Roaming-Like Mechanism for Dehydration of Diol Radicals - The

7 days ago - In the present study, we propose a new low-energy, roaming-like mechanism based on a detailed dispersion-corrected DFT and ab initio leve...
0 downloads 0 Views 992KB Size
Subscriber access provided by University of Winnipeg Library

A: Kinetics, Dynamics, Photochemistry, and Excited States

Roaming-Like Mechanism for Dehydration of Diol Radicals Rubik Asatryan, Yudhajit Pal, Johannes Hachmann, and Eli Ruckenstein J. Phys. Chem. A, Just Accepted Manuscript • DOI: 10.1021/acs.jpca.8b08690 • Publication Date (Web): 28 Nov 2018 Downloaded from http://pubs.acs.org on November 30, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Roaming-Like Mechanism for Dehydration of Diol Radicals Rubik Asatryan,1,* Yudhajit Pal,1,3 Johannes Hachmann, 1,2,3,* and Eli Ruckenstein1 1 Department

of Chemical and Biological Engineering, University at Buffalo,

The State University of New York, Buffalo, NY 14260, USA 2 New

York State Center of Excellence in Materials Informatics, Buffalo, NY 14203, USA

3 Computational

and Data-Enabled Science and Engineering Graduate Program,

University at Buffalo, The State University of New York, Buffalo, NY 14260, USA * Corresponding authors: [email protected], [email protected] Abstract Diol radicals (DRs) are important intermediates in biocatalysis, atmospheric chemistry and biomass combustion. They are particularly generated from photolysis of halogenated diols and a barrierless addition of hydroxyl radical to a double bond of unsaturated alcohols, such as lignols. The energized DRs further isomerize / decompose to form products, including water. Aqueous-phase dehydration in radiolytic and biomimetic systems typically occurs at low temperatures, with or without catalysis, whereas the gas-phase dehydration is usually considered energetically unfavorable. In the present study, we propose a new low-energy, roaming-like mechanism based on a detailed dispersion-corrected DFT and ab initio level analysis of the gas-phase dehydration of DRs obtained from the combination of OH radicals with allyl alcohol (AA, CH2=CHCH2OH) - the simplest relevant model of the unsaturated alcohols. The roaming pathways involve a nearly dissociated OH-group, which subsequently abstracts an H-atom of the remaining fragment to form water and [C3H5O] radical via a transition state (TS) with energy close to the C-O bond fission asymptote. Two types of roaming-like first-order saddle points (SP) are identified for unimolecular dehydration of 1,2- and 1,3-DR radical adducts involving either both hydroxyl groups of diol radicals to generate an oxygen-centered radical, or β-OH-group and a skeletal α–hydrogen atom of the 1,2-DR to form a resonantly stabilized hydroxyallyl radical. Two higher energy conventional (tight) transition states, along with the pathways to 1,2-OH-migration, as well as direct H-abstraction, are also identified and analyzed. Most of the traditional DFT methods that have been successfully employed in literature to locate so-farknown roaming SPs, were also able to identify presented here roaming SPs, in accord with dispersioncorrected double hybrid B2PLYP-D3(BJ) and mPW2PLYPD methods involving MP2-correlation corrections. However, the MP2 method itself failed to locate any of them, which seems to be typical for 1 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

MP2 method for loose TS structures, confirmed here for a flat region of PES connecting direct and roaming saddle points. However, MP2 method correctly locates an identical roaming SP for a larger pcoumaryl alcohol model involving additional hydroxyphenyl substituent at Cγ – atom of AA. Two types of inter-fragmental interactions are identified that stabilize the roaming SPs: (a) H-bonding of the leaving OH-radical either with the H-atom of the remaining OH-group, or with π-cloud of the double bond; (b) direct interaction of π-electrons with the lone-pair electrons of the heteroatom in the leaving OH-group through the TS-ring. The alternative TSs are qualitatively characterized by “collinearity” angle of the OH-radical attack on the O-H/C-H bonds of the substrate in abstraction-like O-H-O geometry, attributed to the improved orbital overlaps. The proposed mechanism presents broader implications to signify, particularly, a larger role in atmospheric and combustion processes, especially biomass pyrolysis. 1. INTRODUCTION Dehydration of mono- and polyhydric alcohols (polyalcohols) is ubiquitous among natural and engineered chemical processes;

1-12

particularly biomass pyrolysis13-20 and combustion,21-26 where radical processes

dominate. Several pyrolysis products of lignins have unsaturated side chains because of dehydration.12,1420

Valuable chemicals are also formed from dehydration of bio-refinery-feedstock polyalcohols, such as

glycerol.8,16,23 The OH radical typically produces dehydration.53-63 It also generates diol radicals, either by abstraction of a skeletal H-atom of diols, or by addition to a double bond of an unsaturated alcohol.22,59 It is also part of the indirect effect of γ-irradiation on biological systems, leading particularly, to the ringopening of the 2-deoxyribose moiety in DNA.5,10 Diol radicals are also generated from photolysis of the halogenated diols and other halohydrins, many of which are carcinogenic and genotoxic food byproducts identified also in water treated with chlorine, chloroamine and combinations of ozone and chlorine.11 Dehydration mechanisms are varied depending on the conditions. The aqueous-phase reactions are typically explained via acid/base-catalysis,5,27-41 although the uncatalyzed dehydration is also known.5 The acid-catalyzed pinacol rearrangement initiated by protonation of a hydroxyl group, for instance, converts 1,2-diols to aldehydes or ketones.31 The naturally occurring regio-specific dehydration of vicinal diols is catalyzed by diol dehydratase enzyme initiated by B12-coenzyme, and involves α,β-dihydroxyalkyl radicals (diol radicals - DR) as intermediates.1,2,20,42-49 In contrast to aqueous-phase processes, the gas2 ACS Paragon Plus Environment

Page 2 of 43

Page 3 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

phase dehydration is typically considered to be energy demanding and unfavorable, even when the intermediates are energized,21,25,53 although a feasible water elimination pathway was recently proposed for alcohol oxidation involving key hydroperoxyalkyl (QOOH) radicals.50,51 Here, we introduce a novel low-energy roaming-like mechanism for the gas-phase dehydration of chemically activated diol-radicals using reaction of OH radical with allyl alcohol (AA, CH2=CHCH2OH) as a simple model of the unsaturated alcohols. The AA (2-propen-1-ol) also replicates the basic structural motif of the lignin monomers – lignols and its side-chains.12,65

Scheme 1. Roaming-like versus conventional pathway for dehydration of diol radicals. We also verify the major roaming pathway for para-coumaryl alcohol (p-CMA, HOPhCH=CHCH2OH), which is one of the three basic lignols explored in our previous studies.12,17 A typical roaming-like mechanism for dehydration of diol radicals is shown in Scheme 1 along with the conventional reaction pathway via tight TS. The 1,3-diol radical is formed from OH-addition to the double bond of AA and further decomposes depending on its initial conformation (cis-gauche or gauchegauche). 3 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 43

Formally, the novel pathways are alternatives to the conventional dehydration reactions (Eqs 1,2 describing formation and dehydration of 1,3-diol and 1,2-diol radicals, respectively), however, they proceed through substantially lower activation barriers due to the loose structures of transition states. CH2=CHCH2OH + HO●  [CH2(OH)CH●CH2OH]* [HO…CH2CH●CH2OH]*[C3H5O]● + H2O

(1)

CH2=CHCH2OH + HO●  [CH2●CH(OH)CH2OH]* [CH2●CH(…OH)CH2OH]*[C3H5O]●+ H2O (2)

The simplest considered here model compounds for 1,2-diol and 1,3-diol radicals are chemically activated 1,2 propane-diol-3-yl (1,3-DR*), and 1,3 propane-diol-2-yl (1,2-DR*) radicals, respectively (asterisks are omitted in future discussions, for simplicity). Various isomers of the product radical [C3H5O]● can be generated from dissociation of the 1,3-DR and 1,2-DR energized adducts (details are provided in Scheme 2, Sec.3.1). In a joint experimental / theoretical study Kamarchik et al. had demonstrated66 that the energized radical [●CH2CH2OH], formed either by photolysis of haloethanols or OH + C2H4 addition reaction,60 undergoes dehydration reaction to form CH2=CH● + H2O products passing through a flat region of PES (explored by CBS-QB3 method) involving the transition complex [CH2=CH2…●OH]. The phenomenon have been characterized as a roaming process.66-70 Since the roaming decomposition of the β-hydroxyethyl radical occurs via H-abstraction of the fairly strong (albeit partly distorted) C-H bond of the ethylene, it was reasonable for us to expect that the same phenomenon can occur with diol radicals, which involve similar and even weaker bonds (and more polarized, as it concerns the second O-H bond dissociation). For comparison, the BDE(O-H) in methanol is 104.6 ± 0.7 kcal/mol, while in ethylene BDE(C-H) = 110.7 ± 0.6 kcal/mol.74 To prove this hypothesis, we undertake a detailed PES study of the unimolecular decomposition of DRs produced by OH+AA reaction, focusing mainly on the dehydration related reaction channels. Another roaming mechanism closely related to the dehydration of diols operates in addition-elimination reaction of chlorine atoms with butenes.75,76 The addition of chlorine atoms to iso-butene is followed by a roaming excursion of Cl and elimination of HCl. This Cl-addition – HCl-elimination pathway has been 4 ACS Paragon Plus Environment

Page 5 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

suggested to proceed from an abstraction-like Cl-H-C geometry rather than a conventional three-center or four-center transition state.75 Quite recently, this roaming mechanism has found new experimental evidence.76 To the best of our knowledge, there are no direct experimental or theoretical studies on roaming dynamics for decomposition of DRs. Therefore, the aforementioned two roaming reaction systems, viz., OH + ethylene and Cl + iso-butene remain the most relevant processes to be compared our results with, in terms of the mechanism and dynamics of roaming. 1.1. Roaming Phenomenon. Roaming is a relatively new and unusual class of reaction mechanisms among unimolecular dissociation reactions of molecules and radicals.64,66-73,75-100 It remains a hot topic following the seminal paper of the Townsend et al. in 2004,78 which had coined the roaming term. The term was assigned to the second molecular mechanism of the photodissociation of formaldehyde to explain the specific energy distribution and vibrational patterns of H2 and CO products. Since then, several new (typically small) molecular and ionic systems have been discovered to involve roaming dynamics.81-84,92,93 A novel molecular mechanism was proposed in our previous study101 for the photochemical dissociation of formaldehyde assisted by the H2 co-product, which leads to the same product set of (CO+H2). Based on our calculations, the dihydrogen catalysis has a relatively low barrier of activation (as low as 64.8 kcal / mol at M06-2X/cc-pVTZ level, compared to 79 kcal/mol for the uncatalyzed dissociation and 86.6 kcal/mol for the radical decomposition).78,83 Intriguingly, the novel TS has a symmetric structure, in contrast to the angular geometry of the tight TS of the molecular dissociation. This allows us to distinguish between vibrational patterns and energy distributions of the different CO and H2 products, which can be probed in roaming experiments.78,80,81 Corresponding experiments that incorporate conditions for product accumulation would also provide the first direct evidence of the proposed dihydrogen catalysis phenomenon. 95,101,123 It should be also noted that a few processes currently defined as roaming, such as the wandering of the NH3-group about alkyl cation in protonated propylamine,99 had been identified much before the term was 5 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

coined, as noted by Harding et al.64 (see also ref.100). The roaming process is generally characterized by a partial cleavage of a chemical bond in a molecule (or a radical) followed by the “roaming” of the two incipient fragments through a flat region of the potential energy surface before they self-react and form products. Typically, the self-reaction of so-called “frustrated” fragments involves H-atom abstraction mediated by a counterpart fragment. A recent quasiclassical dynamics analysis of the roaming “under the microscope”,92-94 particularly highlights that the principal feature of the roaming is that the trajectories enter the plateau region of high potential energy. The new reaction system reported here is believed to involve roaming-like dynamics. A potential energy surface (PES) analysis of the OH radical reaction with allyl alcohol (AA) revealed some unimolecular decomposition channels for diol radical-adducts, which can be well interpreted as roaming. The corresponding first-order saddle-points (SP) share similar traits with other roaming phenomena: (i) “loose” transition state geometries involving an almost dissociated fragment (hydroxyl group in our case) at typical distances 3-3.5 Å, (ii) SP-energies that are very close to the (DR → OH + AA) asymptote, (iii) SPs that are characterized by a low imaginary frequency mode (typically below 200 cm-1) along the reaction coordinate, (iv) SPs occur within a flat region of the PES connecting direct and roaming saddle points, and involving several stationary points of similar energy and structure. Therefore, the nascent OHfragment abstracts an H-atom of the second fragment of the energized DR-adduct and generates thermalized product-radical [C3H5O]● and water. 1.2. Mechanism and Kinetics of the OH+AA Reaction. There are a number of experimental102-112 and a few theoretical106,112 studies on kinetics and mechanism of the model reaction OH+AA. However, the available experiments are mostly focused on the overall kinetics of the reactions, and rate constants, and only a few of them have somewhat analyzed the product formation, and none of them identified the roaming phenomenon. Various isomerization and unimolecular decomposition channels were considered in the most recent and comprehensive theoretical study by Zhang et al.112 at the G2MP2 //MP2/6-311++(d,p) level of theory involving formation and decomposition of both 1,3-DR and 1,2-DR radical adducts (Eqs 3-10). However, 6 ACS Paragon Plus Environment

Page 6 of 43

Page 7 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

only the bond fissions in intermediate radicals were found to be the most affordable reactions, and no potential product formation, including dehydration of radical adducts, was analyzed by authors. CH2=CHCH2OH + OH  [CH2(OH)CHCH2OH / CH2●CH(OH)CH2OH]



 CH2=CHOH + CH2OH

(3)

 CH3CHOH + CH2O

(4)

 CH3CHO + CH2OH

(5)

 CH2OHCHO + CH3

(6)

 CHOHCHOH + CH3

(7)

 CH2CH2OH + CH2O

(8)

 C3H6O2 + H

(9)

 C3H5O + H2O

(10)

Notably, the main product formation channels including the one leading to the glycolaldehyde formation (Eq.6) appear to be high-energy processes. Water elimination (Eq.10) has been simply evaluated via bimolecular (direct) H-abstraction reactions (Eqs 11-14) involving four available types of H-atoms bound to the C1, C2, C3, and O-atoms, respectively, to form CH2=CHCH2O●, CH2=CHC●HOH, CH2=C●CH2OH, and C●H=CHCH2OH intermediate radicals. CH2=CHCH2OH + OH

 CH2=CHCH2O● + H2O

(11)

 CH2=CHC●HOH + H2O

(12)

 CH2=C●CH2OH + H2O

(13)

 CH2=C●CH2OH + H2O

(14)

Note that some other reactions such as the migration of OH-group in the energized adducts, also has not been considered in ref. 112, in addition to the missing unimolecular dehydration pathways discussed in the next sections. Meanwhile, the OH-migration is a known mechanism in biocatalysis;

1,2,42-46

it is also

shown to be an affordable pathway in the atmospheric114 and gas-phase combustion113,115 processes, and can also well occur during the lignin pyrolysis, as we recently demonstrated while studying H + p-CMA chemical activation reactions.12

Indeed, the 1,2-OH shift, provides a feasible switch between two

energized diol radicals, as shown in Fig.5 (Sec.3.3). 7 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

It is important to note also that even the most feasible direct (bimolecular) dehydration channel involving H-abstraction at C1-position of AA to form the most stable α-hydroxy radical (cf., R2 in Fig.1), becomes dominant only at high temperatures, whereas the radical addition and CH2●CH(OH)CH2OH adduct formation from collisional stabilization, prevails at 200-400K temperatures and atmospheric pressure.112 The reaction of allyl alcohol with OH is known to be a pressure dependent process at low pressures (in contrast to OH reactions with other unsaturated alcohols). The PLP-LIF measurements at low pressures in 10-20 Torr argon by Upadhyaya et al.106 provided a rate constant which is ca. 25% lower than that obtained by others.105,109,111 The discrepancy has been explained by difference in P-dependence of the addition channels.112 Rate coefficients for reactions of allyl alcohol and other atmospherically relevant unsaturated alcohols with OH at 231-404K107 display a negative temperature dependence again suggesting that the major pathway is the OH addition to the C=C double bond. Formaldehyde and glycolaldehyde, with molar yields of (98±12%) and (90±12%), respectively, have been found in ref.107 to be the co-products of the OH-addition to AA. Acrolein has also been identified (5±2%) as a product in refs 110 and 111. It should be emphasized, that the types of product and their yields can be altered significantly in pure pyrolysis conditions since almost all of the available experiments on OH+AA reaction are performed in aerobic conditions where the product formation is affected by oxidation reactions. Such reactions are known to be governed by peroxy radicals chemistry (see, e.g., refs 24,51,54), which typically involves the addition of one or two oxygen molecules to a radical intermediate with further transformations of the chemically activated adducts to various oxygenates. Not surprisingly, the formation of dominant product glycolaldehyde has been linked to a peroxy radical pathway.40,104 In order to explain the product distribution in their experiments with OH + allyl alcohol and two other substituted unsaturated alcohols (3-butene-1-ol and 2-methyl-3-butene-2-ol), Takahashi et al. suggested that an initial reaction step is the OH-addition to the terminal C3-atom of the substrate, forming corresponding 1,3-diol radicals (viz., 1,3-DR).104 This is in accord with EPR data by Dixon and Norman that found greater concentration of 1,3-DR radicals than 1,2-DR in reaction of AA with OH radicals 8 ACS Paragon Plus Environment

Page 8 of 43

Page 9 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

generated from Ti3++ H2O2, in acidic solutions.33,34,36-38 This is also consistent with the general view on the dominance of the methylene group reactivity over the substituted carbon atom in olefins like CH2=CHX towards free-radical addition.34 These data also provide evidence on superior reactivity of OHaddition to the double bond versus H-abstraction, since even the most preferred (conjugated) product of the H-abstraction from AA - the radical CH2=CHCH●OH, could not be identified. Recently, De Bruycker et al. suggested that the reactivity of unsaturated alcohols depends on the mutual positions of the C=C bond and OH-group in a hydrocarbon chain. 22 The reactivity of iso-prenol with C=C in γ-position is governed by molecular decomposition to formaldehyde and isobutene. This is opposed to the systems with a C=C double bond in the β- position such as prenol (double methyl-substituted AA), where the radical chemistry dominates and H-abstraction from prenol forms resonantly stabilized radical as dominating conversion path.22 This highlight the radical chemistry of AA being important to explain more complex processes involving more complex biomass components. In the subsequent sections, we report a detailed PES analysis of the most relevant dehydration pathways in OH + AA reaction using DFT and ab initio methods. The employed methods are outlined in Sec. 2. In Sec.3, we provided our detailed PES (DFT) analysis supporting the roaming phenomenon in the diol radicals and large molecules. Since dehydration processes via bimolecular H-abstraction reactions have recently been studied in detail,112 our effort will be focused on the PES for addition-elimination reactions triggered by OH-addition and dehydration of generated DR-adducts. We examined the possible formation of all types of product-radicals. The most roaming-conducive ones are not necessarily the thermodynamically most stable ones. Roaming was identified only for certain reactions: when two OHgroups in both 1,2- and 1,3-DRs are involved in dehydration to form an alkoxy radical; also, when β-OH and a skeletal H-atom at Cα interact in 1,2-DR to generate a resonantly stabilized conjugated product. In Sec.3.1.1 we provide an overview on possible dehydration mechanisms, and evaluate pathways to generate product-radicals. Dehydration channels for of the 1,3-diol radical-adduct (1,3-DR), including a roaming pathway, are detailed in Sec. 3.1.2, whereas the 1,2-DR related issues are addressed in Sec.3.1.3, followed by a general discussion on relevance of the roaming processes (Sec.3.2). The overall PES 9 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 43

analysis is provided in Sec. 3.3, whereas some comparisons with chemical experiments and general implications of proposed mechanisms are reported in Sec.3.4. An overview of methods to predict a roaming stationary point is provided in Supporting Information. 2. METHODS AND COMPUTATIONAL DETAILS Our study is primarily based on Kohn–Sham density functional theory (DFT) using generalized gradient approximation (GGA), hybrid, and double hybrid functionals.120 DFT performance has been extensively studied for various properties and systems, including cases that are in the same problem domain as the presented work.12,

24,54,57,116

Traditional DFT methods, combined with experiments, have been widely

employed for key roaming studies since the phenomenon was discovered.52,66,

73,75,79,84, 93,97,140,148

The

hybrid B3LYP and B3LYP-based composite, BP86 GGA, and B97XD hybrid methods have been particularly used to locate roaming saddle points with surprisingly good agreements with experimental data.52,66,72,73,75,79,89,94,148 The coupled-cluster and/or multireference ab initio methods have been employed for small systems to verify results, in general agreement with DFT data.41,52,84,85,88, 89,91,93,97 Remarkably, we could find no instances in literature on any roaming transition state predicted by MP2 method, which is a common alternative to DFT methods. Since roaming typically occurs when one of the fragments (here OH-radical) is nearly dissociated, we expected the week intermolecular (non-covalent) interactions to play important roles in stabilization of a roaming SP. Hence, the need to choose a computational DFT/ab initio method that best captures these interactions. Based on a recent comprehensive analysis of the DFT performance for benchmark databases which are focused on general main-group thermochemistry, kinetics, and noncovalent interactions, Grimme and co-workers suggested that the dispersion-corrected double hybrid methods (DH), such as B2PLYP-D3(BJ), are best suited ones for non-covalent interactions (ref.117, p.32185). Some dispersioncorrected methods including B97D3 and N12D3 were also highlighted and found to be outperforming MP2 for most of the databases tested (see, also ref.119).

10 ACS Paragon Plus Environment

Page 11 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

In addition to the DFT data, we also provide high-level ab initio wavefunction calculation results to support key findings of our work. These include single point calculations on the stationary points obtained at UB97-D3 level using the coupled cluster theory with singles, doubles, and perturbative triples amplitudes (CCSD(T).120 The latter is considered the “gold standard” of computational quantum chemistry,121,122 largely employed to verify and improve results for open-shell systems and reactions (see, e.g., refs 24,54,123,124). For further comparisons, we employed MP2 wave function method.125 We also monitored T1-diagnostics to assess the degree of multireference character of CCSD(T) results, being generally in a proper range of 0.02-0.03.52 Our analysis of the OH + allyl alcohol PES is based on the following computational protocol: We performed a full geometry optimization using the UB97D3 method,117 in conjunction with the all-electron Pople-type split-valence triple-ζ basis set 6-311+G(2d,p) augmented with diffuse and polarization functions.128 We also employed cc-pVDZ and cc-pVTZ correlation-consistent126 and def2-TZVP basis sets127 to test for basis set dependence. The primary results are provided at UB97D3/6-311+G(2d,p) level of theory, which includes D3-version of Grimme’s dispersion correction with Becke-Johnson damping,129 being consistent with most of the DFT results, particularly those obtained at the popular BP86 and MPW1K138 levels of theory. These results should be at par with CCSD(T)/aug-cc-pV5Z122 as found by Schaefer and co-workers for the symmetric-exchange model reaction OH + H2O → H2O + OH, which is the simplest prototype of the studied in this paper reactions.122 To assess the quality of these baselinelevel results, we reoptimized the geometries using the double hybrid B2PLYPD3(BJ)133 and mPW2PLYPD,130 as well as MP2-related methods (vide infra). We carried out all the DFT and CCSD(T) calculations using the Gaussian 09 (revision D.01) program package,131 utilizing the unrestricted open-shell framework for radicals. We also used QChem 4.4 quantum chemistry package132 to perform some modified MP2 ab initio calculations. The stationary points on the potential energy surface (PES) were verified on the basis of harmonic vibration analysis. Transition states (TS) were characterized as having only one negative eigenvalue in the Hessian matrices. The absence of imaginary frequencies confirms that structures are true minima at their respective levels of 11 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

theory. Our study also employs internal reaction coordinate (IRC) calculations along the reactive mode corresponding to the imaginary frequency of each transition state to ensure the connectivity of all minima and associated saddle points on the PES. All Gaussian calculations were carried out with the standard quadrature ‘‘fine grid’’.117 The validity of the small imaginary frequencies for roaming-type TSs is verified by additional calculations using SuperFineGrid (150,974) and specific grids. The generally good agreement between the results obtained from different DFT methods (see, Table S1, Supporting Information) underscores their robustness and applicability, and thus supports the choices we made in designing our computational protocols. Certain differences between DFT and MP2 results can be explained by extremely flat structure of the PES region and the well-known trend of MP2 to underestimate some key bond distances,72,61,134 as well as some inadequacies in describing potential curves at the TS region.50 An overview of methods to predict a roaming pathway is provided in Supporting Information, where a comprehensive summary of other results is also provided.

3. RESULTS AND DISCUSSION 3.1. Addition-Elimination Reactions. Dehydration Mechanisms of Diol Radicals. OH-addition to the double bond of allyl alcohol generates two types of chemically activated (energized) adduct-radicals: CH2(OH)CH●CH2OH, and CH2●CH(OH)CH2OH, (Eqs 1,2), denoted correspondingly as 1,3-DR (1,3-diol radical) and 1,2-DR (1,2-diol radical). Both radical-adducts contain a pair of OH-groups potentially capable of leaving (to roam). Roaming involves the near dissociation of an OH-group with subsequent abstraction an H-atom of the remaining second fragment (basically the AA molecule) to generate water and a thermalized [C3H5O]● product-radical. Despite the diversity of dehydration mechanisms, the roaming-like saddle points are stabilized only for limited pathways. To identify those pathways, we performed a comprehensive analysis of the PES. All results in this section are obtained at the UCCSD(T)/ 6-311+G(2d,p) //UB97-D3/6-311+G(2d,p) level of theory, unless otherwise stated.

12 ACS Paragon Plus Environment

Page 12 of 43

Page 13 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

3.1.1. Formation of Product-Radicals. Scheme 2 provides an array of possible pathways for unimolecular dehydration of DRs. It involves all possible combinations of OH groups with each other and with skeletal H-atoms to eliminate water to form R1 - R7 isomers of the [C3H5O]● product radical. The mechanisms are specified according to the carbon backbone positions of OH-groups and hydrogen atoms (Fig.1). Some interactions such as (2-1) and (3-1, same as 1-3) between the two OH-groups at C2 and C1, and C3 and C1 positions, respectively, produce the same product-radical (here R1) from different 1,2-DR and 1,3-DR adduct-radicals.

Scheme 2. Addition of hydroxyl radical to the C1 and C2 carbon atoms of the allyl alcohol to form energized 1,2- and 1,3-diol radicals (1,2-DR and 1,3-DR, respectively), and their unimolecular dehydration channels. Notes on arrows specify which hydroxyl groups are involved in a dehydration reaction; leaving (reactive) OH-groups (i) are indicated first; h(i) indicates a skeletal H-atom of the iposition. Roaming pathways are identified only for reactions (1-3), (2-1), and (2-h1) to generate R1 and R2 radicals. The asterisks in the notations for energized adducts are omitted in the text, for simplicity. 13 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Similarly, a dehydration involving an OH group at C2 and a skeletal H-atom at C1 (denoted as 2-h1 in the Scheme 2), and OH at C3 and H-atom at C1 (denoted as 3-h1 and equivalent to 1-h3) also yield the same radical (R2). Notably, the formation energies from alternative pathways differ insignificantly since the isomer adducts are almost isoenergetic (within 1-2 kcal/mol, vide infra, Fig.1). A potential energy diagram on relative stabilities of product-radicals and the key unimolecular dehydration barriers of the DRs is presented in Fig.1. The structures of reagents and relevant stationary points on PES for dehydration of 1,3-DR are illustrated in Fig.2 (Sec.3.1.2), whereas the dehydration of 1,2-DR is discussed later on in the Sec.3.1.3. The well-depths for adduct formation (shown by dotted arrow) are 29.0 and 30.5 kcal/mol for 1,3-DR and 1,2-DR adduct-radicals, respectively. These values are in agreement with similar CCSD(T) data from Zhang et al. based on MP2-geometries, discussed in Introduction.112 Even though all reactions in Scheme 2 produce water and an isomer of [C3H5O]● radical, they are not a family of similar reactions on which to apply empirical rules, such as the Bell-Evans-Polanyi principle, which states that the difference in activation energies between two reactions of the same family is proportional to the difference in their enthalpies of reaction.135,136 The reason is that the TS structures here differ principally. The roaming breaks further the rules. Formation of product-radicals, except for R6, is thermodynamically favored relative to the (OH + AA) entrance level (Fig.1). The product-radicals R2 and R3 are thermodynamically favored even with respect to the thermalized DRs, as follows from Fig.1. Both are highly conjugated resonantly stabilized allylhydroxy radicals. Note also that R2 radical is the simplest model of the highly conjugated phenylsubstituted allyl alcohol radical R(O9) derived from p-CMA (Fig.1) via abstraction of an allylic H-atom by hydrogen atoms, that we have identified previously.12,17 Therefore, on the basis of thermochemistry we expected that the formation of R2 and R3 radicals will be straightforward. However, their generation encounters significant conventional (tight) barriers due to the strained 4-membered-ring TSs. The barrier for R3 formation via TS3, for instance, is almost 20 kcal mol-1 higher than the entrance channel. A high barrier of activation is seen also for the reaction (3-1) via a conventional TS1A transition state to form R1 14 ACS Paragon Plus Environment

Page 14 of 43

Page 15 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

radical (ΔE#=13.1 kcal/mol relative to the entrance level). Thus, formation of R1 - R3 will occur only at very high temperatures.

Figure 1. Energy diagram illustrating formation and major dehydration pathways for energized 1,2-DR and 1,3-DR adducts (Scheme 2). Two tight (conventional) TS1A and TS3 transition states are presented corresponding to reactions (3-1) and (1-h2), respectively, as well as four loose TS1B, TS2B, TS2A, and TS3A transition states leading to the formation of R1 and R2 radicals. The low-energy roaming-like pathways are highlighted in green, for clarity. Relative stabilities of all R1-R7 product-radicals are also provided for comparisons. The key structures involved in the reactions of 1,3-DR via TS1A and TS1B are depicted in Fig.2, whereas the pathways emerged from 1,2-DR are illustrated in Fig.4. Energies are ZPEcorrected, in kcal/mol. However, the radical R1 can alternatively be formed via other low-energy unimolecular channels. Fig.1 shows that the formation of R1 is thermodynamically favored (exothermic by -10.4 kcal/mol relative to the entrance channel) compared to the remaining R4, R5 and R7 radicals (ΔErxn =-5.8, -4.1 and -6.2 15 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

kcal/mol, respectively), as well as the radical R6, the formation of which is even endothermic by 3 kcal/mol. As a result, the unimolecular low temperature formation of radicals R4 - R7 can be safely ignored. The negligible importance of the R4 formation is consonant with the experimental data on slow H-atom abstraction reactions from vinylic C-H bonds in various unsaturated compounds.137 The thermochemistry of R1, R2, R5, and R7 product-radicals is in agreement with G3(MP2) data on direct H-abstraction of the allyl alcohol by OH radicals.112 Some preliminary calculations (not provided) suggested that the formation of hydroxymethyl-vinyl radical (R5) and two cyclic radicals (R4, R6), indeed, encounter high barriers of activation and cannot account for the low-energy dehydration of DRs, thus they are not considered any further. The dehydration barriers for energized DR-adducts depend on the types of product-radicals and the associated steric factors (TS ring-strains) – the accessibility of the reaction centers and effective interfragment interactions. Therefore, the most favored roaming dehydration channels are not the ones necessarily favored by thermochemistry, as seen in Fig.1 (pathways highlighted in green). The dehydration channels involving the combination of both OH-groups to generate R1 seem to be the most relevant processes to stabilize roaming TS. The lowest energy pathways from both adducts involve an oxygen-bound H-abstraction by practically dissociated OH radical (via TS1B and TS2B). While no roaming SP could be located for R3 formation, we have identified a low-energy saddle point TS3A leading to α-hydroxy radical R2 with roaming features (Sec.3.2). 3.1.2. Dehydration of 1,3-Diol Radicals. Fortuitously, the 1,3-DR adduct involving isolated OH-groups, has only a few available dehydration options due to the symmetry when one of the two nearly equivalent C-OH bonds is stretched out. Scheme 2 suggests three possible unimolecular decomposition pathways (two of them are highlighted in green in the PE-diagram, Fig.1), which generate radicals R1, R2, and R7. The key structures for the two alternative conventional and roaming pathways are detailed in Fig.2. The stationary point TS1B can be well characterized as a roaming-like SP with features that are typical, such as the extended distance r(HO-C3) showing the near dissociation of the OH-group at TS (3.05 Å),

16 ACS Paragon Plus Environment

Page 16 of 43

Page 17 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

and a low imaginary frequency of the TS (-139 cm-1).80-84,91 The IRC analysis illustrates the connection between products and diol radicals. The roaming SP, TS1B, originates from CG isomer of the 1,3-DR with cis orientation of the attacked H-O bond relative to C-C-C plane (Fig.2), whereas the high-energy conventional transition state, TS1A, originates from gauche-gauche (GG) isomer with gauche orientation of the attacked H-O bond. In GG 1,3-DR, the two interacting C-OH bonds are almost parallel and expel each other to extend the distance r(H..O) to ca. 2.3Å in comparison to CG 1,3-DR with r(H..O) = 2.1Å. In TS1B, the collapsing Cα-OH (gauche oriented to CCC plane) and intact Cγ-OH (cis-orientated) bonds are located in different planes (Fig.2). To achieve the best overlap of orbitals, the reactive OH-group is required to perform a large excurse (nearly dissociate) to attain a close to collinear orientation of the forming HO-H and the breaking H-Oα bonds (cf. TS1B and TS1A in Fig.2).

Figure 2. Key structures and their ZPE-corrected relative energies in (1-3) dehydration reactions of 1,3DR (Scheme 2, Fig.1), involving cis-gauche (CG) and gauche-gauche (GG) stereoisomers of the radicaladduct, tight (TS1A), and roaming (TS1B) transition state, and corresponding water-complexes of the product-radical R1. The geometry of TS1B is similar to the two roaming SPs discussed in Introduction, for C2H4OH and iso-

17 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

butene systems.66,75 Similar to the roaming TS in Cl + iso-butene reaction,75 the O-H distance in TS1B is about 1.55 Å for the forming HO-H bond with the attacked O-H bond being stretched only to 1.015 Å. In fact, the TS is both “early” and “late” simultaneously since the HO—H bond is in an early stage of its formation (Δro=1.55- 0.96=0.59Å), away from the O-H bond in water molecule, whereas the breaking OγH bond is only somewhat stretched (by 0.042 Å). Meanwhile, the Cα-OH bond is already dissociated. Note that there is an isomeric, almost isoenergetic roaming SP with a different orientation of the reactive OH-group (ν1 = -95.6 cm-1), not provided in Fig.2, for clarity; yet its results are involved in Table S1 in Supporting Information while discussing some methodological issues. The main features of TS1B are reproduced by most of the employed DFT methods including the two double hybrid B2PLYP and mPW2PLYP methods, but not at the MP2 ab initio level. Even though the conventional transition state TS1A also involves a 6-membered ring transition state similar to TS1B, its energy is significantly higher (by over 12 kcal/mol, Fig.1), which can be attributed to its tight structure. The reactive r(OH..C3) distance, in particular, is shorter than that in TS1B by almost one Angstrom (Fig.2). Note that the tight TS1A structure was reproduced by all employed methods including MP2 method (see, Table S2, Supporting Information). Among three available types of skeletal H-atoms in 1,3-DR (Scheme 2), the allylic H-atoms (at C1) are the most reactive ones since their elimination generates resonantly stabilized α-hydroxy radical R2 (CH2CHCH●OH) (Fig.1). Hence, it can be expected that the α-H-abstraction both by γ-OH and β–OH groups would be most feasible. However, the calculations showed that the TS for (3-h1) reaction is not stable and collapses to a VdW-complex. Perhaps, this is due to the rigidity of the allylic moiety against the partial rotation, which is required for the effective interaction of the OH and the reactive H-atom. No saddle point could be located also for the reaction (3-h2) leading to R7-radical formation (Scheme 2) via a 4-member-ring TS. Instead, a VdW-complex between the OH-reagent and β-skeletal H atom of the AA was located, which can lead further to the direct Hβ-abstraction and R7-formation.

18 ACS Paragon Plus Environment

Page 18 of 43

Page 19 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

3.1.3. Dehydration of 1,2-Diol Radicals. In contrast to 1,3-DR, the dehydration of 1,2-diol radical with two adjacent (vicinal) OH-groups, mostly occurs via smaller transition state rings (Fig.3). The transition states particularly involve 5-membered rings when two OH-groups are interacting, and 4-membered rings when a reactive OH abstracts a skeletal H-atom. The ring-strain is expected to increase the barrier heights. Indeed, the reaction (1-h2) involving β-skeletal H-atom abstraction by α-OH group, occurs via 4membered-ring TS3 (Fig.1) and encounters a high barrier of activation (19.7 kcal/mol above the entrance channel). In contrast, a seemingly similar reaction (2-h1), involving abstraction of the α-skeletal H-atom by β-OH group (from C2-center) occurs with a significant reduction of the barrier via expansion of the TS-ring to form a loose, roaming-like TS3A (Fig.3). The “elasticity” of TS3A is likely due to the involvement of the unpaired electrons of the leaving oxygen atom with the π-system of the residue (vide infra). Note that no high energy tight TS could be located for relatively less-strained 5-membered-ring reaction (2-1), as opposed to the 6-membered-ring TS1A for (3-1) reaction (Fig.2). Instead, we obtained two loose saddle points TS2A and TS2B emerging from 1,2-DR. Due to its relaxed structure, the TS2A is only 4 kcal/mol higher than the entrance channel. The TS2B is lower than the entrance level itself by 2.45 kcal/mol. Although both TS2A and TS2B barriers involve 5-membered ring structures and lead to the same R1 + H2O product set, the latter barrier is substantially lower than the first one (by ca. 6.05 kcal/mol) because of the O-H..π stabilizing interactions (facilitated by rotation of the terminal CH2-group), as illustrated in Fig.3 shown by dotted arrow. The H- π interactions partly occurs at the expense of the regular H-bonding interaction with HO-group: a weaker H-bonding in the TS2B and reduced bond-electron density, increases the r(HO…H) distance to 1.70Å compared to that in TS2A structure (1.52Å, Fig.3). Thus, the higher energy of the TS2A vs. TS2B (4.1 kcal/mol above the entrance channel) is due to its tighter structure: a reactive r(O..C2) = 2.80 Å, and r(HO..H) =1.52Å vs. 2.91 Å and1.70Å, respectively. Perhaps, the TS2A can be considered as an underbound version of the conventional TS for reaction (1-2). 19 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 3. Key structures in dehydration of 1,2-DR radical-adduct involving its two (cis-gauche and gauche-gauche) stereoisomers and corresponding tight (TS3) and roaming-like TS3A and isomeric TS2A and TS2B transition states according to reactions (1-h2), (h2-1), (2-1p) and (2-1), respectively. Letter p in the (2-1p) indicates the H…π interaction. Digits represent ZPE-corrected relative energies in kcal/mol (Figs 1 and 5). Intriguingly, all roaming-like (loose) structures (TS2A, TS2B, TS3A) involve elimination of the central βOH-group, as opposed to the TS3, where the terminal OH-group is reactive. Perhaps, the stabilization of

20 ACS Paragon Plus Environment

Page 20 of 43

Page 21 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

these transition states is related to the interactions of the leaving β-OH-group with the π-electron system of the newly forming (exo-cyclic) double bond in R1 and R2 radical-products (Scheme 2). As a result, the unpaired electron, which is initially located on the terminal CH2-group, becomes a part of the π-electron framework in the course of the reaction. The longer-range effect in the loose transition states TS2A, TS2B, TS3A is possibly provided by the lonepair electrons of the O-atom in the leaving OH-group, in contrast to the tight TS3 in which the β-skeletal H-atom is a leaving moiety directly involved in the conjugation, whereas the leaving α-OH-group is isolated by the α-methyl group. Indeed, the collinearity angle φ, which is the measure of the orbital overlap in the reaction center (vide infra) is much smaller for the tight TS3 at 119.3o, at variance to TS3A, TS2A and TS2B for which the φ angle is equal to128.5o, 162.4o and 148.7o, respectively. To check the accuracy of different methods, we extended our calculations with several traditional DFT methods, such as GGA methods BP86-D3, PBE-D3, MN12L, N12, and meta-GGA hybrid method M062X-D3, and so on. Details are reported in the Supporting Information. The roaming-like pathways emerging from 1,2-DR adduct (Fig.1) are predicted only by a group of DFT methods, including B97D3. In particular, the transition states TS2A and TS2B emerging from 1,2-DR are not reproduced by double hybrid B2PLYP and mPW2PLYP and some other methods, including MP2, due perhaps to the too-strained-ring structure of the TS, and need further study. 3.2. How Relevant is the Roaming Dehydration Mechanism. A Qualitative Analysis. The roaming saddle point TS1B for dehydration of chemically activated adduct 1,3-DR (Figs. 2 and 3) is the most relevant transition state for a roaming process predicted by almost all employed here methods including the double hybrid methods (see Table S1, Supporting Information ), therefore in this section it will be analyzed in more detail for comparison purposes. Importantly, all tested methods reproduce the conventional (tight) TSs with similar structures and PES curvature (TableS2). In contrary, the methods diverge in predicting the roaming SPs (DFT) and direct Habstraction TSs (MP2), which have very similar geometries. Remarkably, the dispersion-corrected double

21 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

hybrid DFT method UB2PLYPD3/6-311+G(2d,p) reproduces both types of SPs. This coincides with the fact that it features both DFT and MP2 contributions (see Sec. 2 and 3.2.2). For the comparative qualitative analysis of the roaming and direct dehydration pathways, we thus employed the UB2PLYPD3 method. As expected for a roaming phenomenon, the Cγ-OH covalent bond in TS is almost broken (extended over 3 Å distance, within the van der Waals interaction zone) to generate two associated fragments – a closedshell allyl alcohol (CH2=CHCH2OH) and an HO● radical, which subsequently react to form a productradical (C3H5O●) and water molecule (Scheme 2). The TS1B is also characterized by a low imaginary frequency and the energy within 1-2 kcal/mol of that for the dissociation asymptote, which also are typical for a roaming stationary point. For CH3CHO, the roaming radical transition state is ~1 kcal/mol below the C–C bond fission asymptote,64 whereas the preliminary scan result for CH3…OCH3 fission in dimethyl ether indicates on ~ 2kcal/mol above the roaming saddle point.140 In addition, the TS1B operates through a flat region of PES (vide infra) in accord with literature; in a recent review, Bowman and coworkers have particularly emphasized that among many characteristic features of the roaming, the principal one is that the trajectories enter the plateau region of the high potential energy.92 It is also important to note that the process via TS1B is fairly similar to what occurs with the α-hydroxyethyl radical, C2H4OH●, in a simpler model reaction HO● + C2H4, which is well characterized by numerous experimental kinetics and computational studies as a typical roaming process.66-72 It also resembles the reaction of Cl-atoms with iso-butane also involving non-traditional roaming mechanism.75 3.2.1. Roaming Saddle Point versus Tight Transition State. There has been a dispute on whether the roaming is a pure dynamic process that evades transition state theory (TST) or if it can be described statistically by TST to derive corresponding reaction rate constants. Klippenstein et al. have developed an application of TST to roaming kinetics.64,85,139,140 Bowmen and co-workers have stressed more on the role of dynamic components in the roaming phenomenon (see, e.g., refs 83,84,88). However, in a recent article,92 the latter group have developed a simple kinetic model for dissociation of formaldehyde, and

22 ACS Paragon Plus Environment

Page 22 of 43

Page 23 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

suggested that the TST generally can account for roaming, if it is treated as a separate entity on the PES (here, as a formaldehyde isomer). It was also emphasized that even though roaming follows trajectories that are far from the minimum energy path (MEP), the general nature of a reaction via roaming and via the (tight) TS is similar. The major flux for the conventional TS trajectories occurs at inter-fragment distances around 2.5 Å, whereas that for the roaming trajectories occurs at ca.3.5Å. These data are consonant with our results provided in Fig.4, viz., 1.97Å and 3.13Å C-OH bond distances for conventional and roaming transition states TS1A and TS1B, respectively. Harding et al. explored the separability of roaming and tight transition states141 and proposed a secondorder SP (on the ridge of PES) connecting two first-order SPs, to define an energetic criterion for separability of the two mechanisms. In our case, the two associated tight and roaming-like pathways that lead to the same product-set, are well separated by energy and depend on the initial conformation of the adduct-reagent. An easy to access dehydration of gauche-gauche (GG) stereoisomer of 1,3-DR occurs through a high-energy conventional (tight) barrier TS1A, whereas the roaming saddle point TS1B provides an option for cis-gauche (CG) stereoisomer where the two interacting OH-groups are almost orthogonal to each-other (θ=52o, see, Fig.2). In fact, the roaming process arranges a set of stretched bonds in elastic (expandable) cyclic TS to achieve maximal collinearity of the reactive bonds (vide infra). Note also that all unimolecular dehydration reactions are favored by entropy, however, only the roaming reaction has a positive activation entropy (ΔS# =+4.58 cal/mol.K) due to the loose structure of TS1B. In contrast, the conventional reaction through the tighter transition state TS1A encounters significant decrease of the entropy (ΔS#= -4.63 cal/mol.K). Apparently, positive activation entropy can be seen also for the direct H-abstraction reaction, the transition state of which is similar to that for the roaming reaction: ΔS# = +3.72 and +2.45 cal/mol.K , depending on the orientation of OH-groups in TS (TS4A and TS4B in Fig.4). 3.2.2. Three Alternative Transition States. The roaming and conventional dehydration pathways constitute main alternatives to direct H-abstraction reactions leading to the same set of products. The 23 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

methods employed in current work to identify TS1B are categorized to those predicting only a TS for direct H-abstraction reaction (MP2 and some DFT methods), those predicting only the roaming SP instead (most of the DFT methods, presented in Table 1), and those predicting both. The two employed double hybrid DFT methods, which involve a hybrid-DFT calculation plus an MP2 calculation (using the DFT orbitals), were able to reproduce all channels involving both direct and indirect pathways, namely: two stereo-isomeric direct H-abstraction transition states (TS4A and TS4B), a roaming saddle point (TS1B), and a conventional transition state (TS1A), illustrated in Fig.4. Such results are tempting to explain based simply on the intermediate character of the double hybrid methods, which integrate hybrid DFT with ab initio (MP2) method. In DH methods, parts of the conventional DFT exchange and correlation are replaced by contributions from nonlocal Fock-exchange and second-order perturbative correlation based on the MP2 wave-function. Fig.4 illustrates two direct decomposition and two indirect (addition-elimination) transition states generated from different initial conformations of the diol radicals defined by dihedral angle O-C-C-C, at the double hybrid level of theory. Detailed structures are provided in Supporting Information (Fig. S2). Note that the structure of roaming saddle point at DH level is somewhat looser than that described above by UB97D3 method (cf. Fig.2). The tight transition state, TS1A, is the counterpart of the TS4A direct reaction - both emerged from the GG isomer of the 1,3-DR reagent, whereas the roaming TS1B provides an alternative to the direct reaction via TS4B transition state – both derived from CG isomer of the 1,3DR (Fig.1). Even though, the structure of TS4B is somewhat similar to that for roaming TS1B (Fig.4), the larger imaginary frequency ν1= -1382 cm-1, in the former TS describes it as a bimolecular H-abstraction reaction. The IRC from TS4B, indeed, indicates on a pre-reaction complex of OH with AA reagents, whereas the IRC for roaming saddle point TS1B leads only to the initial diol radical - reagent. This suggests that TS1B is a transition state for a unimolecular dehydration of 1,3DR.

24 ACS Paragon Plus Environment

Page 24 of 43

Page 25 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Figure 4. Comparison of the two direct (TS4A and TS4B) and two indirect (addition-elimination) alternative transition states TS1A and TS1B for dehydration of two stereoisomers of the 1,3-DR to generate the same product set - water and R1-radical (reaction 1-3 in Fig.4). We selected the “collinearity” angle φ (O-H-O angle in Figs 2-4) here as a qualitative (diagnostic) measure of the deviation of roaming TS from an ideal direct H-abstraction TS. The larger φ angle can also suggest larger orbital overlaps, as discussed by Mebel et al.75, 142 A direct H-abstraction by an H(X)-atom is known to occur via a collinear attack (φ=180o), to provide the best orbital overlap.143 However, in case of the OH-attack on either of the H-O and H-C bonds, the φ angle deviates significantly, depending on substituents. For the case of H-abstraction in OH + H2 model reaction we found φ=163.1o at UB2PLYPD3/cc-pVTZ double hybrid DFT level, whereas the reactive bonds are almost collinear (φ=171.8o) when C-H is attacked in H + CH4 reaction. For the symmetric H-abstraction model reaction OH + H2O = H2O + OH we found φ=142.0o, which agrees well with that calculated by Schaefer and coworkers using “golden standard” CCSD(T)/aug-pV5Z method (φ=141.9o).122 Note that the utilization of 6-311+G(2d,p) basis set slightly increases φ angle to 143.1o, whereas B97D3/6-311+G(2d,p) method predicts somewhat larger value φ=146.4o. We observe that the collinearity angle of 143.3o in TS4A for direct H-abstraction in Fig.4 is in good agreement with that for OH + H2O model reaction of similar stereochemistry. The geometry of “loose” transition states are mainly controlled by two types of intermolecular forces: a regular H-bonding with an oxygen atom of the bound OH-group, and interaction of H-atom of the reactive OH-group with π-electrons of the double bond (H-π interactions). At closer distances, the 25 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

covalent interactions of the HO and C2-radical centers dominate through the involvement of OH into an ethylene-oxide-like moiety. Thus, an increase of φ angle to 153.5o in TS4B (cf. 154.8o predicted by MP2/cc-pVDZ) compared to 141.6o in TS4A is attributed to the H-π interactions. The φ angle is further increased in roaming TS1B due to the much looser structure of TS predicted by DH, where the OH-radical does not effectively interact with AA fragment, being away from C3-center at r(HO-H)=1.96Å. The collinear attack of OHgroup is almost ideal in this case with φ =179.9o. The looser structure of TS1B at double hybrid level is also reflected in its lower imaginary frequency (-67.5 cm-1 compared to -138.7 cm-1 and ~-95.6 cm-1 obtained at UB97D3/6-311+G(2d,p) level of theory for TS1B, Fig.2). Consequently, albeit a milder but similar increase of the collinear angle from φ=146.4 to 155.2o is seen in relatively tighter structure of TS1B predicted at hybrid level of theory. 3.2.3. Space Limitations of Roaming Processes. One of the arguments supporting the possibility of the roaming dynamics in OH+AA is that roaming can generally occur also in a limited confinement area, controlled by a part of the remaining fragment. The roaming is known to occur even in a condensed phase.144 The roaming-mediated ultrafast photoisomerization of XBr3 compounds (X=CH, P, B) is the first example of this mechanism in the liquid phase.144 The two Br-atoms of the XBr2 fragment control and limit the space for the roaming Br (and the central X) atom. This makes the Br atom avoid the direct effect of the environment since the process is sufficiently faster than that the mediated by the solvent. Except for the case of the small-size formaldehyde involving the orbiting of the H-atom, the roaming dynamics (rearrangement of fragments) in all other systems are space limited within the week intermolecular interactions zone. The processes typically occur at 3-3.5Å distances of the leaving groups from the remaining fragments of the molecules, which we also see in the case of TS1B. Thus, the OH excursion in energized DR-adducts and abstraction of an H-atom of the AA can also be space-limited and controlled by roaming SPs like TS1B, even if one assumes that the roaming is governed only by dynamic factors and occurs in an off- MEP region of PES.91,149

26 ACS Paragon Plus Environment

Page 26 of 43

Page 27 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

3.3. Overall PES for OH+AA Reaction. Figure 5 provides the low-energy part of the PES for interaction of OH with AA (high-profile PES for decomposition of DRs) explored at dispersion-corrected UB2PLYPD3(BJ)/ 6-311+G(2d,p) double hybrid level. It particularly involves the two key pre-reaction complexes HpibC and HpibG leading to the formation of 1,3-DR and 1,2-DR adducts via transition states TSCC3 and TSGC2, respectively, as well as the formation and isomerization pathways for both DRs. Notably, all transition states are located in a close proximity to each other and to the dehydration stationary points discussed above (cf. Fig.2, Fig.3). 3.3.1. Pre-Reaction Complexes. The PES regions involving VdW-complexes are anticipated to allow the fragments (OH and AA) to spend sufficient time in reasonable proximity and to access the geometries of DRs favorable for internal abstraction via roaming excursion. Our calculations revealed that the lowest dissociation pathway for adduct to form OH and AA fragments, contains double-hydrogen-bonded complexes involving H…O and H…π interactions. These are denoted in Fig.5 as HpibC and HpibG regarding CG and GG-isomers of AA, respectively. The attractive interactions allow the OH radical to roam in a region of the PES near the AA product and potentially abstract an H atom, yielding H2O + [C3H5O●] radical. The direct H-abstraction is exothermic and proceeds via a pre-reaction complex, but encounters a sizable barrier (3.56 and 1.33 kcal/mol relative to HpibG and HpibC equilibrium structures, respectively, Fig.4). The direct H-abstraction barriers for reactions (11-14), evaluated at CCSD(T)//MP2 level by Zhang et al., are even higher than the entrance channel (1.05; 6.18, 6.91 and 6.08 kcal/mol).112 Instead, the OH-addition generates an energized adduct, which easily decomposes to products. TSs for direct H-abstraction (TSGc2 and TSCc3) bifurcate to TS1B (or TS2A and TS2B) saddle points of the similar structure. Indeed, a closer inspection of the partially optimized structures along MEP connecting the HpibC minimum to corresponding DRs obtained from a relaxed scan of the OH-addition to the C2 and C3 skeletal atoms, revealed a mixing of at least three TSs (VRI points). These three TSs, namely TSmig, TS1B and TSCc3 in Fig.4, correspond to 2,3-OH-migration,

27 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

roaming dehydration and OH-addition to the C3-carbon atom, respectively, and have very similar geometry. The interaction of OH and AA particularly bifurcates to TS1B, followed by product formation.

Figure 5. The overall low-energy portion of PES for OH + AA involving formation and dehydration of 1,2-DR and 1,3-DR adducts to generate R2 radical. The major roaming pathway via TS1B is highlighted in red, for clarity. Two additional roaming channels TS2A and TS2B produced by 1,2-DR (highlighted in blue) are also provided for comparisons (see text and Fig.1). A single (green) arrow indicates on H-π stabilizing interaction. A very shallow VdW-minimum is predicted along the addition pathway at DH-levels when a full optimization of an intermediate configuration is performed with OH-radical located orthogonal to the radical plane, on top of the AA, at 5Å distance from C2-center. This shallow minimum, however, is not

28 ACS Paragon Plus Environment

Page 28 of 43

Page 29 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

reproduced by other DFT methods including B97D3. Therefore, a similar optimization at B97D3 level leads to the spontaneous formation of the 1,3-DR and 1,2-DR adducts. Thus, the roaming is facilitated further by flatness of the corresponding PES regions, which along with the attractive VdW interactions forming pre-reaction complexes would provide conditions for roaming dissociation of a DR adduct. 3.3.2. Migration of OH-Group. As mentioned above, there are some other stationary points located in the close proximity of the roaming region. The one with the very close to TS2B geometry is the TSmig for 2,3OH migration leading to the isomerization (interconversion) of 1,2- and 1,3-DRs. The TS is located wellbelow the entrance channel (by 5.78 kcal/mol); it is easily reproduced by all employed methods. Note that all TS-searches for an OH-addition reaction initiated from HpibC pre-reaction complex were collapsing to the transition state TSmig of the very similar geometry. Perhaps, the barriers for OH-addition are insignificant and the 1,3-DR adduct is practically formed in a spontaneous manner. 3.4. Some Implications and Comparisons with Product-Oriented Experiments. Most often the roaming accounts for only a minor fraction of the total products. However, under favorable conditions, it can even be a dominant pathway, as it occurs in case of the unimolecular photochemical decomposition of acetaldehyde.89 To the best of our knowledge, there are no experimental data on gas-phase product formation in the AA + OH reaction in anaerobic conditions to make comparisons with theoretical results. A few available experiments are performed in the presence of O2 (air) and involve oxidation processes, which are typically governed by the rich chemistry of the peroxy radicals (see e.g., refs 24,54,104). Hence, the information on detected products is not relevant for direct comparisons with pyrolysis processes. Some direct information can be obtained from pyrolysis of diol molecules. Laino et al. observed formation of oxy-cyclic compounds while studying pyrolysis of the propylene glycol and triacetin.147 Such structures, indeed, can be formed from high temperature reactions of diol radicals, as suggested in Scheme 2 above. Acrolein was the main product of the reaction OH + AA detected by Atkinson and co-workers.105,145 Its yield has been evaluated as 5.5±0.7% at 296±2 K, using a multiplicative correction factor of 1.25145 to 29 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

account for secondary reactions via OH + acrolein product (for recent data on this reaction see ref.54). A similar yield for acrolein was reported by Orlando et al.107 As an only pathway to generate acrolein, Atkinson and co-workers considered a bimolecular Habstraction of α-H atom by OH radical to form α-hydroxy radical CH2=CH-C●HOH (denoted R2 in Scheme 2 and Fig.1), followed by a reaction with O2 to produce acrolein:105 CH2=CH-CH2OH +OH  H2O + CH2=CH-C●HOH CH2=CH-C●HOH +O2  HO2 + CH2=CH-CHO However, the discussed above theoretical results from Zhang et al. suggested that the bimolecular Habstraction reactions are relatively slow and dominating only at elevated temperatures.112 This is in accord with anticipated minor role of the direct H-abstraction reactions compared to OH radical addition to the C=C bonds.34,105,137 In addition, the experimental rate coefficients provided in refs 110 and 111 show a negative temperature dependence that is also consistent with a reaction mechanism involving the addition of OH to the double bond. Note that Atkinson and co-workers105 have recognized the superior reactivity of OH-addition over H-abstraction, however, they assumed that the first generation products of the OH addition to the C=C double bond of AA are HOCH2CHO + HCHO, and thus concluded that the OH addition cannot account for acrolein formation. Moreover, the products HOCH2CHO + HCHO have not even been detected by authors, considering them amenable to analysis by gas-chromatography. As a result, a low-energy pathway involving a chemically activated addition-elimination channel is required to explain the acrolein formation. The current paper suggests a comprehensive mechanism on addition-elimination in OH + AA reaction leading to diverse products. The low-energy roaming pathways via TS1B and TS2B transition states can particularly provide a perfect explanation for production of acrolein via formation of chemically activated R1 intermediate radical (Scheme 2 and Fig.1), followed by H-elimination or H-abstraction by O2 (relevant to the aerobic conditions of the ref. 105). CH2=CH-CH2OH +OH  [CH2=CH-CH2OH…OH]*  H2O + CH2=CH-CH2O● CH2=CH-CH2O● + O2 HO2 + CH2=CH-CHO

30 ACS Paragon Plus Environment

Page 30 of 43

Page 31 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

Remarkably, the homogeneous oxidation of 1,3-propane diol at low temperatures (400-500K) exclusively (ca. 90% selectivity) produces acrolein and water, as shown by Díaz et al.26 Asmus et al.146 reported that a significant amount of the OH + propane-1,2-diol reaction (20.7%) occurs via abstraction of a primary H-atom from methyl group to form 1,2-DR radical C●H2CH(OH)CH2OH. This, again, suggests an important role of the discussed above dehydration channels in these reactions, to be explored further. Jiang et al. recently revisited reactions of OH●/H● with vicinal diols, particularly, propane-1,2-diol, in water, also assuming that the abstraction of a primary H-of the methyl group is the main pathway to generate 1,2-DR radical, followed by β-fragmentation and formation of ●CH2OH + CH3CHO species.40 Our preliminary calculations at UB97D3 level, however, rule out such a process as highly unlikely and energy demanding since the β-fragmentation via an initial H-transfer from β-OH to C3 position occurs through a barrier of 27.2 kcal/mol height. Moreover, the overall barrier is even higher at 47.5 kcal/mol (including also ca. 20 kcal/mol endothermicity of the R2 radical formation), which makes it prohibitively high - well above the entrance channel. Thus, the actual fate of the intermediate radicals remains elusive. Fig.1 provides alternative scenarios for decomposition of 1,2-DR radical to be explored further. Finally, the reaction of OH with allyl alcohol can serve as a model for more complex systems, such as lignols, and even the lignin macromolecule containing unsaturated side chains linked to hydroxyl groups. To explore the effect of enlargements, we studied the roaming pathway in a lignol model. One of the Hatom in AA was replaced by an electron donor phenyl group to form cinnamyl alcohol (PhCH=CHCH2OH). The electron pumping Ph-substitution further reduces the roaming barrier (TS1B) by almost 2.3 kcal/mol compared to that for the reaction of allyl alcohol. Thus, the electron-donation is supportive and signifies the role of roaming channel in the biomass pyrolysis. 4. Conclusions A new roaming-like mechanism is proposed based on a comprehensive analysis of the dehydration channels in chemical activation reaction of OH-radical with allyl alcohol - a simple model of the 31 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

unsaturated alcohols, using an array of DFT and ab initio methods. We identified several OH-addition – H2O-elimination pathways, which occur via two high-energy conventional (tight) transition states, four low-energy saddle points with roaming features, and a low-energy 2,3-OH-migration TS for interconversions of two 1,2-DR and 1,3-DR adduct radicals. These reactions compete with direct Habstraction reactions. The roaming SPs are characterized by: (a) extremely loose structures involving an almost dissociated OH-fragment; (b) energies just above/below the reagents asymptote, (c) low imaginary frequencies, and (d) flatness of the PES region to easily switch between direct and roaming transition structures. A roaming-like SP can be stabilized depending on the initial conformation of reagents (like in the case of two stereoisomers of 1,3-DR radical-adduct). Two types of intermolecular interactions can affect stabilization of roaming SPs: (a) H-bonding of the leaving OH-radical either with the H-atom of the remaining OH-group, or with π-cloud of the double bond; (b) direct interaction of π-electron fragment (initially involving an unpaired electron) with the lone-pair electrons of the heteroatom in the leaving OHgroup through the TS-ring. Additional arguments in favor of the roaming mechanism and some other conclusions include: i)

The existence of the roaming saddle points on PES is commonly accepted, and the reaction dynamics is considered to operate through such roaming SPs.

ii)

Most of the key experimental studies on roaming phenomenon are accompanied by detailed DFT analysis, in general agreement with high-level ab initio results (including multiconfigurational methods employed for limited systems). Meanwhile, to the best of our knowledge, no MP2 calculation has ever provided any roaming saddle point to signify roaming.

iii)

The roaming SP has intermediate structure between direct and tight TSs on PES, at least for the 1,3-DR radicals studied here, which is demonstrated in Fig.4 at double hybrid B2PLYPD3 level.

iv)

The high-energy conventional (tight) TSs are easy to identify on PES. Therefore, all the methods predicted very similar geometries and fairly close relative energies.

32 ACS Paragon Plus Environment

Page 32 of 43

Page 33 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

v)

We attempted to categorize the employed methods according to their ability to locate a roaming pathway, bearing in mind that the high profile area of PES connecting direct and roaming saddle points is extremely flat, and there is a variety of stationary points with close geometries.

vi)

The methods can be divided into three main groups. The first group involves MP2-related methods predicting the direct H-abstraction pathway and failed to identify roaming SP. In contrary, the second group mostly involving traditional DFT methods, perfectly predicted roaming SPs, but failed to locate direct H-abstraction pathways. The third group involves double hybrid methods predicting both types of TSs.

vii)

Since the roaming area spans intermolecular, van der Waals region, the dispersion interactions appear to be essential.

viii)

Double hybrids with dispersion-corrections give the best performance in our studies, as had been suggested by Grimme et al. based on the comprehensive analysis of the modern DFT methods.117 We additionally found that BP86, PBE-D3, B97D3, MPW1K, and N12 traditional DFT methods are affordable for roaming studies in larger systems. Notably, even MP2 method identifies a roaming SP when a larger model such as OH + p-CMA is employed (Fig.S1, Supporting Information).

Acknowledgments This work is partially funded by National Science Foundation under Grant CBET 1330311. RA acknowledges the Ruckenstein Fund at the University at Buffalo (UB) for the continuous support. JH is supported by start-up funding from the UB School of Engineering and Applied Sciences as well as the Department of Chemical and Biological Engineering. Computing time on the high-performance computing infrastructure “Rush” was provided by the UB Center for Computational Research. Supporting Information Available: An overview of methods to predict roaming pathways; the structures of alternative (direct vs. roaming) transition states; three additionally identified VdW-

33 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

complexes, as well as Cartesian coordinates for optimized stationary points on PES, are provided. This material is available free of charge via the Internet at http://pubs.acs.org. References 1. Speranza, G., Buckel, W. & Golding, B. T. Coenzyme B12-dependent Enzymatic Dehydration of 1,2-Diols: Simple Reaction, Complex Mechanism. J. Porphyrins Phthalocyanines, 2004, 8, 290300. Dehydration of alcohols 2. Toraya T. Radical Catalysis in Coenzyme B12-Dependent Isomerization (Eliminating) Reactions, Chem. Rev. 2003, 103, 2095-2127. 3. Nimlos M.R., Blanksby S.J., Ellison G.B., Evans R.J., Enhancement of 1,2-Dehydration of Alcohols by Alkali Cations and Protons: a Model for Dehydration of Carbohydrates, J. Analyt. Appl. Pyrolysis, 2004, 66, 3-27. 4. von Sonntag C. Free-Radical Reactions of Carbohydrates as Studied by Radiation Techniques, Adv. Carbohydr. Chem. Biochem., 1980, 37, 7-77. 5. Steenken S., Davies M.J., Gilbert B.C. Pulse Radiolysis and Election Spin Resonance Studies of the Dehydration of Radicals from 1,2-Diols and Related Compounds, J. Chem Soc., Perkin Trans. II, 1986, 1003-1010. 6. Komaszylo(née Siedlecka) J., Kania M., Masnyk M., et al.. Isoprenoid Alcohols are Susceptible to Oxidation with Singlet Oxygen and Hydroxyl Radicals, Lipids 2016, 51, 229–244. 7. Shulgin I.L., Ruckenstein E. Preferential Hydration and Solubility of Proteins in Aqueous Solutions of Polyethylene Glycol, Biophys. Chem. 2006, 120, 188-198. 8. Ott L., Bicker M., Vogel H. Catalytic Dehydration of Glycerol in Sub- and Supercritical Water: a New Chemical Process for Acrolein Production, Green Chem., 2006, 8, 214-220. 9. Nolan M.R., Bejile A., Enombo S.-L., Shanks B.H. Directing Polyol Dehydration via Modification of Acid Catalysts with Metals, Topics in Catalysis, 2016, 59, 29–36. 10. Schulte-Frohlinde D. Mechanism of Radiation-Induced Strand Break Formation in DNA and Polynucleotides, Adv. Space Research, 1986, 6, 89-96. 11. Richardson S.D. In: R.A. Meyers (Ed.), Encyclopedia of Environmental Analysis and Remediation, Wiley, 1998, 1398–1421. 12. Asatryan R., Bennadji H., Bozzelli J.W., Ruckenstein E., Khachatryan L. Molecular Products and Fundamentally Based Reaction Pathways in the Gas-Phase Pyrolysis of the Lignin Model Compound p-Coumaryl Alcohol, J. Phys Chem A, 2017, 121, 3352-3371. 13. Westmoreland P.R., Fahey P.J. Dehydration and Dehydrogenation Kinetics of OH Groups in Biomass Pyrolysis, Chem. Eng. Trans, 2016, 50, 73-78. 14. Meier D., Faix O. State of the Art of Applied Fast Pyrolysis of Lignocellulosic Materials - A Review, Biores. Technol. 1999, 68, 71-77. 15. Kibet, J.; Khachatryan, L.; Dellinger B. Molecular Products and Radicals from Pyrolysis of Lignin, Environ. Sci. & Technol. 2012, 46, 12994-13001. 16. Paine III J.B., Pithawalla Y.B., Naworal J.D., Thomas Jr. C.D. Carbohydrate Pyrolysis Mechanisms from Isotopic Labeling Part 1: The Pyrolysis of Glycerin: Discovery of Competing Fragmentation Mechanisms Affording Acetaldehyde and Formaldehyde and the Implications for Carbohydrate Pyrolysis, J. Analyt. Appl. Pyrolysis, 2007, 80, 297-311. 34 ACS Paragon Plus Environment

Page 34 of 43

Page 35 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

17. Xu M., Khachatryan L., Baev A., Asatryan R. Radicals from the Gas-Phase Pyrolysis of a Lignin Model Compound: p-Coumaryl Alcohol, RSC Advances, 2016, 6, 62399-62405. 18. SriBala G., Carstensen H.-H., Van Geem K.M., Marin G.B. Measuring Biomass Fast Pyrolysis Kinetics: State of the Art. WIREs Energy Environ. 2018; e326. https://doi.org/10.1002/wene.326. 19. Khachatryan L., Xu M., Wu A., Pechagin M., Asatryan R. Radicals and Molecular Products from the Gas-Phase Pyrolysis of Lignin Model Compounds. Cinnamyl Alcohol, J. Anal. Appl. Pyrolysis 2016, 121, 75-83. 20. Liu W.-Q., Amara P., Mouesca J.-M., et al. 1,2-Diol Dehydration by the Radical SAM Enzyme AprD4: A Matter of Proton Circulation and Substrate Flexibility, J. Am. Chem. Soc. 2018, 140, 1365−1371. 21. Moc J., Simmie J.M., Curran H.J. The Elimination of Water from a Conformationally Complex Alcohol: A Computational Study of the Gas Phase Dehydration of n-Butanol, J. Mol. Str., 2009, 928, 149-157. 22. De Bruycker R., Herbinet O., Carstensen H.-H., Battin Leclerc F., Van Geem K.M., Understanding the Reactivity of Unsaturated Alcohols: Experimental and Kinetic Modeling Study of the Pyrolysis and Oxidation of 3-Methyl-2-Butenol and 3-Methyl-3-Butenol, Combust. Flame, 2016, 171, 237-251. 23. Nimlos M.R., Blanksby S.J., Qian X., Himmel M.E., Johnson D.K., Mechanisms of Glycerol Dehydration, J. Phys. Chem. A, 2006, 110, 6145-6156. 24. Asatryan R., Bozzelli J.W. Chain Branching and Termination in Low Temperature Combustion of n-Alkanes, J. Phys. Chem. A. 2010, 114, 7693-7708. 25. Sarathy S.M., Oßwald P., Hansen N., Kohse-Hoingghaus K. Alcohol Combustion Chemistry, Prog. Energy & Combust. Sci. 2014, 44, 40-102. 26. Díaz E., Sad M.E., Iglesia E. Homogeneous Oxidation Reactions of Propanediols at Low Temperatures, ChemSusChem, 2010, 3, 1063-1070. 27. Walling C., Johnson R.A. Fenton's Reagent. VI. Rearrangements During Glycol Oxidations, J. Am. Chem. Soc., 1975, 97, 2405. 28. Livingston R., Zeldes H. Paramagnetic Resonance Study of Liquids During Photolysis. III. Aqueous Solutions of Alcohols with Hydrogen Peroxide, J. Am. Chem. Soc. 1966, 88, 4333-4336. 29. von Sonntag C., Thorns E. Strahlenchemie von alkoholen. XV. γ-Radiolyse von Äthylenglykol in wäßriger Lösung Z. Naturforsch., B, 1970, 25, 1405. 30. Burchill C.E., Perron K.M. Radiation-Induced Rearrangement of Ethylene Glycole in Aqueous Solution, Can. J. Chem., 1971, 49, 2382-2389. 31. Bansal K.M., Gratzel M., Henglein A., Janata E. Polarographic and Optical Absorption Studies of Radicals Produced in the Pulse Radiolysis of Aqueous Solutions of Ethylene Glycol, J. Phys. Chem., 1973, 77,16-19. 32. Samuni A., Neta P. Hydroxyl Radical Reaction with Phosphate Esters and the Mechanism of Phosphate Cleavage1, J. Phys. Chem. 1973, 77, 2425-2429. 33. Golding B., Radom L. Facilitation of Intramolecular 1,2-Shifts in Radicals by Protonation, and the Mechanism of Reactions Catalysed by 5‘-Deoxyadenosylcobalamin, J.C.S. Chem. Comm., 1973, 939-941. 34. Dixon W.T., Norman R.O.C. Electron Spin Resonance Studies of Oxidation. Part I. Alcohols, J. Chem. Soc., 1963, 3119-3124. 35. George P., Glusker J.P., Bock C.W., An Ab Initio Computational Molecular Orbital Study of 35 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

36. 37.

38.

39.

40. 41.

42. 43.

44.

45. 46.

47. 48. 49.

50.

51.

Radical, Protonated Radical (Radical Cation), and Carbocation Species That Have Been Proposed in Mechanisms for the Transfer Process in the Enzyme-Coenzyme B12-Catalyzed Dehydration of 1,2-Dihydroxyethane, J. Am. Chem. Soc. 1997, 119, 7065-7074. Buley A.L., Norman R.O.C., Pritchett J. Electron Spin Resonance Studies of Oxidation. Part VIII. Elimination Reactions of Some Hydroxyalkyl Radicals, J. Chem. Soc. B, 1966, 849-852. Gilbert B.C., Norman R.O.C., Williams P.S. Electron Spin Resonance Studies. Part 58. The Formation and Reactions of Some Aliphatic Radical-Cations in Aqueous Solution, J. Chem. Soc., Perkin Trans. 2, 1980, 647-656. Gilbert B.C., Larkin J.P., Norman R.O.C. Electron Spin Resonance Studies. Part XXXIII. Evidence for Heterolytic and Homolytic Transformations of Radicals From 1,2-Diols and Related Compounds, J. Chem. Soc., Perkin Trans. 2, 1972, 794-802. Steenken S. Oxidation of Phenolates and Phenylenediamines by 2-Alkononyl Radicals Produced From 1,2-Dihydroxy- and 1-Hydroxy-2-Alkoxyalkyl Radicals, J. Phys. Chem., 1979, 83, 595599. Jiang D., Barata-Vallejo S., Golding B.T., Ferrerib C., Chatgilialoglu C. Revisiting the Reaction of Hydroxyl Radicals with Vicinal Diols in Water, Org. Biomol. Chem., 2012, 10, 1102-1107. Sivaramakrishnan R., Michael J.V., Harding L.B., Klippenstein S.J. Resolving Some Paradoxes in the Thermal Decomposition Mechanism of Acetaldehyde, J. Phys. Chem. A 2015, 119, 7724−7733. Golding B., Radom L. On the Mechanism of Action of Adenosylcobalamin, J. Am. Chem. Soc., 1976, 98, 6331-6338. Smith D.M., Golding B., Radom L. Toward a Consistent Mechanism for Diol Dehydratase Catalyzed Reactions: An Application of the Partial-Proton-Transfer Concept, J. Am. Chem. Soc. 1999, 121, 5700-5704. George P., Siegbahn Per E. M., Glusker J.P., Bock C.W. The Dehydration Step in the EnzymeCoenzyme-B12 Catalyzed Diol Dehydrase Reaction of 1,2-Dihydroxyethane Utilizing a Hydrogen-Bonded Carboxylic Acid Group as an Additional Cofactor: A Computational Study, J. Phys. Chem. B 1999, 103, 7531-7541. Smith D.M., Golding B., Radom L. Understanding the Mechanism of B12-Dependent Diol Dehydratase: A Synergistic Retro-Push-Pull Proposal, J. Am. Chem. Soc. 2001, 123, 1664-1675. Sandala G.M., Smith D.M., Coote M.L., Golding B., Radom L. Insights into the HydrogenAbstraction Reactions of Diol Dehydratase: Relevance to the Catalytic Mechanism and Suicide Inactivation, J. Am. Chem. Soc. 2006, 128, 3433-3444. Buckel W., Golding B. T. Radical Enzymes in Anaerobes. Ann. Rev. Microbiol., 2006, 60, 27-49. Eda M., Kamachi T., Yoshizawa K., Toraya T. Theoretical Study on the Mechanism of Catalysis of Coenzyme B12-Dependent Diol Dehydratase, Bull. Chem. Soc. Jpn., 2002, 75, 1469–1481. Toraya T., Honda S., Mori K. Coenzyme B12-Dependent Diol Dehydratase Is a Potassium IonRequiring Calcium Metalloenzyme: Evidence That the Substrate-Coordinated Metal Ion Is Calcium, Biochemistry 2010, 49, 7210–7217. Welz O., Klippenstein S.J., Harding L.B., Taatjes C.A., Zádor J. Unconventional Peroxy Chemistry in Alcohol Oxidation: The Water Elimination Pathway, J. Phys. Chem. Lett. 2013, 4, 350−354. Osborn D.L. Reaction Mechanisms on Multiwell Potential Energy Surfaces in Combustion (and Atmospheric) Chemistry, Annu. Rev. Phys. Chem. 2017, 68, 233–260. 36 ACS Paragon Plus Environment

Page 36 of 43

Page 37 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

52. Harding L.B., Klippenstein S.J., Jasper A.W. Ab initio Methods for Reactive Potential Surfaces, Phys. Chem. Chem. Phys., 2007, 9, 4055–4070. 53. Xu S., Lin M.C. Theoretical Study on the Kinetics for OH Reactions with CH3OH and C2H5OH, Proc. Combust. Inst. 2007, 31, 159-166. 54. Asatryan R., da Silva G., Bozzelli J.W. Quantum Chemical Study of the Acrolein (CH2CHCHO) + OH + O2 Reactions, J. Phys. Chem. A 2010, 114, 8302-8311. 55. Huynh, L.K.; Barriger, K.; Violi, A. Kinetics Study of the OH + Alkene → H2O + Alkenyl Reaction Class. J. Phys. Chem. A 2008, 112, 1436−1444. 56. Albarran G., Schuler R.H. Concerted Effects of Substituents in the Reaction of OH Radicals with Aromatics: The Cresols, J. Phys. Chem. A 2005, 109, 9363-9370. 57. da Silva G., Bozzelli J.W., Asatryan R. Hydroxyl Radical Initiated Oxidation of s-Triazine: Hydrogen Abstraction Is Faster than Hydroxyl Addition, J. Phys. Chem. A 2009,113, 8596-8606. 58. Greenwald, E. E.; North, S. W.; Georgievskii, Y.; Klippenstein, S. J. A Two Transition State Model for Radical-Molecule Reactions: Applications to Isomeric Branching in the OH-Isoprene Reaction. J. Phys. Chem. A 2007, 111, 5582−5592. 59. So S., Wille U., da Silva G. Atmospheric Chemistry of Enols: A Theoretical Study of the Vinyl Alcohol + OH + O2 Reaction Mechanism, Environ. Sci. Technol. 2014, 48, 6694−6701. 60. Yamada, T.; Bozzelli, J.W. Kinetic and Thermodynamic Analysis on OH Addition to Ethylene: Adduct Formation, Isomerization, and Isomer Dissociations. J. Phys. Chem. A 1999, 103, 7646– 7655. 61. Zádor, J.; Jasper, A.W.; Miller, J.A. The Reaction Between Propene and Hydroxyl. Phys. Chem. Chem. Phys. 2009, 11, 11040−11053. 62. Vakhtin, A.B.; Lee, S.; Heard, D. E.; Smith, I.W.M.; Leone, S.R. Low-Temperature Kinetics of Reactions of the OH Radical with Propene and 1-Butene Studied by a Pulsed Laval Nozzle Apparatus Combined with Laser-Induced Fluorescence. J. Phys. Chem. A 2001, 105, 7889−7895. 63. Kwok, E.S.C.; Atkinson, R. Estimation of Hydroxyl Radical Reaction Rate Constants for Gasphase Organic Compounds Using a Structure-Reactivity Relationship: an Update. Atmos. Environ. 1995, 29, 1685−1695. 64. Harding L.B., Georgievskii Y., Klippenstein S.J. Roaming Radical Kinetics in the Decomposition of Acetaldehyde, J. Phys. Chem. A 2010, 114, 765–777. 65. Hatakeyama H., Hatakeyama T. Lignin Structure, Properties, and Applications. In: Biopolymers. Adv. Polym. Sci., Abe A., Dusek K., Kobayashi S. (eds) Springer, Berlin, Heidelberg, 2009, 232, 1-63. 66. Kamarchik E., Koziol L., Reisler H., Bowman J.M., Krylov A.I. Roaming Pathway Leading to Unexpected Water + Vinyl Products in C2H4OH Dissociation, J. Phys. Chem. Lett. 2010, 1, 3058. 67. Ratliff B.J., Womack C.C., Tang X.N., Landau W.M., Butler L.J., Szpunar D.E. Modeling the Rovibrationally Excited C2H4OH Radicals from the Photodissociation of 2-Bromoethanol at 193 nm J. Phys. Chem. A 2010, 114, 4934–4945. 68. Ratliff B.J., Alligood B.W., Butler L.J., Lee S.-H., Lin J.J.-M. Product Branching from the CH2CH2OH Radical Intermediate of the OH + Ethene Reaction, J. Phys. Chem. A 2011, 115, 9097–9110. 69. Womack C.C., Booth R.S., Brynteson M.D., Butler L.J., Szpunar D.E. Characterizing the Rovibrational Distribution of CD2CD2OH Radicals Produced via the Photodissociation of 2Bromoethanol-d4, J. Phys. Chem. A 2011, 115, 14559–14569. 37 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

70. Womack CC., Ratliff B.J., Butler L.J., Lee S.-H., Lin J.J.-M. Photoproduct Channels from BrCD2CD2OH at 193 nm and the HDO +Vinyl Products from the CD2CD2OH Radical Intermediate, J. Phys. Chem. A 2012, 116, 6394−6407. 71. McKown B.G., Ceriotti M., Womack C.C., Kamarchik E., Butler L.J., Bowman J.M. Effects of High Angular Momentum on the Unimolecular Dissociation of CD2CD2OH: Theory and Comparisons with Experiment, J. Phys. Chem. A 2013, 117, 10951. 72. Brynteson M.D., Butler L.J. Predicting the Effect of Angular Momentum on the Dissociation Dynamics of Highly Rotationally Excited Radical Intermediates, J. Chem. Phys, 2015, 142, 054301. 73. Matsugi A., Roaming Dissociation of Ethyl Radicals, J. Phys. Chem. Lett. 2013, 4, 4237-4240. 74. Blanksby S.J., Ellison G.B. Bond Dissociation Energies of Organic Molecules, Acc. Chem. Res. 2003, 36, 255-263. 75. Joalland B., Shi Y., Kamasah A., Suits A.G., Mebel A.M. Roaming Dynamics in Radical Addition–Elimination Reactions, Nature Comms., 2014, 5, 4064. 76. Chen L-W., Hung C.-M., Matsui H., Lee Y.-P. New Experimental Evidence to Support Roaming in the Reaction Cl + Isobutene (i-C4H8), Nature, Sci. Reports, 2017, 7, 40105, 1-7. 77. Zhang, X.; Rheinecker, J. L.; Bowman, J.M. Quasiclassical Trajectory Study of Formaldehyde Unimolecular Dissociation: H2CO  H2+CO, H+HCO, J. Chem. Phys. 2005, 122, 114313. 78. Townsend, D.; Lahankar, S.A.; Lee, S. K.; Chambreau, S.D.; Suits, A.G.; Zhang, X.; Rheinecker, J.; Harding, L.B.; Bowman, J.M. The Roaming Atom: Straying From the Reaction Path in Formaldehyde Decomposition, Science 2004, 306, 1158−1161. 79. Kim H.-R. Excursion, Roaming and Migration of Hydrogen Atom during Dissociation of Formaldehyde, Bull. Korean Chem. Soc. 2014, 35, 1285-1292. 80. Suits, A.G. Roaming Atoms and Radicals: a New Mechanism in Molecular Dissociation. Acc. Chem. Res. 2008, 41, 873−881. 81. Bowman, J.M.; Shepler, B.C. Roaming Radicals, Annu. Rev. Phys. Chem. 2011, 62, 531−553. 82. Herath, N.; Suits, A.G. Roaming Radical Reactions, J. Phys. Chem. Lett. 2011, 2, 642−647. 83. Bowman, J.M.; Suits, A.G. Roaming Reactions: The Third Way, Phys. Today, 2011, 64, 33−37. 84. Bowman, J. M. Roaming, Mol. Phys. 2014, 112, 2516−2528. 85. Harding L.B., Klippenstein S.J. Roaming Radical Pathways for the Decomposition of Alkanes, J. Phys. Chem. Lett. 2010, 1, 3016–3020. 86. Maeda S., Taketsugu T., Ohno K., Morokuma K. From Roaming Atoms to Hopping Surfaces: Mapping Out Global Reaction Routes in Photochemistry, J. Am. Chem. Soc. 2015, 137, 3433−3445. 87. Houston P.L., Kable S.H. Photodissociation of Acetaldehyde as a Second Example of the Roaming Mechanism, PNAS, 2006, 103, 16079–16082. 88. Shepler B.C., Braams B.J., Bowman J.M. Quasiclassical Trajectory Calculations of Acetaldehyde Dissociation on a Global Potential Energy Surface Indicate Significant Nontransition State Dynamics, J. Phys. Chem. A, 2007, 111, 8282–8285. 89. Heazlewood B.R., Jordan M.J.T., Kable S.H., Selby T.M., Osborn D.L., Shepler B.C., Braams B.J., Bowman J.M. Roaming is the Dominant Mechanism for Molecular Products in Acetaldehyde photodissociation, PNAS 2008, 105, 12719–12724. 90. Goncharov V., Herath N., Suits A.G. Roaming Dynamics in Acetone Dissociation, J. Phys. Chem. A, 2008, 112, 9423–9428. 38 ACS Paragon Plus Environment

Page 38 of 43

Page 39 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

91. Chen C., Braams B., Lee D.Y., Bowman J.M., Houston P.L., Stranges D. Evidence for Vinylidene Production in the Photodissociation of the Allyl Radical, J. Phys. Chem. Lett. 2010, 1, 1875–1880. 92. Houston P.L., Conte R., Bowman J.M. Roaming Under the Microscope: Trajectory Study of Formaldehyde Dissociation, J. Phys. Chem. A 2016, 120, 5103−5114. 93. Homayoon Z., Bowman J.M. Quasiclassical Trajectory Study of CH3NO2 Decomposition via Roaming Mediated Isomerization Using a Global Potential Energy Surface, J. Phys. Chem. A 2013, 117, 11665−11672. 94. Mauguièrea F.A.L., Collins P., Kramer Z.C., Carpenter B.K., Ezra G.S., Farantos S.C., Wiggins S. Roaming: A Phase Space Perspective, Annu. Rev. Phys. Chem. 2017, 68, 499–524. 95. Asatryan R. Molecular Hydrogen Assisted Transport of H Atoms”, Chem. Phys. Lett. 2010, 498, 263-269. 96. Grubb M.P., Warter M.L., Johnson K.M., North S.W. Ion Imaging Study of NO3 Radical Photodissociation Dynamics: Characterization of Multiple Reaction Pathways, J. Phys. Chem., 2011, 115, 3218–3226. 97. Xiao H., Maeda S., Morokuma K. Excited-State Roaming Dynamics in Photolysis of a Nitrate Radical, J. Phys. Chem. Lett., 2011, 2, 934–938. 98. North S.W. Roaming in the Dark, Nature Chemistry 2011, 3, 504–505. 99. Audier, H.E.; Morton, T.H. Rearrangements in Metastable Ion Decompositions of Protonated Propylamines, Org. Mass Spec. 1993, 28, 1218-1224. 100. van Zee, R.D.; Foltz, M.F.; Moore, C.B. Evidence for a second molecular channel in the fragmentation of formaldehyde, J. Chem. Phys. 1993, 99, 1664−1673. 101. Asatryan R., Ruckenstein E. Dihydrogen Catalysis: A Remarkable Avenue in the Reactivity of Molecular Hydrogen, Catal. Rev.: Sci. Eng., 2014, 56, 403–475. 102. Gordon, S.; Mulac, W.A. Reaction of the OH(X2Π) Radical Produced by Pulse Radiolysis of Water Vapor, Int. J. Chem. Kinet. 1975, Symp. No.1, 289−299. 103. Cometto, P.M.; Dalmasso, P.R.; Taccone, R.A.; Lane, S.I.; Oussar, F.; Daële, V.; Mellouki, A.; Bras, G.L. Rate Coefficients for the Reaction of OH with a Series of Unsaturated Alcohols between 263 and 371 K. J. Phys. Chem. A 2008, 112, 4444. 104. Takahashi K., Hurley M.D., Wallington T.J. Kinetics and Mechanisms of OH-Initiated Oxidation of Small Unsaturated Alcohols, Int. J. Chem. Kinet. 2010, 42, 151–158. 105. Papagni, C.; Arey, J.; Atkinson, R. Rate Constants for the Gas-Phase Reactions of OH Radicals with a Series of Unsaturated Alcohols. Int. J. Chem. Kinet. 2001, 33, 142−147. 106. Upadhyaya, H.P.; Kumar, A.; Naik, P.D.; Sapre, A.V.; Mittal, J.P. Kinetics of OH Radical Reaction with Allyl Alcohol (H2CCHCH2OH) and Propargyl Alcohol (HCCCH2OH) Studied by LIF, Chem. Phys. Lett. 2001, 349, 279−285. 107. Orlando, J.J.; Tyndall, G.S.; Ceazan, N. Rate Coefficients and Product Yields from Reaction of OH with 1-Penten-3-ol, (Z)-2-Penten-1-ol, and Allyl Alcohol (2-Propen-1-ol), J. Phys. Chem. A 2001, 105, 3564−3569. 108. Holloway, A.L.; Treacy, J.; Sidebottom, H.; Mellouki, A.; Daële, V.; Le Bras, G.; Barnes, I. Rate Coefficients for the Reactions of OH Radicals with the Keto/Enol Tautomers of 2,4Pentanedione and 3-Methyl-2,4-Pentanedione, Allyl Alcohol and Methyl Vinyl Ketone Using the Enols and Methyl Nitrite as Photolytic Sources of OH, J. Photochem. Photobiol. A 2005, 176, 183−190. 39 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

109. Parker, J.K.; Espada-Jallad, C. Kinetics of the Gas-Phase Reactions of OH and NO3 Radicals and O3 with Allyl Alcohol and Allyl Isocyanate. J. Phys. Chem. A 2009, 113, 9814−9824. 110. Davis, M. E.; Burkholder, J. B., Rate Coefficients for the Gas-Phase Reaction of OH with (Z)-3Hexen-1-ol, 1-Penten-3-ol, (E)-2-Penten-1-ol, and (E)-2-Hexen-1-ol Between 243 and 404 K, Atmospheric Chem. Phys., 2011,11, 3347-3358. 111. Le Person, A.; Solignac, G.; Oussar, F.; Daele, V.; Mellouki, A.; Winterhalter, R.; Moortgat, G. K.; Le Bras, G. Gas Phase Reaction of Allyl Alcohol (2-Propen-1-ol) with OH Radicals and Ozone, Phys. Chem. Chem. Phys. 2009, 11, 7619−7628. 112. Zhang Y., Chao K., Sun J., Su Z., Pan X., Zhang J., Wang R. Theoretical Study on the Gas Phase Reaction of Allyl Alcohol with Hydroxyl Radical, J. Phys. Chem. A 2013, 117, 6629−6640. 113. Green, Jr. W.H., Wijaya C.D., Yelvington P.E., Raman S. Predicting Chemical Kinetics with Computational Chemistry: is QOOH→HOQO Important in Fuel Ignition? Mol. Phys. 2004, 102, 371. 114. Asatryan R., Bozzelli J.W. Formation of a Criegee Intermediate in the Low-Temperature Oxidation of Dimethyl Sulfoxide, Phys. Chem. Chem. Phys. 2008, 10, 1769-1780. 115. Asatryan R., Raman S., Bielenberg P.A., Peterson B., Bozzelli J.W., Weissman W. Fundamentally Based Mechanism for Oxidation of Iso-Octane. Proceedings, 7th US National Combustion Meeting, Atlanta, GA, USA. 2011, vol.1, 382-395. 116. Hachmann J., Frazier B.A., Wolczanski P.T., Chan G.K.-L. A Theoretical Study of the 3dM(smif)₂ Complexes: Structure, Magnetism, and Oxidation States, ChemPhysChem, 2011, 12, 3236–3244. 117. Goerigk L., Hansen A., Bauer C., Ehrlich S., Najibia A., Grimme S. A Look at the Density Functional Theory Zoo with the Advanced GMTKN55 Database for General Main Group Thermochemistry, Kinetics and Noncovalent Interactions. Phys. Chem. Chem. Phys., 2017, 19, 32184—32215. 118. Asatryan R., Ruckenstein E., Hachmann J. Revisiting the Polytopal Rearrangements in Pentacoordinate d(7)-Metallocomplexes: Modified Berry Pseudorotation, Octahedral Switch, and Butterfly Isomerization, Chem. Sci. 2017, 8, 5512-5525. 119. Goerigk L. Treating London-Dispersion Effects with the Latest Minnesota Density Functionals: Problems and Possible Solutions, J. Phys. Chem. Lett. 2015, 6, 3891−3896. 120. Cramer C.J., Essentials of Computational Chemistry: Theories and Methods, Wiley, New York, 2002. 121. Recent Progress in Coupled Cluster Methods: Theory and Applications, ed. P. Càrsky, J. Paldus and J. Pittner, Springer, 2010. 122. Gao A., Li G., Peng B., Xie Y., Schaefer H.F. The Symmetric Exchange Reaction OH + H2O → H2O + OH: Convergent Quantum Mechanical Predictions, J. Phys. Chem. A 2016, 120, 10223−10230. 123. Asatryan R., Bozzelli J.W., Ruckenstein E. Dihydrogen Catalysis: A Degradation Mechanism for N2-Fixation Intermediates, J. Phys. Chem. A, 2012, 116, 11618-11642. 124. Asatryan R., Ruckenstein E. Effect of "Reducible" Titania Promotion on the Mechanism of HMigration in Pd/SiO2 Clusters, Catal. Lett., 2016, 146, 398–423. 125. Head-Gordon M., Pople J.A., Frisch M.J. MP2 Energy Evaluation by Direct Methods, Chem. Phys. Lett., 1988, 153, 503–506. 40 ACS Paragon Plus Environment

Page 40 of 43

Page 41 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

126. Dunning Jr T.H., Gaussian Basis Sets for Use in Correlated Molecular Calculations. I. The Atoms Boron through Neon and Hydrogen, J. Chem. Phys., 1989, 90, 1007–1023. 127. Schaefer A., Huber C., Ahlrichs R. Fully Optimized Contracted Gaussian-Basis Sets of Triple Zeta Valence Quality for Atoms Li to Kr, J. Chem. Phys., 1994, 100, 5829-35. 128. Binkley J.S., Pople J.A., Hehre W.J. Self-Consistent Molecular Orbital Methods. 21. Small Split-Valence Basis Sets for First-Row Elements, J. Am. Chem. Soc., 1980, 102, 939-947. 129. Grimme S., Ehrlich S. Goerigk L. Effect of the Damping Function in Dispersion Corrected Density Functional Theory, J. Comp. Chem. 2011, 32, 1456-65. 130. Schwabe T., Grimme S. Double-Hybrid Density Functionals with Long-range Dispersion Corrections: Higher Accuracy and Extended Applicability, Phys. Chem. Chem. Phys., 2007, 9, 3397-3406. 131. Gaussian 09, Rev. D.01, M.J. Frisch, G. W. Trucks, H.B. Schlegel, G.E. et al., Gaussian, Inc., Pittsburgh PA, 2013. 132. Shao Y., Gan Z., Epifanovsky E., et al., Advances in Molecular Quantum Chemistry Contained in the Q-Chem 4 Program Package, Mol. Phys., 2015, 113, 184-215. 133. Goerigk L., Grimme S. Efficient and Accurate Double-Hybrid-Meta-GGA Density Functionals—Evaluation with the Extended GMTKN30 Database for General Main Group Thermochemistry, Kinetics, and Noncovalent Interactions, J. Chem. Theory Comput. 2011, 7, 291-309. 134. Nguyen T.N., Putikam R., Lin M.C. A Novel and Facile Decay Path of Criegee Intermediates by Intramolecular Insertion Reactions via Roaming Transition States, J. Chem. Phys. 2015, 142, 124312. 135. Bell, R.P. The Theory of Reactions Involving Proton Transfers, Proc. R. Soc. London, Ser. A 1936, 154, 414–429. 136. Evans, M.G.; Polanyi, M. Some Applications of the Transition State Method to the Calculation of Reaction Velocities, Especially in Solution, Trans. Faraday Soc. 1935, 31, 875–894. 137. Atkinson R.J., Phys. Chem. Ref. Data 1994, Monograph, 2, 1. 138. Lynch, B. J.; Fast, P. L.; Harris, M.; Truhlar, D.G. Adiabatic Connection for Kinetics, J. Phys. Chem. A 2000, 104, 4811. 139. Klippenstein S.J., Georgievskii Y., Harding L.B. Statistical Theory for the Kinetics and Dynamics of Roaming Reactions, J. Phys. Chem. A 2011, 115, 14370–14381. 140. Sivaramakrishnan R., Michael J.V., Wagner A.F., Dawes R., Jasper A.W., Harding L.B., Georgievskii Y., Klippenstein S.J. Roaming radicals in the thermal decomposition of dimethyl ether: Experiment and theory, Combust. & Flame 2011, 158, 618–632. 141. Harding L.B., Klippenstein S.J., Jasper A.W. Separability of Tight and Roaming Pathways to Molecular, J. Phys. Chem. A 2012, 116, 6967−6982. 142. Mebel, A.M., Morokuma, K., Lin, M.C. Ab initio Molecular Orbital Study of Potential Energy Surface for the Reaction of C2H3 with H2 and Related Reactions. J. Chem. Phys. 1995, 103, 3440–3449. 143. Temelso B., Sherrill C.D., Merkle R.C., Freitas Jr. R.A. High-Level Ab Initio Studies of Hydrogen Abstraction from Prototype Hydrocarbon Systems, J. Phys. Chem. A 2006, 110, 1116011173. 144. Mereshchenko A.S., Butaeva E.V., Borin V.A., Eyzips A., Tarnovsky A.N. Roaming-Mediated Ultrafast Isomerization of Geminal Tri-Bromides in the Gas and Liquid Phases, Nature Chem., 2015, 7, 562-568. 41 ACS Paragon Plus Environment

The Journal of Physical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

145. Atkinson, R.; Aschmann, S.M.; Carter, W.P.L.; Winer, A.M.; Pitts, J.N., Jr. Alkyl Nitrate Formation From the Nitrogen Oxide (NOx)-Air Photooxidations of C2-C8 n-Alkanes, J. Phys. Chem. 1982, 86, 4563-4569. 146. Asmus K.-D., Mockel H., Henglein A. Pulse Radiolytic Study of the Site of OH. Radical Attack on Aliphatic Alcohols in Aqueous Solution, J. Phys. Chem. 1973, 77, 1218-1221, 147. Laino T., Tuma C., Moor P., Martin E., Stolz S., Curioni A. Mechanisms of Propylene Glycol and Triacetin Pyrolysis, J. Phys. Chem. A 2012, 116, 4602−4609. 148. Tsai P.-Y., Li H.-K., Kasai T., Lin K.-C. Roaming as the Dominant Mechanism for Molecular Products in the Photodissociation of Large Aliphatic Aldehydes, Phys. Chem. Chem. Phys., 2015, 17, 23112—23120. 149. Rehbein J., Wulff B. Chemistry in motion – off the MEP, Tetrahedron Lett., 2015, 56, 69316943.

42 ACS Paragon Plus Environment

Page 42 of 43

Page 43 of 43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

“TOC Graphic “

43 ACS Paragon Plus Environment