Experimentally Determined Site-Specific Reactivity of the Gas-Phase

Article pubs.acs.org/JPCA

Experimentally Determined Site-Specific Reactivity of the Gas-Phase OH and Cl + i‑Butanol Reactions Between 251 and 340 K Max R. McGillen,†,‡,⊥ Geoffrey S. Tyndall,*,§ John J. Orlando,§ Andre S. Pimentel,∥ Diogo J. Medeiros,∥,# and James B. Burkholder*,† †

Chemical Sciences Division, Earth System Research Laboratory, National Oceanic and Atmospheric Administration, 325 Broadway, Boulder, Colorado 80305, United States ‡ Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, Colorado 80309, United States § Atmospheric Chemistry Observations and Modeling Laboratory, National Center for Atmospheric Research, Boulder, Colorado 80307, United States ∥ Departamento de Química, Pontifícia Universidade Católica do Rio de Janeiro, Rio de Janeiro, Brazil S Supporting Information *

ABSTRACT: Product branching ratios for the gas-phase reactions of i-butanol, (CH3)2CHCH2OH, with OH radicals (251, 294, and 340 K) and Cl atoms (294 K) were quantified in an environmental chamber study and used to interpret i-butanol site-specific reactivity. iButyraldehyde, acetone, acetaldehyde, and formaldehyde were observed as major stable end products in both reaction systems with carbon mass balance indistinguishable from unity. Product branching ratios for OH oxidation were found to be temperaturedependent with the α, β, and γ channels changing from 34 ± 6 to 47 ± 1%, from 58 ± 6 to 37 ± 9%, and from 8 ± 1 to 16 ± 4%, respectively, between 251 and 340 K. Recommended temperature-dependent sitespecific modified Arrhenius expressions for the OH reaction rate coefficient are (cm3 molecule−1 s−1): kα(T) = 8.64 × 10−18 × T1.91exp(666/T); kβ(T) = 5.15 × 10−19 × T2.04exp(1304/T); kγ(T) = 3.20 × 10−17 × T1.78exp(107/T); kOH(T) = 2.10 × 10−18 × T2exp(−23/T), where kTotal(T) = kα(T) + kβ(T) + kγ(T) + kOH(T). The expressions were constrained using the product branching ratios measured in this study and previous total phenomenological rate coefficient measurements. The site-specific expressions compare reasonably well with recent theoretical work. It is shown that use of i-butanol would result in acetone as the dominant degradation product under most atmospheric conditions.

1. INTRODUCTION The atmospheric oxidation of volatile organic compounds (VOCs) has long been implicated in the formation of tropospheric ozone, secondary organic aerosol, and other secondary pollutants, for example, peroxyacylnitrates. Biofuels are a subset of VOCs, whose use is expected to increase in the decades to come. Large-scale production and consumption of fuel leads to its emission into the atmosphere either through incomplete combustion or leakage and evaporation. Biofuels are therefore expected to become a growing fraction of VOCs present in the atmosphere, as has been observed with widespread use of ethanol biofuel.1,2 The reactivity of biofuels in the atmosphere with the OH radical and the stable products formed in their degradation has a direct impact on air quality. For biofuels such as i-butanol, (CH3)2CHCH2OH, competing reaction pathways exist that lead to the formation of different stable end products, each of which may have different effects upon air quality. The competing H atom abstraction reaction channels and expected © XXXX American Chemical Society

stable end products for i-butanol under high NOx (NOx = NO + NO2) conditions are outlined in Mechanism 1 (Scheme 1). In addition, the initial site-specific reactivity of the OH reaction is expected to be temperature-dependent3−6 and therefore lead to a different distribution of stable end products depending on the location and time of its emission into the atmosphere. Quantifying the end-product yields and temperature dependence enables the site-specific reactivity of each reaction channel to be evaluated. Presently, there is limited experimental data available on the site-specific reactivity for the OH + i-butanol reaction. In a room-temperature environmental chamber study, Andersen et al.7 reported an acetone endproduct yield, β-channel, of 61%. Using a different approach, McGillen et al.4 used their measured total phenomenological rate coefficients, unpublished product yield data, and Received: September 13, 2016 Revised: November 14, 2016 Published: November 22, 2016 A

DOI: 10.1021/acs.jpca.6b09266 J. Phys. Chem. A XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry A

Samples were irradiated in discrete steps, the number of which ranged from 5 to 10 per experiment. FTIR spectra were recorded with multipass optics housed within the chamber, providing a path length of 32.6 m. Absorption spectra consisted of 200 coadditions recorded at 1 cm−1 resolution after each photolysis period. Gas concentrations were monitored periodically throughout an experiment using a GC-FID. GC-FID samples were taken by flowing gas from the over-pressurized reaction chamber through 1/8 in. OD Teflon tubing at a flow rate of ∼150 cc min−1 and into a sample loop of a six-port valve connected to the GC column. The sampling loop consisted of 28 cm of 0.5 mm i.d. PEEK tubing. The GC column was a 30 m, 0.32 mm i.d., 0.25 μm film thickness HP-5 column. The carrier gas flow was 4.5 cc min−1 (He). The GC column was maintained at 28 °C for 4 min, ramped at 8 °C min−1 to 65 °C, and then ramped at 40 °C min−1 to 300 °C and held at that temperature for 2 min. The GC-FID detector response was calibrated using manometrically prepared standards sampled from the chamber. Infrared absorption measurements were also used to cross check the calibration whenever possible. Product branching ratios were determined from the measured profile of the end-product concentration versus Δ[i-butanol], which in the absence of secondary chemistry should be a straight line with a slope equal to the product branching ratio. It was necessary, however, to account for the dark loss of i-butanol and secondary losses affecting reaction products, as described below, to obtain accurate branching ratios. 2.1. OH + i-Butanol. OH radicals were produced via UV/ vis photolysis of either methylnitrite, CH3ONO:

Scheme 1. Mechanism 1

structure−activity relationships (SARs) to estimate the temperature-dependent rate coefficients for each of the reaction channels, with the α-channel estimated to be the site of greatest reactivity over the temperature range of ∼250−1000 K. In this work, the stable end products associated with the α, β, and γ reaction pathways were measured at 251, 294, and 340 K in a series of environmental chamber studies. The measured yields, carbon mass balance, and complementary Cl atom reaction product yields at 294 K establish rigorous constraints on the reactive system enabling an experimental determination of the site-specific reactivity over a range of atmospherically relevant temperatures. By assessing product branching ratios over the atmospheric temperature range we are able to show that under most atmospheric conditions, acetone formation from both first- or second-generation oxidation processes will dominate the product distribution for this compound. The present results are also used to update the site-specific reactivity estimates provided by McGillen et al.,4 which takes a global approach to defining k(T) over a broad temperature range. With new and direct measurements of reaction branching ratios, we reduced the reliance of this approach on SARs and improved the accuracy with which branching ratios are estimated.

CH3ONO + hν → CH3O + NO

(1)

CH3O + O2 → CH 2O + HO2

(2)

HO2 + NO → OH + NO2

(3)

or i-propylnitrite, (CH3)2CHONO:

2. EXPERIMENTAL METHODS Experiments were performed in a 47 L jacketed stainless steel environmental reaction chamber8 coupled to Fourier transform infrared (FTIR) absorption and gas chromatography-flame ionization (GC-FID) detection systems. The experimental apparatus has been used extensively in previous kinetic and mechanistic studies9,10 and is only briefly described below. Stable end-product yields for the OH + i-butanol reaction in the presence of O2 and NO were measured at 251, 294, and 340 K. Product yields for the Cl + i-butanol reaction were measured at 294 K in an O2/N2 bath gas in the presence and absence of NO. The chamber temperature was maintained by circulating ethanol (251 K experiments) or water (340 K experiments) from a temperature-controlled reservoir through the outer jacket of the chamber. Experiments were performed by introducing the reactant, ibutanol, together with precursors for OH or Cl (alkylnitrites (RONO) and Cl2, respectively) from a gas manifold into the chamber. NO was added to promote RO2 to RO radical conversion, facilitating the formation of stable end products from the β and γ reaction sites. Bath gas (N2/O2) was added rapidly bringing the total pressure to 760−800 torr, while helping to mix the reactants and oxidant precursors. Gas mixtures were then irradiated using the filtered output of a 250 W xenon arc lamp, providing radiation from 290−420 nm.

(CH3)2 CHONO + hν → (CH3)2 CHO + NO

(4)

(CH3)2 CHO + O2 → CH3C(O)CH3 + HO2

(5)

as the OH radical precursor. Individual photolysis periods were from 1 to 20 min with total photolysis times of ∼3 to 60 min and a total experiment duration of 1−2 h. Initial concentrations of reactants were: [i-butanol] 6−29 × 1013 molecules cm−3, [methylnitrite or isopropylnitrite] 6−21 × 1013 molecules cm−3, [NO] 0−21 × 1013 molecules cm−3, ∼300 torr N2, ∼470 torr O2. Several test experiments were conducted at higher initial [i-butanol] and [NO]; however, it was found that at these higher concentrations stable endproduct yields for the major reaction channels shown in Mechanism 1 (Scheme 1) were suppressed as a consequence of side chemistry and at 251 K dark losses of reactant and products. Two OH radical sources were used to enable the isolation of the acetaldehyde and acetone end products. Also, the chromatographic methods used did not effectively separate CH3ONO from acetaldehyde. This precluded the use of CH3ONO as the OH precursor for determining the i-butanol γsite yield, whose stable end product was acetaldehyde. The use of (CH3)2CHONO enabled the detection of acetaldehyde, but B

DOI: 10.1021/acs.jpca.6b09266 J. Phys. Chem. A XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry A

Table 1. Site-Specific Branching Ratios Determined in This Work for the Reactions of OH Radicals and Cl Atoms with iButanol

OH reaction reaction site α β γ

251 K 29a

38b 50b

7a

Cl reaction

294 K 34 ± 6c 58 ± 6c 8 ± 1c

35a

47b 43b

10a

340 K 41 ± 8c 48 ± 3c 11 ± 1c

46a 21a

48b 50b

294 K 47 ± 1c 37 ± 9c 16 ± 4c

48 ± 4 35 ± 4 17 ± 4

i-Propylnitrite experiments. Methylnitrite experiments. Final normalized values with estimated errors of the standard deviation between the two α channel experimental values and for the β and γ channels the standard deviation between the experimental and normalized values. a

b

c

at higher i-butanol conversion, as curvature in the product yield profiles. Secondary reaction was accounted for using the analytical expression:11

the production of acetone as a (CH3)2CHONO degradation product prevented the determination of the acetone yield from the β-site of i-butanol. Consequently, experiments with both OH precursors were used to obtain accurate branching ratios for all major reaction pathways in i-butanol oxidation. In this study, dark heterogeneous nitrosation reactions were observed:

⎫ ⎛ kRH − k prod ⎞⎧ 1 − ([RH]t /[RH]0 ) ⎬ ⎟⎨ F=⎜ k prod / kRH kRH ⎠⎩ ([RH]t /[RH]0 ) ⎝ − ([RH]t /[RH]0 ) ⎭

(6)

(CH3)2 CHCH 2OH + CH3ONO wall

(7)

(CH3)2 CHCH 2OH + (CH3)2 CHONO

Cl 2 + hν → Cl + Cl

wall

⎯⎯⎯→ (CH3)2 CHOH + (CH3)2 CHCH 2ONO

⎪

where kRH is the rate coefficient of OH + i-butanol, kprod is the rate coefficient of OH + product, [RH]0 is the initial [ibutanol], [RH]t is the [i-butanol] at time t, giving F, a multiplicative factor that accounts for the secondary loss of the product. Once F is applied to [product] and Δ[i-butanol] is adjusted for dark loss, eq I, a linear relationship, is expected. 2.2. Cl + i-Butanol. Cl atoms were produced by the photolysis of molecular chlorine:

wall

⎯⎯⎯→ CH3OH + (CH3)2 CHCH 2ONO

⎪

⎪

(II)

(CH3)2 CHCH 2OH + NO2 ⎯⎯⎯→ (CH3)2 CHCH 2ONO + products

⎪

(8)

(9)

Experiments were performed at 294 K in the presence and absence of NO in an O2/N2 bath gas. Unlike the OH experiments, in which NOx was always present, Cl-initiated experiments could be used to probe reaction products either in the presence or absence of NOx. A disadvantage to performing the Cl-initiated reaction in the presence of NOx is that it produces OH during the experiment, which can complicate Clinitiated experiments in general. Performing experiments in the presence and absence of NO provides a check on the contributions of OH chemistry and whether or not this is being quantified correctly. Initial concentrations used were: [i-butanol] 3.5−7.0 × 1014 molecules cm−3, [Cl2] 2.1 × 1015 molecules cm−3, [NO] when present 7.0−10.5 × 1014 molecules cm−3, O2 150−400 torr, N2, 350−550 torr, and a total pressure of 700−760 torr. Both FTIR and GC detection methods were used in two experiments, one with and one without NO. The FTIR and GC measurements gave product yields that agreed to within ∼5%. As with the OH-initiated experiments, it was necessary to account for secondary loss of i-butyraldehyde and acetaldehyde. However, the presence of NO led to secondary production of OH radicals, which reacted with the aldehydes and complicated the analysis. 2.3. Materials. CH3ONO and (CH3)2CHONO were synthesized by adding sulfuric acid (50% in water) dropwise to a 0 °C saturated solution of NaNO2 in methanol or isopropanol, respectively. The gaseous effluent was trapped at −78.5 °C and was pumped on at this temperature to remove volatile impurities such as NO2. i-Butanol (≥99.5%), i-

These reactions contributed to the overall observed loss of ibutanol and needed to be accounted for in the determination of the loss attributable to reaction with the OH radical. Because both NO2 and R′ONO were typically present in our experiments, the relative contributions from reactions 6−8 were not resolved. The nitrosation reactions occurred on the chamber walls in the presence of NO2 and alkylnitrites, but the exact mechanism is unknown. The dark loss of i-butanol was, however, quantified from the observed formation of (CH3)2CHCH2ONO, RONO, and its temporal profile, such that Δ[i‐butanol]t = [i‐butanol]0 + [RONO]0 − [i‐butanol]t − [RONO]t

(I)

where [i-butanol]0 and [RONO]0 are the initial concentrations of i-butanol and i-butylnitrite, [i-butanol]t and [RONO]t are their concentrations at time t, and Δ[i-butanol]t is the loss of starting material due to reaction with OH. Note that initial concentrations refer strictly to the concentrations measured directly before photolysis is initiated, since dark reactions are occurring continuously throughout the experiment. Corrections as a result of these processes varied both between experiments and with experimental duration, and by the end of the experiment the magnitude of this correction ranged from 4 to 64% depending on the conditions. It was necessary to account for the secondary reaction of OH with i-butyraldehyde and acetaldehyde in the branching ratio determination. Secondary reactions were observed, particularly C

DOI: 10.1021/acs.jpca.6b09266 J. Phys. Chem. A XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry A

when the i-butylnitrite formed in reactions 6−8 is photolyzed, where the photolytic loss rate of i-butylnitrite was assumed to be the same as methylnitrite. This assumption is supported by similar photolytic loss of methylnitrite and i-propylnitrite throughout our experiments, indicating that the overall photolysis rate of the alkylnitrites is not strongly dependent on the identity of alkyl substitutions. Figure 2, which shows a graphical representation of the magnitude of the corrections for each end product, shows that this source of i-butyraldehyde was minor at all temperatures. The α-channel branching ratio was obtained from a linear least-squares analysis of the [i-butyraldehyde] concentration profiles, after accounting for secondary loss and production, and was found to increase from 34 ± 6 to 47% ± 1 over the temperature range of 251−340 K. Estimated uncertainties are the 1σ standard deviation between two α-channel determinations at each temperature (using both methylnitrite and ipropylnitrite OH sources). These uncertainties are considered a good estimate of the absolute error, since reproducibility between experiments was found to be a larger source of uncertainty than the precision of an individual determination. 3.1.2. Measurement of the β-Product Branching Channel. The branching ratio of the β-product channel was determined by monitoring acetone formation relative to the loss of ibutanol. Acetone is formed in the following mechanism (Scheme 3). Data for [acetone] versus Δ[i-butanol] obtained at 251, 294, and 340 K are shown in Figure 1 and listed in Table 4. Acetone is not consumed under the conditions of our experiments, primarily because it is much less reactive than i-butanol. However, acetone is formed when the first-generation product i-butyraldehyde is oxidized by OH, which in the absence of NO2 would produce acetone in a nearly 100% yield following oxidation at its aldehydic and tertiary hydrogen sites. The secondary acetone formation was accounted for in the data analysis as

butyraldehyde (≥99.5%), acetone (>96%), acetaldehyde (>99.5%), i-butylnitrite (>95%), NO (>99.5%), N2 (liquid N2 boil-off), O2 (99.99%), and Cl2 (>99.5%) were used as supplied.

3. RESULTS AND DISCUSSION In this section, the product branching ratio measurements are presented and interpreted for H-atom abstraction at the α, β, and γ reaction sites of i-butanol (see Mechanism 1 (Scheme 1)). H-atom abstraction from the OH moiety was not explicitly included in this analysis, as it would be indistinguishable in the end-product analysis from reaction at the α and/or β site. However, it is expected to have a minor yield,