged analogues are now described. Sample kinetic plots are provided in Fig. S7 (ESI†) for oxidation of 5-ammonium-2-methylphenyl radical cations (m/z 107) and in Fig. S8 (ESI†) for 5-(N,N,N-trimethylammonium)-2-hydroxyphenyl radical cations (m/z 151). For reactions of 5-ammonium-2-methylphenyl radical cations (m/z 107) the non-zero horizontal offset (shown in Fig. S7b, ESI†) is ca. 40% of the isolated m/z 107 ion population. Interestingly, k1st values for 5-ammonium-2-methylphenyl radical (m/z 107) and 5-ammonium-2-hydroxyphenyl radical (m/z 109) signal decay are separable with 2σ uncertainty, where the k1st for the 5-ammonium-2-hydroxyphenyl radical cations is reproducibly greater by ca. 15%. In the case of trimethylammonium-tagged o-hydroxyphenyl radical + O2 reactions, the m/z 151 ion population can be completely depleted by O2 reaction, suggesting that a pure population of trimethylammonium-tagged o-hydroxyphenyl radicals are formed from PD of the precursor. This observation is consistent with our previous investigation of trimethylammonium-tagged o-methylphenyl + O2 reaction kinetics17 and may be attributed to the greater number of internal degrees of freedom from the trimethylammonium substituent thus reducing the propensity for isomerisation.
Second-order rate coefficients (k2nd, cm3 molecule−1 s−1) and reaction efficiencies (Φ%) derived from fitted pseudo-first order rate coefficients (k1st) are reported in Table 2. Collision frequencies were calculated using the Langevin collision model.39 Kinetic measurements were conducted at background O2 ([O2] = 6.4 ± 0.4 × 109 molecule cm−3) and increased O2 concentrations ([O2] = 1.6–2.2 × 1011 molecule cm−3). Repeated kinetic measurements provided consistent results and statistical uncertainties from fitting k1st were typically 2σ ≤ 10%. These results indicate stable absolute O2 concentrations within the ion trap as indicated by the linear relationship between k1st and [O2] (Fig. S9, ESI†). However, the uncertainty in the ion-trap pressure ultimately results in an upper limit of 50% uncertainty in the trap [O2], and consequently, a 50% uncertainty for the reported second-order rate coefficients and reaction efficiencies in Table 2.
Distonic radical | [O2] (molecule cm−3) | k 2nd (cm3 molecule−1 s−1) | Collision frequency (cm3 molecule−1 s−1) | Φ (%) |
---|---|---|---|---|
a Rate coefficients reported in ref. 17. Ions of m/z 149 were generated by PD of the 3-bromo-N,N,N,4-trimethylbenzenaminium cation. | ||||
5-Ammonium-2-hydroxyphenyl | Low [109] | 2.9 × 10−11 | 5.9 × 10−10 | 4.9 |
High [1011] | 3.2 × 10−11 | 5.5 | ||
5-Ammonium-2-methylphenyl | Low [109] | 2.6 × 10−11 | 5.9 × 10−10 | 4.4 |
High [1011] | 2.6 × 10−11 | 4.4 | ||
5-(N,N,N-Trimethylammonium)-2-hydroxyphenyl | 6.6 × 109 | 2.5 × 10−11 | 5.7 × 10−10 | 4.4 |
5-(N,N,N-Trimethylammonium)-2-methylphenyla | 2.2 × 109 | 2.9 × 10−11 | 5.7 × 10−10 | 5.1 |
8.5 × 109 | 2.6 × 10−11 | 4.5 |
Reaction efficiencies for all species reported in Table 2 are all approximately equal to 5%, similar to reported reaction efficiencies for neutral phenyl radicals and a range of positively charged distonic phenyl radical ions, including trimethylammonium and pyridinium-tagged phenyl radicals,27 and distonic o-methylphenyl radicals.17 For the 5-ammonium-2-hydroxyphenyl + O2 reaction mechanism discussed further below, proximity of the ortho-OH-substituent to the peroxyl-radical site in the o-hydroxyphenylperoxyl radical provides a notably low-energy reaction pathway (refer to Fig. 6) that competes with dissociation of the peroxyl radical intermediate toward separated reactants, however, it does not appear to significantly affect measured reaction efficiencies compared to the other values reported in Table 2 and for the phenyl-type radical + O2 reactions cited above. This moderate reaction efficiency of ∼5%, consistent for a range of phenyl-type radicals,17,27 indicates that the rate of reaction is not controlled by the microcanonical rate for forward dissociation pathways. Instead, it may result from an entropic bottleneck after formation of the non-covalent complex between the phenyl radical and O2 (ref. 65–67) that reflects reaction flux back to the free reactants. More experiments and insights are required to address this question.
Addition of O2 to the neutral o-hydroxyphenyl radical produces the o-hydroxyphenylperoxyl radical species (N2) that is 49.0 kcal mol−1 below the energy of separated reactants (N1), as shown in Fig. 6. Close proximity of the OH substituent to the peroxyl radical site in the o-hydroxyphenylperoxyl radical (N2) allows for a 1,5-H shift viaTSN2 → N3 with an incredibly small 1.7 kcal mol−1 barrier to the o-hydroperoxyphenoxyl radical (N3). Elimination of OH from the hydroperoxyl group in N3viaTSN3 → N4 (11.5 kcal mol−1 barrier) results in o-BQ + OH with a reaction exothermicity of 51.9 kcal mol−1. Comparing this to the charge-tagged case, the o-hydroxyphenylperoxyl radical analogue (C2) is 44.8 kcal mol−1 below the energy of the separated reactants (C1). The barrier to the 1,5-H shift in C2 and subsequent OH loss from C3 is 27.4 kcal mol−1 below reactants and the resulting 4-ammonium-2-benzoquinone + OH (C4) products are formed with an exothermicity of 38.9 kcal mol−1 (13.0 kcal mol−1 less than in the neutral case). As shown by Fig. S10 in the ESI,† the reaction enthalpy for charge-tagged o-BQ + OH is reduced by separation of the charge tag and ring structure via inclusion of methylene linkages, indicating that differences shown in Fig. 6 are (in part) due to a through-space charge effect. Still, intermediates and transition states for both cases shown in Fig. 6 are well below the energy of the reactants and, therefore, OH-elimination is expected to be facile. This mechanism is consistent with the appearance of o-BQ (m/z 108) in neutral flow-tube experiments and ammonium-tagged o-BQ (m/z 124) in the distonic radical cation experiments.
Fig. 7 Potential energy schematic for CO loss from o-BQ (N4) along the singlet C6H4O2 surface. G3X-K 0 K enthalpies are provided in kcal mol−1 relative to o-BQ. |
Fig. S11 (ESI†) shows the potential energy scheme for CO elimination from ammonium-tagged o-BQ. The mechanisms shown in Fig. S11a (ESI†) feature barriers that exceed the entrance channel (5-ammonium-2-hydroxyphenyl + O2) by 2.8 kcal mol−1viaTSC6a → C8 and 6.7 kcal mol−1viaTSC6b → C8. In Fig. S11b (ESI†), however, the highest barrier is 35.0 kcal mol−1viaTSC4 → C5 (3.8 kcal mol−1 below 5-ammonium-2-hydroxyphenyl + O2). The decomposition reactions shown in Fig. S11 (ESI†) are less likely to proceed due to the reduced exothermicity of the charge-tagged o-BQ + OH and barriers to decarbonylation approaching the entrance channel limit. Collisional activation of the charge-tagged o-BQ intermediate should provide the activation energy required to generate charge-tagged CPO + CO, consistent with a loss of 28 Da from CID of m/z 124 ions shown in Fig. 4c. The appearance of a small m/z 96 ion peak in Fig. 4b (<1%), prior to isolation of the m/z 124 ion, may result from decomposition of the high-energy portion of the nascent m/z 124 ion ensemble. It is likely that further exploration of o-BQ decomposition is required to reveal additional competitive pathways resulting in CPO + CO.
Rearrangement of o-hydroxyphenylperoxyl (N2) toward 7-hydroxyoxepinoxyl (N23) and 6-carboxy-1-oxo-hex-2,4-dienyl radicals (N21) via dioxirane-hydroxycyclohexadienyl intermediates are described in Fig. 8. The reactions of these intermediates represent plausible unimolecular pathways to both m/z 80 and 108 ions in the ALS experiments. Reactions toward the 7-hydroxyoxepinoxyl radical proceeds through TSN2 → N22 18.1 kcal mol−1 above the barrier to o-BQ (TSN3 → N4 in Fig. 6) with an exothermicity of 102.3 kcal mol−1. Formation of the 6-carboxy-1-oxo-hex-2,4-dienyl radical (N21) occurs viaTSN2 → N20 at 11.6 kcal mol−1 above TSN3 → N4. Reactions of charged-tagged o-hydroxyphenylperoxyl toward hydroxyoxepinoxyl and 6-carboxyoxohexdienyl, shown in Fig. S13 (ESI†), generally parallel those described by Fig. 8. The rate limiting steps toward the two hydroxyoxepinoxyl (TS3c) and carboxyoxohexdienyl radicals (TS1c) are 22.2 and 5.1 kcal mol−1, respectively, above the barrier to charge-tagged o-BQ (TS C3 → C4, Fig. 6).
In the case where either hydroxyoxepinoxyl or carboxyoxohexdienyl radicals are produced, their decomposition could possibly follow pathways described by Fig. S14 and S15 (ESI†). Likely products, by analogy to phenyl radical oxidation,16,61 include o-BQ + OH, CPO + HOCO, 3-hydroxy-2-benzoquinone + H, and 2-hydroxy-cyclopentadienone + HCO with barriers far below the reactants. The absence of peaks at m/z 96 and 124 within ALS experimental results (Fig. 1) and the high barriers to hydroxyoxepinoxyl and carboxyoxohexdienyl intermediates indicate that at most only a small fraction of the reaction flux follows these channels. Furthermore, preliminary RRKM modelling of the o-hydroxyphenylperoxyl radical (N2) decomposition, utilizing MultiWell,45 indicates H-migration and OH-loss to form o-BQ (N4, shown in Fig. 6) comprehensively outcompetes the pathways toward hydroxyoxepinoxyl and carboxyoxohexadienyl radicals (shown in Fig. 8, N2 toward TS1n and TS4n). The sums of states for salient transition states and corresponding rate coefficients are provided in Table S2 (ESI†). The oxepinoxyl pathways (viaTS1n and TS4n) experience comparatively tight transition states with state counts several orders of magnitude lower than any other along the o-BQ pathway. This is in accord with our previous statement that the prevailing mechanism is formation of o-BQ via an o-hydroxyperoxylphenoxyl radical intermediate (N3). The appearance of m/z 80 in ALS experiments is explained by o-BQ decomposition, supported by distonic experiments that show connectivity between the analogous charge-tagged species.
Second order rate coefficients (k2nd) for 5-ammonium-2-hydroxyphenyl (m/z 109) + O2 were measured to have a 5% reaction efficiency. Additional kinetic measurements for O2 reactions with PD generated 5-ammonium-2-methylphenyl and 5-(N,N,N-trimethylammonium)-2-hydroxyphenyl radical cations and a previous investigation of trimethylammonium-tagged o-methylphenyl + O2 reaction kinetics17 demonstrate for this small set that the identity of the charged-tag and ortho-substituent does not significantly affect the reaction efficiency (ca. 5%).
Quantum chemical calculations are in accord with our experimental observations, where a 1,5-H shift in the o-hydroxyphenylperoxyl adduct and subsequent OH elimination is the minimum energy pathway for both o-hydroxylphenyl + O2 and the ammonium-tagged counterpart. Decomposition of the o-BQ toward CPO does encounter large barriers. However, the indication from preliminary kinetic modelling is that production of o-BQ is the dominant unimolecular pathway.
The prevailing mechanism for decomposition of the o-hydroxyphenylperoxyl radical produced by O2 addition is via 1,5-H migration and OH loss from the hydroperoxyphenoxyl radical intermediate to produce o-BQ. Its decomposition via ring opening, cyclisation, and CO elimination is the likely pathway to CPO. These proposed pathways to o-BQ and CPO serve as source of OH and CO species in reactive environments.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5cp02953h |
This journal is © the Owner Societies 2016 |