Crossed molecular beam studies of bimolecular reactions of atomic oxygen with nitrogen-bearing organic molecules (nitriles and N-heterocyclic)

Abstract

In this contribution, dedicated to the memory of Prof. Gian Gualberto Volpi, we provide a short review of recent work carried out in our laboratory on reactive scattering studies of the reaction dynamics of atomic oxygen with nitrogen-bearing organic molecules. Specifically, we focus on the polyatomic bimolecular reactions of atomic oxygen, both in the ground and first excited state, O(³P) and O(¹D), with the simplest unsaturated nitriles, namely HCCCN (cyanoacetylene) and CH₂CHCN (cyanoethylene, or acrylonitrile), and with the simplest six-member ring N-heterocyclic compound, pyridine (C₅H₅N). Using the crossed molecular beam (CMB) scattering technique with universal electron-impact ionization mass-spectrometric detection and time-of-flight analysis to measure product angular and velocity distributions, the primary product channels and their branching fractions were determined, thus assessing the central role played by intersystem-crossing (ISC) in this class of reactions. The experimental work was synergistically accompanied by theoretical calculations of the relevant triplet and singlet potential energy surfaces (PESs) to assist the interpretation of experimental results and elucidate the reaction mechanism, including extent of ISC. Cyanoacetylene and cyanoethylene are of considerable interest in astrochemistry being ubiquitous (and relatively abundant) in space including comets and the upper atmosphere of Titan. Being oxygen the third most abundant element in space, the title reactions are of considerable relevance in the chemistry of extraterrestrial environments. In addition, they are also important in combustion chemistry, because thermal decomposition of pyrrolic and pyridinic structures present in bound N-containing fuels generates N-bearing compounds including, in particular, the above two nitriles.

1 Introduction

In 1968 Gian Gualberto Volpi (1928–2017, Linceo from 1994), together with collaborators Giorgio Liuti, Vincenzo Aquilanti, and Franco Vecchiocattivi, founded at the University of Perugia the research group “Elementary Chemical Processes” and began an exciting scientific journey that today still continues in Volpi’s legacy (Boato and Volpi 1999). In 1990, within the various group activities that were inspired by Volpi with enthusiastic support and collaboration, we undertook a broad research program on the investigation of the dynamics of elementary chemical reactions by exploiting a newly built, advanced “crossed molecular beam” (CMB) instrument equipped with a rotatable quadrupole mass-spectrometer (MS) detector and a time-of-flight analysis system (Beneventi et al 1986a, b; Alagia et al. 1995; Casavecchia et al. 1999a, b; Casavecchia 2000). Establishing a research line on experimental reaction dynamics had been one of the main dreams of Volpi ever since he arrived in Perugia. The initial CMB instrument, built in the early 1980s and first employed for high-resolution elastic scattering experiments (Beneventi et al 1986a,b), was optimized for reactive scattering studies in 1990 (Balucani et al. 1991; Alagia et al. 1993a, b, 1995, 1996; Casavecchia et al. 1999a, b; Casavecchia 2000) and continuously updated over the years. In 2004, most notably, a major improvement, consisting in the implementation of soft ionization with tunable energy electrons for product detection (Capozza et al. 2004; Casavecchia et al. 2005; Balucani et al. 2006), permitted us to overcome the main limitation of the method until that time, namely, the problem of serious (elastic/inelastic) interferences in product detection due to dissociative ionization of products, reactants, and background gases, that made very difficult, and often impossible, to identify all the primary reaction products of multichannel bimolecular reactions. This problem had, in particular, severely hampered the application of the CMB-MS method to studies of bimolecular reactions exhibiting a variety of competing product channels, as are most reactions of atomic and molecular radicals with polyatomic molecules occurring in atmospheric, combustion, and astrochemical environments. Among these reactions, those of ground state atomic oxygen, O(³P), with unsaturated hydrocarbons are a most notable example. In fact, although the first kinetic investigation of the O(³P) reaction with the simplest alkene, ethylene, dates back to the mid-1950s (Cvetanović 1955), and various dynamical techniques, including the CMB method with hard (200 eV) electron ionization detection (Schmoltner et al. 1989) and various spectroscopic methods in pump-probe experiments (Endo et al. 1986) were subsequently applied in the following decades, it was not until 2004, with the implementation in the Perugia laboratory of soft electron ionization in CMB experiments with MS detection, that it became possible to probe, on the same footing, all competing product channels of multichannel bimolecular reactions, such as those of O(³P) with acetylene (Capozza et al. 2004) and ethylene (Casavecchia et al. 2005), and to derive the product branching fractions (BFs). The initial studies on O(³P) reactions with 2C unsaturated hydrocarbons (UHCs) (the simplest alkyne and alkene) were subsequently followed by studies involving 3C (alkyne, alkene, and diene) and 4C (alkenes and dienes, including conjugated dienes) UHCs, as well as the simplest aromatic hydrocarbons (benzene and toluene) (see further below).

From a more general perspective, we emphasize that during the 1990s in the field of reaction dynamics, after successful work in several laboratories on a series of 3-atom and 4-atom reactions of great fundamental interest, such as H + D₂/HD (Schnieder et al. 1995; Yuan et al. 2018), F + H₂/D₂/HD (Qiu et al. 2006; Chen et al. 2021), Cl + H₂/D₂/HD (Alagia et al. 1996; Wang et al. 2008), OH + D₂ (Alagia et al. 1993a; Strazisar et al. 2000; Liu et al. 2012), and OH + CO (Alagia et al. 1993b; Laganà et al. 2012), in which the experimental results were usually accompanied by detailed comparisons with the results of quasi-classical trajectory and/or accurate quantum scattering calculations on ab initio potential energy surfaces (PESs), the interest started to shift toward more complex (polyatomic) bimolecular reactions of importance also in areas of practical relevance, such as combustion and astrochemistry (including the atmosphere of Titan). These reactions involved mostly, among others, not only O, C, and N atoms (Casavecchia et al. 1999a, b; Leonori et al. 2008; Balucani et al. 2009) [which are the 3rd, 4th, and 5th most abundant elements in the interstellar medium (ISM)], but also S atoms (Leonori et al. 2009) and CN (Leonori et al. 2010) and OH radicals (Laganà et al. 2012; Liang et al. 2023a).

To improve current kinetic models of the above important environments, detailed information on the involved key elementary, neutral–neutral reactions is required. In particular, knowledge of the identity of the primary reaction products and their BFs is central. A most useful and powerful approach to elucidate the complex mechanisms of multichannel bimolecular reactions is to study them using a synergistic experimental and theoretical approach that can allow us to identify unambiguously all the primary reaction products, determine their BFs, and elucidate the overall reaction mechanism. This was achieved by combining CMB-MS experiments and high-level theoretical calculations of the underlying PESs, as well as dynamical and/or statistical calculations on these PESs of the product BFs as a function of energy (temperature) (Leonori et al. 2015; Caracciolo et al. 2019a; Cavallotti et al. 2022). For O(³P) reactions, pronounced intersystem crossing (ISC) effects were observed in most systems. We remind that the quantum effect of ISC, formally forbidden within the nonrelativistic quantum theory, derives from the fact that the first excited state of the oxygen atom is a ¹D state that correlates with a singlet PES. The latter starts higher in energy and goes much lower than the triplet one. So, a crossing between the two PESs occurs and a transition from the triplet to the singlet PES is possible, enhancing the number of possible products by opening up new reaction channels. ISC effects for 3C- and 4C-UHCs and aromatics were theoretically elucidated using coupled triplet-singlet PESs and statistical Rice-Ramsperger-Kassel-Marcus/Master Equation (RRKM/ME) calculations of product BFs, with inclusion of ISC effects, in collaboration with Carlo Cavallotti (Politecnico of Milan). For some of these reactions, once the theoretical results were found validated by a satisfactory comparison with experimental results in CMB conditions, the theoretical approach was exploited to predict product BFs and, often, also channel-specific rate constants as a function of temperature (Leonori et al. 2015; Caracciolo et al. 2019a; Cavallotti et al. 2022), which is the needed information for improving current kinetic models of combustion as well as of astrochemical environments. Critical in most of our studies on multichannel bimolecular reactions has been the use of soft electron-ionization detection (Casavecchia et al. 2009; Caracciolo et al. 2019a). In other laboratories, soft photo-ionization detection exploiting tunable vacuum-ultra-violet (VUV) synchrotron radiation was used both in pulsed CMB-MS experiments (Lee et al. 2009) and in kinetic experiments (Taatjes et al. 2008; Osborn et al. 2008).

Over the past 20 years or so, we have been using the above strategy of coupling synergistically experiment and theory, initially in collaborations with Joel Bowman (Emory University) for O(³P) + C₂H₄ (Fu et al. 2012a, b; Balucani et al. 2015a) and later on, for more complex reactions, with Carlo Cavallotti (Politecnico of Milan), Marzio Rosi (University of Perugia), and Dimitrios Skouteris (Master-Up, Perugia), to investigate a variety of bimolecular reactions of relevance in combustion chemistry, astrochemistry, and the atmosphere of Titan, such as:

• (i) Reactions of ground state atomic oxygen, O(³P) (also excited O(¹D) in some cases) with unsaturated aliphatic [acetylene (Leonori et al. 2014), ethylene (Fu et al. 2012a, b; Balucani et al. 2015a), allene (Leonori et al. 2012a), propene (Cavallotti et al. 2014; Leonori et al. 2015), propyne (Vanuzzo et al. 2016a, b; Gimondi et al. 2016), 1-butene (Caracciolo et al. 2019a), 1,2-butadiene (Caracciolo et al. 2019b), 1, 3-butadiene (Cavallotti et al. 2022)], and aromatic hydrocarbons [benzene (Cavallotti et al. 2020; Vanuzzo et al. 2021), toluene (Balucani et al. 2024)], the prototype N-heterocyclic, pyridine (Recio et al. 2022), and ubiquitous nitriles [cyanoacetylene (Liang et al. 2023b), cyanoethylene (Pannacci et al. 2023)].

• (ii) Reactions of excited atomic nitrogen, N(²D), with saturated [methane (Balucani et al. 2009), ethane (Balucani et al. 2010)] and unsaturated hydrocarbons [acetylene (Balucani et al. 2000), ethylene (Balucani et al. 2012), propyne (Mancini et al. 2021), allene (Vanuzzo et al. 2022b)], with aromatic hydrocarbons [benzene (Balucani et al. 2023), toluene (Vanuzzo et al. 2024)], pyridine (Recio et al. 2021), cyanoacetylene (Liang et al. 2022), and cyanoethylene (Vanuzzo et al. 2022a).

• (iii) Reactions of carbon atoms, C(³P), and CN radicals with unsaturated hydrocarbons [acetylene (Casavecchia et al. 2001; Leonori et al. 2008, 2010), ethylene (Geppert et al. 2003; Leonori et al. 2012b; Balucani et al. 2015b)]; of S(¹D) with saturated (CH₄) and unsaturated hydrocarbons (C₂H₂, C₂H₄) (Leonori et al. 2009; Berteloite et al. 2011), and of CN with also nitriles [cyanoacetylene (de Aragão et al. 2024), cyanoethylene (Marchione et al. 2022)].

In this contribution, we provide a short review of some of the most interesting results very recently obtained in our laboratory on the dynamics of the multichannel reactions of O(³P) with N-bearing unsaturated hydrocarbons, specifically, the simplest unsaturated nitriles, cyanoacetylene and cyanoethylene (acrylonitrile), and with the simplest 6-member ring, N-heterocyclic molecule, namely, pyridine. As already noted, these systems are of relevance in areas ranging from combustion to astrochemistry, including biology and astrobiology.

2 Experimental method

The basic methodology of crossed molecular beam reactive scattering with mass spectrometric detection is well established (Lee 1987a, b; Herschbach 1987; Casavecchia 2000). The CMB-MS technique is arguably the most powerful experimental method for studying the dynamics of bimolecular reactions at the microscopic level. In fact, MS, either by electron-impact ionization or photo-ionization, is a universal detection method. This is particularly advantageous, actually crucial when studying polyatomic multichannel reactions, because the different products can be probed on the same footing, while this is not possible, for instance, using laser spectroscopy methods. Critical in MS product detection for multichannel reactions is the use of soft ionization, either using tunable low energy electrons (Casavecchia et al. 2005, 2009, 2015) or tunable VUV photons, as can be afforded by third generation synchrotrons (Yang et al. 1997; Lee et al. 2009) or even table-top lasers [in this case the selection of photon energies is limited, but the approach still remains powerful in many cases (Albert and Davis 2013)]. The great advantage of soft ionization in CMB-MS experiments is the gained capability of limiting or, often, even suppressing completely the problem of dissociative ionization of reactants, products, and background gases.

A description of our CMB apparatus, including the implementation of soft electron ionization, has been provided in previous articles and reviews. (Alagia et al. 1995; Casavecchia 2000; Casavecchia et al. 2009, 2010; Caracciolo et al. 2019c). We wish to emphasize that, because absolute electron-ionization (EI) cross sections are usually known or can be reliably estimated for most species, once measured in the laboratory (LAB) the product angular and velocity distributions of the various competing product channels, the use of soft EI allows us to derive the product branching fractions from the observed product number densities and velocities and their conversions into product angular and translational energy distributions in the center-of-mass (CM) reference frame (Casavecchia et al. 2009, 2015).

A most recent technical improvement of our apparatus has seen the replacement of two freon-baffled diffusion pumps (effective pumping speed of 1200 and 1600 l/s) and an old cryopump (3500 l/s, 20 K), used to pump the main, large scattering chamber, with two magnetically suspended turbomolecular pumps (pumping speed 1850 l/s each) backed by a large dry roots pump (110 m³/h) and a new cryopump (3600 l/s, 10 K) (Murray et al. 2020) Although the ultimate vacuum of the main chamber is only slightly improved (by more than a factor of two), the new pumping speed is overall higher and the vacuum cleaner, which reduces the main chamber effusion in the ultra-high-vacuum (UHV) detector (kept below 1 × 10⁻¹⁰ mbar by extensive turbo- and cryo-pumping).

In our CMB apparatus, two supersonic beams of atomic and molecular species are produced with narrow angular and velocity spread and made to cross in a high-vacuum chamber (maintained in the 10⁻⁷ mbar (= 10⁻⁵ Pa) pressure range in operating conditions), typically at 90°, but also 45° or 135° are uniquely possible for reaching lower and higher, respectively, collisions energies while maintaining the same beam characteristics (Balucani et al. 2006; Leonori et al. 2008). Product angular distributions, N(Θ), are measured by modulating the molecular beam at 160 Hz for background subtraction. When measuring TOF distributions, N(Θ,t), a chopper wheel is placed in front of the detector, which is composed of a tuneable electron-impact ionizer followed by a quadrupole mass filter and a Daly type ion detector (Daly 1960). Single-shot TOF is used for beam characterization, while the higher duty-cycle pseudo-random chopping method is used for measuring the product TOF distributions. The ionizer is located in the innermost region of the triply differentially pumped UHV detector chamber. The detector chamber can be rotated in the collision plane, around the axis orthogonal to the plane of the two beams passing through the collision center.

The product N(Θ) and N(Θ,t) distributions are measured in the LAB reference frame, but for the physical interpretation of the scattering results a coordinate transformation from the LAB to the center-of-mass (CM) reference frame is required (Lee 1987b). For each reaction channel the relation between LAB and CM product flux is given by I_LAB(Θ,v) = I_CM(θ,u)v²/u², where Θ and v are the LAB scattering angle and velocity, respectively, while θ and u are the corresponding CM quantities. Since the EI mass-spectrometer detector measures the product number density, N_LAB(Θ,v), rather than the flux, I_LAB(Θ,v), the actual relation between the LAB density and the CM flux is given by N_LAB(Θ) = I_CM(θ,u)v/u². Analysis of the LAB data is performed by forward convolution of trial CM distributions over the experimental conditions (beam divergences in angle and velocity, and detector angular resolution). The CM reactive differential cross section I_CM(θ,u) is commonly factorized into the product of the velocity (or translational energy) distribution, P(u) [or P(E′_T)], and the angular distribution, T(θ): I_CM(θ, E) = T(θ) × P(E′_T). In some cases the coupling between the T(θ) and P(E′_T) functions needs to be accounted for. The T(θ) and P(E′_T) functions contain all the information about the reaction dynamics. When multiple reaction channels contribute to the signal at a given mass-to-charge (m/z) ratio, as in the reaction systems discussed here, a more complex situation arises. In these cases a weighted total CM differential cross section reflecting the possible contributions for a specific m/z value is used in the data analysis of the LAB distributions, that is, I_CM(θ,E′_T) = ∑_iw_i × [T(θ) × P(E′_T)]_i, with the parameter w_i representing the relative contribution of the integral cross section of the ith channel (Casavecchia et al. 2009, 2015). The T(θ) and P(E′_T) functions and the relative weight w_i for each channel are iteratively adjusted until calculated LAB angular distributions and TOF spectra reproduce the experimental ones. Once the T(θ) and P(E′_T) functions for the various product channels are characterized, the branching fraction of each primary product can be estimated by using the procedure introduced by Schmoltner et al. (1989) and widely employed by us in the study of a variety of multichannel reactions of O(³P) with UHCs (Casavecchia et al. 2015; Caracciolo et al. 2019a; Vanuzzo et al. 2021).

2.1 The reactions of O(³P,¹D) with unsaturated nitriles

The simplest unsaturated nitriles, cyanoacetylene (HC₃N) and cyanoethylene (acrylonitrile) (C₂H₃CN), are particularly important in combustion chemistry and astrochemistry. The inclusion of the oxidation processes of cyanoacetylene and acrylonitrile in models that simulate the combustion of coals and other low-rank fossil fuels is important to account for dangerous emissions. This because the nitrogen content of many fuels is essentially ascribed to the presence of pyrrolic and pyridinic structures (Snyder 1969; Brandenburg and Latham 1968; Wallace et al. 1989), but their thermal decomposition generates many nitrogen-bearing compounds, including cyanoacetylene and acrylonitrile (Mackie et al. 1990; Lifshitz et al. 1989; Hore and Russell 1998; Terentis et al. 1992), that can undergo subsequent oxidation to NO_x (Finlayson-Pitts and Pitts 1986). For the above reasons, it is worth investigating the reactions of cyanoacetylene and acrylonitrile with common oxidant species in combustion, including atomic oxygen. Information on the primary products and branching fractions for the reactions O(³P) + HC₃N and O(³P) + C₂H₃CN are expected to be useful for improving combustion models.

In addition, the O(³P) + cyanoacetylene and O(³P) + acrylonitrile reactions are also relevant in the chemistry of the interstellar medium, being these two nitriles ubiquitous in space. In fact, HC₃N was first detected in 1971 in the galactic star-forming region Sgr B2 (Turner 1971) and has since been observed in a variety of interstellar environments, including molecular clouds, solar-type protostars, circumstellar envelopes and external galaxies (Turner 1971; Walmsley et al. 1986; van Dishoeck et al. 1995; Jaber Al-Edhari et al. 2017; Suzuki et al. 1992; Aladro et al. 2011, 2015; Lindberg et al. 2011; Costagliola et al. 2011). HC₃N is also one of the few molecules observed in protoplanetary disks (Chapillon et al. 2012) and it has been detected in cometary comas (Bockelée-Morvan et al. 2000; Hänni et al. 2021; Biver et al. 2015) as well as in the upper atmosphere of Titan, the massive moon of Saturn (Teanby et al. 2007). In addition to being ubiquitous, interstellar HC₃N has a relatively large abundance with respect to H₂ (which is, by far, the most abundant molecule in space). Regarding acrylonitrile, it is the first molecule with a C=C double bond detected in the ISM (Ceccarelli et al. 2017; McGuire 2022). After the first detection in Sagittarius (Sgr) B2 (Ceccarelli et al. 2017), acrylonitrile has been identified in a variety of interstellar environments, such as the hot molecular core Sgr B2(N) (Turner 1971; Walmsley et al. 1986), Orion-KL (van Dishoeck et al. 1995), the TMC-1 dark cloud (Jaber Al-Edhari et al. 2017), the circumstellar envelope of the C-rich star IRC + 10216 (Jaber Al-Edhari et al. 2017), and the L1544 prototypical prestellar core (Aladro et al. 2011), as well as in Titan’s atmosphere (Lindberg et al. 2011; Aladro et al. 2015; Costagliola et al. 2011; Chapillon et al. 2012; Bockelée-Morvan et al. 2000; Hänni et al. 2021; Biver et al. 2015). Being O(³P) quite abundant in various regions of the ISM, its reactions with the above two nitriles might contribute to control the abundance of HCCCN and CH₂CHCN in various extraterrestrial environments, and should be included in astrochemical models, where instead, so far, these reactions have been overlooked.

It should be noted that also atomic oxygen in its first excited state, O(¹D), can have an important role in governing the chemistry of several extraterrestrial environments, being its reactions called into play to elucidate the formation routes of complex organic molecules in cometary comas (Teanby et al. 2007) and interstellar ice (Mumma and Charnley 2011). Therefore, the study of the reaction of cyanoacetylene and acrylonitrile with also O(¹D) can contribute to enrich current astrochemical models.

The goal of our study on the reactions of O(³P) with HC₃N and C₂H₃CN was to provide useful information on the nature of the primary products and their BFs for inclusion in improved astrochemical and combustion models.

We add that, recently, we have investigated the reactions of the above two nitriles not only with atomic oxygen, but also with other radicals that are abundant in extraterrestrial environments where HC₃N and C₂H₃CN have been identified, such as N(²D) (Liang et al. 2022) (Titan and comets) and CN (de Aragão et al. 2020, 2021, 2022) (Titan, interstellar clouds, and comets).

Below, we summarize the main results obtained from our recent CMB and theoretical studies on the dynamics and mechanism of these two reactions.

2.2 O(³P,¹D) + HCCCN (cyanoacetylene)

Despite the relevance of the O(³P) + HC₃N reaction in both astrophysical and combustion environments, there were very few experimental/theoretical studies on this reaction prior to our recent study (Liang et al. 2023b).

According to electronic structure calculations of the relevant triplet and singlet PESs from our recent work (Liang et al. 2023b), and also from previous, less detailed work (Xie et al. 2006), for the O(³P) + HC₃N reaction there are five possible exothermic channels, one nearly thermoneutral, while several others are substantially endothermic. The most exothermic ones, which include the product channels (in bold) found to be dominant in our study, are the following:

$$\begin{array}{*{20}l} {{\mathbf{O}}(^{{\mathbf{3}}} {\mathbf{P}}) + {\mathbf{HCCCN}}} & { \to \,^{{3}} {\text{HCCN}} + {\text{CO}}} & {\Delta {\text{H}}^\circ_{0} = - {249 }( - {257}) \, [ - {253}]\;{\text{kJ/mol}}} & {({\text{1a}})} \\ {} & { \to \,^{{\mathbf{1}}} {\mathbf{HCCN}} + {\mathbf{CO}}} & {{\mathbf{\Delta H}}^\circ_{{\mathbf{0}}} = - {\mathbf{199}} \, ( - {\mathbf{209}}) \, [ - {\mathbf{203}}]\;{\mathbf{kJ}}{/}{\mathbf{mol}}} & {({\mathbf{2a}})} \\ {} & { \to \,^{{3}} {\text{CCO}} + {\text{HCN}}} & {\Delta {\text{H}}^\circ_{0} = - {1}0{9 }( - {115}) \, [ - {112}]\;{\text{kJ/mol}}} & {({\text{3a}})} \\ {} & {\, \to \,{\mathbf{OCCCN}} + {\mathbf{H}}} & {{\mathbf{\Delta H}}^\circ_{{\mathbf{0}}} = - {\mathbf{82}} \, ( - {\mathbf{95}})\;{\mathbf{kJ}}{/}{\mathbf{mol}}} & {({\mathbf{4a}})} \\ {} & {} & {} & {} \\ \end{array}$$

The indicated standard enthalpies of reaction at 0 K, ΔH°₀, are those calculated at the CCSD(T) level and CCSD(T)/CBS level (values in parenthesis) (Liang et al. 2023b). In square brackets are also the values from available enthalpies of formation at 0 K (Ruscic et al. 2004, 2005; Ruscic and Bross 2020).

Figure 1 shows a simplified scheme of the triplet (red lines) and singlet (blue lines) PESs for the reactions O(³P,¹D) + HC₃N, with only the main product channels depicted. In the figure, ISC regions between the triplet and singlet PES are indicated with an ellipse in the entrance channel (ISC_t-s) and with a circle in the exit channel (ISC_s-t). For a more detailed PES portraying also the other energetically open channels (which, however, were not observed to occur), see Fig. 6 in ref. Liang et al. (2023b) and Fig. S1 in its Supporting Information.

Fig. 1

(adapted from Liang et al. 2023b) (color figure online)

Schematic, simplified (see text) triplet (red lines) and singlet (blue lines) potential energy surfaces for the O(³P,¹D) + HC₃N reactions calculated at the CCSD(T)/aug-cc-pVTZ//B3LYP/aug-cc-pVTZ level of theory (adapted from Liang et al. 2023b). The reaction pathways explored by the reactive flux are indicated in bold red and blue solid lines on triplet and singlet PES, respectively; they are the channels 1, 2, and 4; channel (3) leading to ³CCO + HCN products, although more exothermic than channels (4) is not reported due to the presence of an interconversion barrier so high that the BF is negligible (see Liang et al. 2023b). The intersystem crossing (ISC) region where the surface-hopping from the triplet to the singlet PES (ISC_t-s) in the entrance channel is more likely to occur, is indicated with an ellipse. Indicated with a circle is also the possible singlet-to-triplet crossing in the exit channel (ISC_s-t), which has been assumed to be negligible

The corresponding channels (1b)–(4b) for the O(¹D) reaction are more exoergic by 190 kJ/mol (the ³P-¹D energy separation), and therefore are all strongly exothermic:

$$\begin{array}{*{20}l} {{\mathbf{O}}(^{{\mathbf{1}}} {\mathbf{D}}) + {\mathbf{HCCCN}}} & { \to \,^{{3}} {\text{HCCN}} + {\text{CO}}} & {\Delta {\text{H}}^\circ_{0} = - {439 }( - {447}) \, [ - {443}]\;{\text{kJ/mol}}} & {({\text{1b}})} \\ {} & { \to \,^{{\mathbf{1}}} {\mathbf{HCCN}} + {\mathbf{CO}}} & {{\mathbf{\Delta H}}^\circ_{{\mathbf{0}}} = - {\mathbf{389}} \, ( - {\mathbf{399}}) \, [ - {\mathbf{393}}]\;{\mathbf{kJ}}{/}{\mathbf{mol}}} & {({\mathbf{2b}})} \\ {} & { \to \,^{{3}} {\text{CCO}} + {\text{HCN}}} & {\Delta {\text{H}}^\circ_{0} = - {299 }( - {3}0{5}) \, [ - \,{3}0{2}]{\text{ kJ/mol}}} & {({\text{3b}})} \\ {} & { \to \,{\mathbf{OCCCN}} + {\mathbf{H}}} & {{\mathbf{\Delta H}}^\circ_{{\mathbf{0}}} = - {\mathbf{272}} \, ( - {\mathbf{285}})\;{\mathbf{kJ}}/{\mathbf{mol}}} & {({\mathbf{4b}})} \\ \end{array}$$

While channel (2a) can only be formed via ISC from the triplet to the underlying singlet PES in the entrance channel of the reaction, channels (1a) and (3a) (not shown in Fig. 1) can occur only adiabatically on the triplet PES, and channel (4a) can occur on both the triplet and singlet PES. We have searched for all channels (1a–4a)/(1b–4b), but have observed reactive signal only attributable to channels (2a)/(2b), that is, formation of CO + ¹HCCN (cyanomethylene, also termed cyanocarbene) and channels (4a)/(4b), that is, formation of OCCCN (cyanoketyl) + H. We emphasize that channels a and b (from O(³P) and O(¹D), respectively) can be differentiated because of their different thermochemistry (see below).

Reactive scattering signals were measured at m/z = 66 (OCCCN⁺) and m/z = 38 (CCN⁺). Because for m/z = 38 there were interferences from daughter ions of the HC₃N reactant elastically/inelastically scattered by the oxygen seeded beam, the use of soft ionization (28 eV electrons were sufficient) was crucial for obtaining reliable results on the product channels. Notably, the presence of some O(¹D) in our atomic oxygen beam, which is mainly composed by O(³P) (about 90%) (Alagia et al. 1997), permitted us to obtain information on also the O(¹D) + HC₃N reaction dynamics. In Fig. 2a, b we show product angular distributions measured at m/z = 66 (top panel) and m/z = 38 (bottom panel), and the TOF spectra at the same two masses (at the exemplary LAB angle Θ = 28°), respectively. Figure 2c shows the velocity vector diagram of the experiment.

Fig. 2

(adapted from Liang et al. 2023b) (color figure online)

a LAB angular distributions, N(Θ), measured at m/z = 66 (top panel) and m/z = 38 (bottom panel) for the O(³P,¹D) + HC₃N reactions at E_c = 31.1 kJ/mol. The black curves represent the calculated total angular distribution, N(Θ), when the weighted best-fit CM functions of Fig. 3 are used for the O(³P) and O(¹D) contributions to the OC₃N product (top panel) and to the OC₃N and HCCN products (bottom panel). The relative contributions from O(³P) and O(¹D) reactions to m/z = 66 and 38 are indicated and color coded. b Product LAB TOF distributions, N(Θ,t), measured at m/z = 66 (top panel) and m/z = 38 (bottom panel) at the LAB angle Θ = 28° for the reactions O(³P,¹D) + HC₃N at E_c = 31.1 kJ/mol. Open circles: experimental data. Black curves: calculated total TOF distributions when using the weighted best-fit CM functions of Fig. 3 for the O(³P) and O(¹D) contributions to OC₃N and HCCN formation, respectively. The distinct contributions from O(³P) and O(¹D) reactions to the calculated total TOF distributions are also indicated (line and color notations as in a). c Newton (velocity vector) diagram of the experiment. Here the various circles delimit the maximum speed that the indicated products from the O(³P,¹D) + HC₃N reactions at E_c = 31.1 kJ/mol can attain if all the available energy is channeled into product translational energy. Red line: Newton circles for the H-displacement channel leading to OC₃N from the O(³P) reaction. Blue line: same for the O(¹D) reaction. Magenta line: Newton circle for the ¹HCCN product from the O(³P) reaction (via ISC). Violet line: Newton circle for the ³HCCN product from the adiabatic O(³P) reaction. Green line: Newton circle for the ¹HCCN product from the adiabatic O(¹D) reaction

It should be noted that while m/z = 66 corresponds to the parent ion of the heavy coproduct, OC₃N, of channel (4a) and possibly also (4b), m/z = 38 (CCN⁺) corresponds to the (− 1) daughter ion of the HCCN product from channels (1a)/(1b), (2a)/(2b), and possibly also to the (− 28) daughter ion of the OC₃N product from channels (4a)/(4b). HCCN products were detected at the (− 1) daughter ion m/z = 38 rather than at the parent ion mass of 39 for signal-to-noise reasons. The m/z = 66 LAB angular distribution shown in Fig. 2a (top panel) corresponding to the heavy coproduct OC₃N of the H-displacement channels (4a)/(4b) is peaked around the CM angle (Θ_CM = 44.1°) and is quite narrow being confined within a small Newton circle (Fig. 2c). The single peak structure in the TOF spectrum (peaked around 300 μs) (Fig. 2b, top panel) is what is expected for the heavy coproduct of the possible H-displacement channels (4a)/(4b). The contribution of OC₃N from channels (4a)/(4b) is also visible at m/z = 38, through its daughter ion C₂N⁺, in the recorded LAB distributions (see Fig. 2a, bottom panel) and especially in the TOF spectra (Fig. 2b, bottom panel). In order to fit the data at m/z = 66 it was necessary to use two sets of CM functions. These are shown in Fig. 3 (top two panels), and they were associated to the H-displacement channels (4a), leading to OC₃N from O(³P) on the triplet PES, and (4b), leading to OC₃N from O(¹D) on the singlet PES. A possible small contribution of OC₃N from O(³P) via ISC was difficult to evaluate, but it is expected to be small. As can be seen from Fig. 2a (bottom panel) the m/z = 38 LAB angular distribution is characterized by the same prominent peak centered around Θ_CM of the m/z = 66 distribution (top panel), but it is clear that the m/z = 38 distribution is not confined between 18° and 68° as the m/z = 66 distribution, exhibiting significant additional intensity in the two wings. This is confirmed clearly by the TOF spectra at m/z = 38 (Fig. 2b, bottom panel), where it is evident that, in addition to the pronounced peak centered at ca. 300 μs due to the fragmentation of the OC₃N product in the ionizer, there are also two distinct fast peaks, with the fastest one located at ca. 90 μs and the other at ca. 180 μs. The wings of the m/z = 38 angular distribution and the two fast peaks in the m/z = 38 TOF spectra could only be fitted by invoking two additional (with respect to the m/z = 66 data) reactive contributions that, on the basis of energy and linear momentum conservation, can be unambiguously attributed to the HCCN products from the (1a)/(1b) and (2a)/(2b) channels.

Fig. 3

(from Liang et al. 2023b) (color figure online)

Best-fit CM angular, T(θ) (left panels), and translational energy, P(E′_T) (right panels), distributions for all (four) contributions used to reach the best-fit of the experimental data shown in Fig. 2a, b for the O(³P, ¹D) + HC₃N reactions at E_c = 31.1 kJ/mol. Line and color notations are as in Fig. 2. The shaded areas represent the error bars determined for the best-fit CM functions. Solid arrows in right-hand-side (rhs) panels indicate the total available energy (E_TOT = ΔH⁰₀ − E_c) for each specified product channel from the O(³P) and O(¹D) reactions. The average fraction of the total available energy, for each channel, released as product translational energy, $\langle {{\text{f}}_{{\text{T}}} } \rangle$, is also indicated

From the shape of the CM angular distributions of Fig. 3 we can infer that formation of OC₃N and of HCCN from O(³P) proceeds via a long-lived complex mechanism (Levine and Bernstein 1987), while via a strongly osculating complex mechanism from O(¹D), which reflects a substantially shorter intermediate complex lifetime in the case of the strongly exothermic O(¹D) reactions.

The interpretation of the experimental results was assisted by dedicated electronic calculations of the triplet/singlet HC₃ON PESs (see Fig. 1) and statistical RRKM/ME calculations of product BFs, carried out separately on the adiabatic triplet and adiabatic singlet PESs (Liang et al. 2023b).

It was found that the O(³P) reaction dynamics with HC₃N is dominated by ISC from the entrance triplet PES to the underlying singlet PES (see Fig. 1), leading to the spin-forbidden ¹HCCN + CO product channel, with an estimated BF = 0.90 ± 0.05, while the H-displacement channel, formed adiabatically on the triplet PES, is minor, yet substantial, with BF = 0.10 ± 0.05. The experimentally derived BFs are summarized in Table 1 for both O(³P) and O(¹D) reactions. This conclusion for the O(³P) reaction was derived from the observation that adiabatic theoretical calculations predict 98% of H channel on the triplet PES and 99% of CO channel on the singlet PES (see Table 2 in Liang et al. 2023b), which leads to a theoretical BF of 89% and 11% for the CO and H channels from O(³P), respectively, to be compared with the experimental values of 90% and 10% for CO-forming and H-displacement channels, respectively, when assuming an extent of ISC of 90% (Liang et al. 2023b). From our data analysis we reached the conclusion that nearly all HCCN formed from the O(³P) reaction is actually the spin-forbidden excited cyanomethylene, ¹HCCN, from channel (2a) reached via ISC from the triplet to the singlet PES, rather than ground-state ³HCCN from the adiabatic channel (1a). That is, the main reaction channel is the C–C bond breaking channel forming CO and ¹HCCN, and this means that the three-carbon chain of cyanoacetylene is not maintained when attacked by O(³P) (by an extent of 90%), and this is due to efficient triplet to singlet ISC (ISC_t-s) in the entrance channel (see Fig. 1).

Table 1 Product branching fractions (BFs) for the distinct O(³P) + HCCCN and O(¹D) + HCCCN reactions (at E_c = 31.1 kJ/mol) (from Liang et al. 2023b)

However, as discussed in ref. (Liang et al. 2023b) we cannot tell experimentally, within the resolution of our experiment, whether additional singlet to triplet ISC is occurring in the exit channel (ISC_s-t) (see circle in Fig. 1) that could lead to production of ground state cyanomethylene ³HCCN + CO. Answer to the above question could come from a detailed treatment of ISC in both the entrance and exit channel of the O(³P) + HC₃N reaction, but this was outside the scope of our work. Interestingly, however, the excited ¹HCCN product is expected, in any case, to decay spontaneously to ground state ³HCCN in astrophysical environments [because of an expectedly short phosphorescence lifetime (Szczepaniak 2017)] and, ultimately, the main product of the O(³P) + HC₃N reaction would be ground state ³HCCN (+ CO). This is particularly relevant in astrophysical environments, where it is ground-state ³HCCN that has been actually observed. However, also in combustion environments at atmospheric pressure ¹HCCN is expected to be rapidly quenched to ground state ³HCCN.

As shown in Table 1 also for the O(¹D) reaction the dominant product channel is ¹HCCN + CO. The BF of 0.94 ± 0.03 was found to agree with the adiabatic calculated value of 0.97. That is, the O(¹D) reaction proceeds adiabatically on the singlet PES. Using the same arguments already invoked in the discussion of the O(³P) reaction, the ultimate coproduct of CO will be ³HCCN also for the O(¹D) reaction.

Our theoretical results indicated that the main reaction mechanism is addition of O(³P) to the C1 carbon of the triple C≡C bond of HCC-CN, occurring with an entrance barrier of 9 kJ/mol (see Fig. 1), which makes this reaction relevant not only in combustion environments, but also in relatively warm regions of the ISM, such as circumstellar envelopes and PDRs (photon dominated regions), and, possibly, also the upper atmosphere of Titan, where it could represent an efficient mechanism of formation of cyanomethylene. We emphasize that the main product of the O(³P) + HC₃N reaction, cyanomethylene (³HCCN), has been detected toward IRC + 10216 where HC₃N is particularly abundant and O atoms are present (Tenenbaum et al. 2006; Agúndez and Cernicharo 2006; Zhang et al. 2017) as well as in the upper atmosphere of Titan (Martens et al. 2008).

Our study led us to propose to include the O(³P) + cyanoacetylene reaction both as a possible destruction pathway of HC₃N and a possible formation route of HCCN, and also of OCCCN, both in extraterrestrial environments and in the upper atmosphere of Titan. In particular, since both HC₃N and HCCN are present in IRC + 10216, we proposed to search in this environment also for OCCCN (cyanoketyl), which is the other main product of the reaction (see Table 1). We remark that the O(³P) + HC₃N reaction is not present in astrochemical databases, such as KIDA (Wakelam et al. 2012) and UMIST (McElroy et al. 2013). Finally, we remind that nitriles are key intermediates in the formation of species with biological potential, such as nucleobases and aminoacids, both on Earth and in extraterrestrial environments.

2.3 O(³P,¹D) + CH₂CHCN (acrylonitrile)

Despite the significant relevance in both astrophysical and combustion environments, very few experimental/theoretical studies existed prior to our recent work (Pannacci et al. 2023) on the O(³P,¹D) reactions with C₂H₃CN, similarly to the case of the reactions with HC₃N.

According to electronic structure calculations of the relevant triplet and singlet PESs for the O(³P) + C₂H₃CN reactions from our recent work (Pannacci et al. 2023), there are about ten possible exothermic channels, while several others are nearly thermoneutral or substantially endothermic. The most exothermic ones, which include the two channels (in bold) found to be dominant in our study, are the following:

$$\begin{array}{*{20}l} {{\text{O}}(^{{3}} {\text{P}}) + {\text{CH}}_{{2}} {\text{CHCN}}} & { \to \,^{{1}} {\text{CH}}_{{3}} {\text{CN}} + {\text{CO}}} & {\Delta {\text{H}}^\circ_{0} = - {463 }( - {478})\;{\text{kJ/mol}}} & {({\text{5a}})} \\ {} & { \to \,^{{\mathbf{1}}} {\mathbf{CH}}_{{\mathbf{2}}} {\mathbf{CNH}} + {\mathbf{CO}}} & {{\mathbf{\Delta H}}^\circ_{{\mathbf{0}}} = - {\mathbf{348}} \, ( - {\mathbf{365}})\;{\mathbf{kJ}}/{\mathbf{mol}}} & {({\mathbf{6a}})} \\ {} & { \to \,^{{1}} {\text{OCCHCN}} + {\text{H}}_{{2}} } & {\Delta {\text{H}}^\circ_{0} = - {319} {\text{kJ}}/{\text{mol}}} & {({\text{7a}})} \\ {} & { \to \,^{{2}} {\text{CH}}_{{2}} {\text{CN}} + {\text{HCO}}} & {\Delta {\text{H}}^\circ_{0} = - {126 }( - {138})\;{\text{kJ/mol}}} & {({\text{8a}})} \\ {} & { \to \,^{{2}} {\text{COCN}} + \,^{{2}} {\text{CH}}_{{3}} } & {\Delta {\text{H}}^\circ_{0} = - {69} {\text{kJ/mol}}} & {({\text{9a}})} \\ {} & { \to \,^{{\mathbf{2}}} {\mathbf{HCOCHCN}} + {\mathbf{H}}} & {{\mathbf{\Delta H}}^\circ_{{\mathbf{0}}} = - {\mathbf{62}} \, ( - {\mathbf{74}})\;{\mathbf{kJ}}{/}{\mathbf{mol}}} & {({\mathbf{10a}})} \\ {} & { \to \,^{{1}} {\text{CH}}_{{2}} {\text{O}} + \,^{{3}} {\text{HCCN}}} & {\Delta {\text{H}}^\circ_{0} = - {58 }( - {68})\;{\text{kJ/mol}}} & {({\text{11a}})} \\ \end{array}$$

The reported enthalpies of reaction are from electronic structure calculations from Ref. Pannacci et al. (2023) at the CCSD(T)/aug-cc-pVTZ level and at the CCSD(T)/CBS level (values in parentheses) (only for the most relevant channels). Regarding the O(³P) + acrylonitrile reaction, channel (11a) correlates adiabatically with the triplet PES, while channels (5a)–(9a) are expected to occur on the singlet PES after ISC. Channel (10a) can occur both adiabatically on the triplet PES and via ISC on the singlet PES.

Figure 4 depicts a simplified scheme of the triplet (red lines) and singlet (blue lines) PESs for the reactions O(³P,¹D) + CH₂CHCN, where are portrayed only the two most exothermic channels leading to CO formation (5a, 6a), the most exothermic H-displacement channel (10a), and the second most exoergic channel on the triplet PES, that leads to formaldehyde formation (11a). The two bolded product channels are those found to be dominant (Pannacci et al. 2023). In the figure, ISC between the entrance triplet and the underlying singlet PES is indicated with a dotted ellipse. For a more detailed representation of the PESs showing also the other energetically open channels (that, however, were not observed to occur), see Figs. 5 and 6 in Ref. Pannacci et al. (2023).

Fig. 4

(adapted from Pannacci et al. 2023) (color figure online)

Schematic representation of the simplified triplet (in red) and singlet (in blue) potential energy surfaces for the O(³P,¹D) + acrylonitrile reactions with energy evaluated at the CCSD(T)/aug-cc-pVTZ level of theory and at the CCSD(T)/CBS level of theory (data in parentheses) for the most relevant channels (see text). In both cases the zero of the energy is represented by the O(³P) + CH₂CHCN reactants. The pathways leading to the two observed, most important channels (one reached adiabatically on the triplet and the other reached via ISC on the singlet PES) for the O(³P) + CH₂CHCN reaction are shown with thicker lines, together with a possible identification of the region of ISC (black dotted ellipse). For the more detailed PESs, see Pannacci et al. (2023)

The corresponding channels (5b)–(10b) for the O(¹D) reaction, being 190 kJ/mol more exoergic than the corresponding channels from O(³P), are all strongly exothermic:

$$\begin{array}{*{20}l} {{\text{O}}(^{{1}} {\text{D}}) + {\text{CH}}_{{2}} {\text{CHCN}}} & { \to \,^{{1}} {\text{CH}}_{{3}} {\text{CN}} + {\text{CO}}} & {\Delta {\text{H}}^\circ_{0} = - {653 }( - {668})\;{\text{kJ/mol}}} & {({\text{5b}})} \\ {} & { \to \,^{{\mathbf{1}}} {\mathbf{CH}}_{{\mathbf{2}}} {\mathbf{CNH}} + {\mathbf{CO}}} & {{\mathbf{\Delta H}}^\circ_{{\mathbf{0}}} = - {\mathbf{538}} \, ( - {\mathbf{555}})\;{\mathbf{kJ}}{/}{\mathbf{mol}}} & {({\mathbf{6b}})} \\ {} & { \to \,^{{1}} {\text{OCCHCN}} + {\text{H}}_{{2}} } & {\Delta {\text{H}}^\circ_{0} = - {5}0{9} \;{\text{kJ/mol}}} & {({\text{7b}})} \\ {} & { \to \,^{{2}} {\text{CH}}_{{2}} {\text{CN}} + {\text{HCO}}} & {\Delta {\text{H}}^\circ_{0} = - {316 }( - {328})\;{\text{kJ/mol}}} & {({\text{8b}})} \\ {} & { \to \,^{{2}} {\text{COCN}} + \,^{{2}} {\text{CH}}_{{3}} } & {\Delta {\text{H}}^\circ_{0} = - {259}\;{\text{kJ/mol}}} & {({\text{9b}})} \\ {} & {\, \to \,^{{2}} {\text{HCOCHCN}}\, + \,{\text{H}}} & {\Delta {\text{H}}^\circ_{0} = - {252 }( - {264})\;{\text{kJ/mol}}} & {({1}0{\text{b}})} \\ \end{array}$$

As can be seen from Fig. 4 the two CO forming channels can only be formed on the singlet PES, so they are adiabatic for the O(¹D) reaction, while for the O(³P) reaction they can only be formed via ISC (see Fig. 4 for the CO channel). In contrast, the HCO, CH₃, and H forming channels can originate from both the triplet and singlet PES. We probed all the above channels, but reactive signal was observed and attributable only to channels (6a)/(6b) and channels (10a)/(10b), that is, the channels leading to CH₂CNH (ketenimine) + CO and to HCOCHCN + H.

Experimentally, reactive scattering signals were detected at m/z = 68 (C₃H₂NO⁺), 67 (C₃HNO⁺), 66 (C₃NO⁺), 41 (C₂H₃N⁺), 40 (C₂H₂N⁺), 39 (C₂HN⁺), and 29 (HCO⁺). For reasons of S/N ratio, product LAB angular distributions, N(Θ), and TOF distributions, N(Θ, t), were measured in detail only for m/z = 67 and m/z = 40 (using an electron energy of 28 eV), and they are shown in Fig. 5a, b, respectively. Figure 5c depicts, instead, the velocity vector diagram of the experiment, where only the limiting circles of the observed products are indicated. The circle associated to the CH₂O (+ ³HCCN) product from the O(³P) reaction [channel (11a)] was also added to the Newton diagram because useful in the discussion of the results.

Fig. 5

(adapted from Pannacci et al. 2023) (color figure online)

a LAB angular distribution of (top) the HCOCHCN product [channel (10a/10b)] detected at m/z = 67 (C₃HNO⁺), and (bottom) the HCOCHCN and CH₂CNH products detected at m/z = 40 (C₂H₂N⁺) for the O(³P,¹D) + CH₂CHCN reactions at E_c = 31.4 kJ/mol. The filled circles are experimental data (with ± 1σ error bars), while the heavy solid black curve represents the calculated total distribution using the best-fit CM functions shown in Fig. 6. Red, blue, and green solid curves (bottom panel) represent the separate contributions of the HCOCHCN and CH₂CNH products from channel (10a), (6a), and (6b), respectively, to the calculated total LAB angular distribution. The red curve of HCOCHCN from O(³P) in the top panel corresponds to the red curve in the bottom panel. b TOF distributions at Θ = 32° of: (top) HCOCHCN detected at m/z = 67, and (bottom) HCOCHCN and CH₂CNH detected at m/z = 40. Empty circles are experimental data, while the red curve for m/z = 67 and the black curve for m/z = 40 represent the calculated total distribution using the best-fit functions of Fig. 6. (bottom): Red, blue, and green solid curves represent the separate contributions of the labeled products to the total TOF distribution. The red curve of HCOCHCN from O(³P) in top panel corresponds to red curve in the bottom panel. c Newton diagram of the experiment. The radius of each circle represents the maximum velocity that the indicated products HCOCHCN, CH₂CNH, and CH₂O can attain in the CM system if all available energy is channeled into product recoil energy (see text)

As discussed below, from the experimental data we have characterized channels (6a)/(6b) and (10a)/(10b) by measuring angular and TOF distributions at m/z = 67 and 40, respectively, and we have determined their relative yields (BFs). In fact, signal at m/z = 68, 67, and 66 was attributed to the H-displacement channel, dominantly from the O(³P) reaction. This because the theoretical predictions of the BFs on the singlet PES clearly indicate a very small contribution of the HCOCHCN + H reactive channel (10b) (BF = 1.2% at the collision energy of the experiment) (see Pannacci et al. 2023). Therefore, we have neglected, within our experimental sensitivity (i.e., BF $\le$ 2–3%), the reactive channel (10b). The contribution of HCOCHCN from O(³P) via ISC is very hard to evaluate, but it is expected to be small and, therefore, it was not considered in the data analysis. The measurements on the H-displacement channel were performed at the (− 1) daughter ion (i.e., m/z = 67) for reasons of S/N ratio. On the other hand, possible contribution at this m/z value from channels (7a)/(7b) (corresponding to H₂ elimination) was ruled out on the basis of the shape of the angular and TOF distributions at this m/z ratio (see Pannacci et al. 2023).

As can be seen in Fig. 5a (bottom panel), in contrast to the m/z = 67 angular distribution (top panel), the angular distribution at m/z = 40 is characterized by a prominent peak centered at around Θ_CM = 45°, superimposed on two additional wings. The central peak corresponds to a daughter ion of the HCOCHCN product coming from the H-displacement channel (10a), whose presence is also evident from the slow broad peak in the corresponding m/z = 40 TOF spectrum (see Fig. 5b, bottom panel), which is identical to that observed at m/z = 67 (Fig. 5b, top panel). The two additional wings in the angular distribution and the fast peaks in the TOF spectra at m/z = 40 can be fitted by invoking two other reactive contributions associated with the CO-forming channels (6a) and (6b), observed via the daughter ion C₂NH₂⁺ of ketenimine (CH₂CNH). The assignment of the signal registered at m/z = 40 to ketenimine was made on the basis of the topology of the singlet PES and considering the RRKM/ME predictions of product BFs on the singlet PES (see Pannacci et al. 2023). The situation is similar to what observed in the case of the O(³P,¹D) reaction with HC₃N (see Sect. 3.1). In fact, by analyzing the TOF spectra (Fig. 5b), it should be noted that the fingerprint of the CH₂CNH + CO channel from the reaction of O(³P) and O(¹D) is particularly clear at the reported angle Θ = 32°. The fact that the angular distribution of CH₂CNH [channels (6a) and (6b)] is much wider than that of HCOCHCN (channel (10a) and that the TOF features associated to the two contributions of CH₂CNH is much faster than that of HCOCHCN, is due to two reasons: (i) the main one is that CH₂CHCN is scattered over a much larger Newton circle (see Fig. 5c) than HCOCHCN because of linear momentum conservation, being its coproduct, CO, much heavier than H; (ii) additionally, channels (6a) and (6b) are much more exothermic than channel (10a). The calculated best-fit LAB angular and TOF distributions for the indicated channels in Fig. 5a, b are obtained from the best-fit CM product angular and translational energy distribution shown, together with their error bounds, in Fig. 6.

Fig. 6

Best-fit product angular, T(θ), (left) and translational energy, P(E′_T), (right) distributions in the CM reference system for the O(³P,¹D) + acrylonitrile reactions. The shaded areas represent the error bars determined for the best-fit CM functions. The vertical arrow in the graphs of P(E′_T) indicates the total energy (E_TOT = ΔH⁰₀ − E_c) for the corresponding reactive channel. The average fraction of the total available energy, for each channel, released as product translational energy, $\langle {{\text{f}}_{{\text{T}}} } \rangle$, is also indicated (from Pannacci et al. 2023)

Adiabatic statistical calculations of product BFs for the decomposition of the main triplet and singlet intermediates were carried out (Pannacci et al. 2023). Combining experimental and theoretical results, it was concluded that the O(³P) reaction leads to two main product channels, among a variety of possible open channels: (i) CH₂CNH (ketenimine) + CO (dominant, BF = 0.87 ± 0.05), formed via efficient ISC from the entrance triplet PES to the underlying singlet PES, (ii) HCOCHCN + H (minor, BF = 0.13 ± 0.05), occurring adiabatically on the triplet PES. The experimentally derived BFs for both O(³P) and O(¹D) reactions are reported in Table 2. As can be seen the derived BFs for O(³P) + acrylonitrile show strong preference for formation of CH₂CNH + CO.

Table 2 Product branching fractions (BFs) for the distinct O(³P) + acrylonitrile and O(¹D) + acrylonitrile reactions (at E_c = 31.4 kJ/mol) (from Pannacci et al. 2023)

Notably, both ketenimine and acrylonitrile have been detected toward the same star-forming regions (as in Sagittarius B2(N) hot core) (Gardner and Winnewisser 1975). Ketenimine is often neglected by astrochemical modelers in the most common astrochemical databases, such as KIDA (Wakelam et al. 2012) and UMIST (Woodall et al. 2007), with respect to its more abundant CH₃CN (acetonitrile) isomer, and also the reaction O(³P) + acrylonitrile is overlooked in models.

It should be noted that the most exothermic of all channels, more than channel (6a), namely, channel (5a) forming CO + CH₃CN, is statistically highly unfavored with respect to the isomeric channel (6a) leading to CO + ketenimine because of the much higher interconversion barrier (¹TS4) located at − 138 kJ/mol (with respect to reactants) when compared to ¹TS3 located at − 235 kJ/mol, that is about 100 kJ/mol lower (see Fig. 4). This is another example where the detailed reaction dynamics, which is dictated by the detailed features of the PES, determines what the most probable product channel actually is, against the common, simplistic and often unwarranted sense often used in models of taking as dominant product channel the most exothermic one.

Our study suggests to include the O(³P) + acrylonitrile reaction both as a possible destruction pathway of CH₂CHCN and a possible formation route of CH₂CNH. The O(¹D) + CH₂CHCN reaction mainly leads to formation of CH₂CNH + CO adiabatically on the singlet PES (see Table 2). This result can improve models related to the chemistry of interstellar ices and cometary comas, where O(¹D) reactions are believed to play an important role. Overall, our results are expected to be useful to improve models of combustion and extraterrestrial environments.

2.4 O(³P,¹D) + C₆H₅N (pyridine)

Recent CMB experiments backed by theoretical calculations of the reaction PESs and of product distributions on multichannel O(³P) reactions with a variety of unsaturated hydrocarbons containing 2C, 3C, and 4C atoms (simple alkynes, alkenes, dienes, and conjugated dienes) and with simple aromatic hydrocarbons (benzene, toluene) have shown that ISC between the weakly coupled entrance triplet and underlying singlet PESs plays a central role in the dynamics and kinetics of these reactions (Casavecchia et al. 2015; Pan et al. 2017; Caracciolo et al. 2019c; Vanuzzo et al. 2021; Cavallotti et al. 2022). These O(³P) reactions were all found to mainly proceed via an addition–elimination mechanism and to exhibit from a small to a sizeable entrance energy barrier (from a few kJ/mol up to 20 kJ/mol). In all these reactions the electrophilic addition of O(³P) to the unsaturated bond(s) leads to a covalently bound triplet diradical intermediate(s) in the triplet PES. Since triplet and singlet PESs always cross in these systems, a nonadiabatic jump (ISC) can occur from the entrance triplet PES to the underlying singlet PES. It was found that ISC takes place near the minimum of the initial triplet intermediate (more than one intermediate can be present when three or more C atoms are present in the molecule), which is reached only after the triplet entrance barrier has been surmounted. In these systems, ISC mainly affects the number of exit channels and the BFs of the various product channels. The extent of ISC usually increases with increasing lifetime of the triplet intermediate, which, in turn, typically increases with increasing stability of the diradical and lowering of the collision energy (Pan et al. 2017). Interestingly, when we studied the reaction of O(³P) with one of the simplest heterocyclic molecules, namely pyridine, which is isoelectronic with benzene, and compared the two systems, we observed a hitherto unobserved ISC-based mechanism (Recio et al. 2022). The study of the O(³P) + pyridine reaction was motivated by the importance of this system in several fields including biology, combustion, and astrochemistry (Wang et al. 2012; Parker et al. 2015; Puzzarini and Barone 2020). Specifically, we found that ISC does occur for ortho-, meta-, and para-addition of the O atom to the carbons of the aromatic ring of pyridine, after the corresponding entrance barrier has been surmounted, as in the case of the reaction of O(³P) with the isoelectronic molecule of benzene. However, for these types of addition the extent of ISC was found to be small, while it was found that ISC can also occur, much more efficiently, for O atom addition to the heteroatom, specifically, to the lone pair of the nitrogen atom (ipso addition) before the entrance barrier. Figure 7 shows a simplified, schematic representation of the triplet/singlet PESs for meta and ipso addition of O(³P) to pyridine. In the figure we emphasize the minimum energy crossing point (MECP) for the two cases of ipso and meta addition (MECP_i and MECP_m, respectively), and portray schematically only the pathways to the main product channels of the O(³P) + C₅H₅N reaction, namely, the strongly exothermic spin-forbidden CO elimination channel, involving ring-contraction with formation of pyrrole + CO, occurring on the singlet PES reached from the triplet PES via ISC, and the H-displacement channel occurring adiabatically on the triplet PES or, nonadiabatically (via ISC), on the singlet PES, and leading to the weakly exothermic pyridoxyl + H products. The fact that ISC is much more efficient for ipso addition than for meta addition is due to the fact that the MECP for ISC in the case of ipso addition is located at an energy (8.0 kJ/mol) lower than the entrance barrier (16.5 kJ/mol) for meta addition (the most favourable addition site to the ring C among the three possible ortho, meta, and para sites) (Fig. 7). The CMB experiments were supported by detailed theoretical work, consisting in high-level, quantum electronic structure calculations of the triplet and singlet PESs and related RRKM/ME simulations of product BFs, with the inclusion of ISC for both meta (the preferred attack position on the C sites) and ipso addition (to the N atom) (Recio et al. 2022). The theoretical calculations of the triplet and singlet PESs for the O(³P,¹D) + pyridine reaction indicated that the main energetically allowed product channels at E_c = 37.2 kJ/mol for the O(³P) reaction are (in bold are the two main observed channels):

Fig. 7

(adapted from Recio et al. 2022) (color figure online)

Simplified representation (schematic) of the triplet (red lines) and singlet (blue lines) PESs of the O(³P,¹D) + pyridine reaction for O(³P) meta and ipso addition. a Addition of O(³P) on the C atom in meta position (the most favorable addition site to the ring carbons). b Ipso addition of O(³P) on the lone pair of the N atom of pyridine. The location of the minimum energy crossing point (MECP) between triplet and singlet PES following meta (MECP_m) and ipso (MECP_i) addition, where ISC occurs, is indicated with a circle. Only the two main observed channels for the O(³P) reaction are portrayed. The dashed interconversion blue lines on the singlet PES leading to the H and CO forming channels are simplified and qualitative (for a detailed complete view of the triplet/singlet PESs with all the interconversion pathways (maxima and minima) along the full singlet PES, see Recio et al. 2022)

These same channels can also be formed from the O(¹D) + C₅H₅N reaction, with the reaction exothermicities in this case being larger by 190 kJ/mol because of the energy difference between excited state ¹D and ground state ³P of atomic oxygen. It should be noted that while the o-, m-, and p-pyridoxyl + H products can be formed on either triplet or singlet PES, the pyrrole + CO products can be formed from O(³P) only via ISC from the entrance triplet PES to the singlet PES, or from O(¹D) adiabatically. Particularly relevant to the O(¹D) reaction, besides the H and CO forming channels, are also the so called “3-body channel” (14), generated by the fast uni-molecular decay of internally hot pyrrole before reaching the detector, and the channel (13) leading to pyrrolyl + HCO, that for O(¹D) are both strongly exothermic.

In the following we summarize the main experimental results and findings of our recent study. The O(³P,¹D) + C₅H₅N reactions were investigated in CMB experiments at E_c = 37.2 kJ/mol using a supersonic oxygen beam containing about 90% of O(³P) and 10% of O(¹D) (Alagia et al. 1997). Data analysis was carried out in a similar way as for the related O(³P,¹D) + C₆H₆ (benzene) reaction (Vanuzzo et al. 2021). Notably, it was found that there are similarities as well as differences in the derived best-fit CM functions for the related O(³P,¹D) + pyridine and O(³P,¹D) + benzene systems (Recio et al. 2022).

Reactive scattering signals were observed at m/z = 94 (C₅H₄NO⁺), 93 (C₅H₃NO⁺), 67 (C₄H₅N⁺), and 66 (C₄H₄N⁺). From LAB product angular and TOF distribution measurements at m/z = 94, 67, and 66, we characterized the occurrence of the H-displacement channel(s), from data at m/z = 67 the pyrrole formation channels as well as the H-displacement channel(s), and from data at m/z = 66 the 3-body and HCO channels, as well as the H-displacement channel(s). In Fig. 8 we show some data at m/z = 66 and 67 (the angular distributions, N(Θ), and exemplary TOF distributions, N(Θ,t), at the LAB angle of 60°), to demonstrate the dynamics of the observed reaction channels from O(³P) and O(¹D) reactions (for the complete data set, see Recio et al. 2022).

Fig. 8

(adapted from Recio et al. 2022)

(lhs): Comparison of the LAB angular distributions at m/z = 67 [parent ion of pyrrole product and (− 27) daughter ion of pyridoxyl product] and at m/z = 66 [(− 1) daughter ion of pyrrole, parent ion of pyrrolyl, and (− 28) daughter ion of pyridoxyl] for the O(³P,¹D) + pyridine reaction at E_c = 37.2 kJ/mol. (rhs): TOF distributions at the two same masses, at the LAB angle of 60°. The black solid lines over the experimental points on both angular and TOF distributions are the global calculated best-fit curves when using the weighted best-fit CM product angular and translational energy distributions (see Fig. 4 in the Supporting Information of Ref. Recio et al. 2022). The distinct contributions of the various products from the distinct O(³P) and O(¹D) reactions are indicated and color coded

As mentioned above, the C₅H₄NO (pyridoxyl) product, besides at the parent ion (m/z = 94), was also detected (more readily) at m/z = 66 and also at m/z = 67, originating from strong dissociative ionization of the parent ion C₅H₄NO⁺ in the ionizer. As can be seen in Fig. 8(lhs) the N(Θ)s of m/z = 66 and 67 exhibit a pronounced, slightly backward-biased distribution with respect to the O atom direction, centered around the CM angle. The N(Θ) of the C₅H₄NO products at m/z = 94 (not shown here—see supporting Fig. 1 in Recio et al. 2022) and their contributions at m/z = 66 and 67 are confined to a narrow LAB angular range because of linear momentum conservation, as the H coproduct is very light. However, in contrast to the m/z = 94 N(Θ), at both m/z = 66 and 67, while the partial contribution of the coproduct of the H displacement channel has the same shape as at m/z = 94, a strong reactive signal is also present at LAB angles well outside the range where the heavy coproducts of H channels are confined, thus suggesting the presence of at least another channel, besides the H channel. The different channels were identified from the combined analysis of angular and TOF distributions; in particular the TOF spectra N(Θ,t), although are essentially bimodal [see Fig. 8 (rhs)], carry the fingerprint of all observed channels mentioned above [two from O(³P) and four from O(¹D)] (see Recio et al 2022 for the complete data set and analysis).

Quantitative information on the reaction dynamics was obtained by moving from the LAB coordinate system to the CM reference frame and analyzing the best-fit product ${\text{T}}(\uptheta )$ and ${\text{P}}({{\text{E}}}_{{\text{T}}}^{\mathrm{^{\prime}}})$ distributions. In Fig. 8 the solid curves superimposed on the experimental distributions are the calculated curves obtained using the best-fit CM ${\text{T}}(\uptheta )$ and ${\text{P}}({{\text{E}}}_{{\text{T}}}^{\mathrm{^{\prime}}})$ functions reported in supporting Fig. 4 of Recio et al. (2022) for all six observed channels [the two channels from O(³P) and the four from O(¹D)].

The procedure adopted to determine reaction yields (product BFs) for this system was similar to that used for the O(³P,¹D) + C₆H₆ system (Vanuzzo et al. 2021). Despite the large number of exothermic channels, our study (Recio et al. 2022) showed that the dominant product channels from the O(³P) reaction are only channels (11b) and (12a), leading to H and CO elimination, respectively, with channel (11b) being minor (${\text{BF}}={0.02}_{-0.01}^{+0.02}$) and coming from the triplet PES via adiabatic reaction, and channel (12a) being dominant (${{\text{BF}}=0.98}_{-0.02}^{+0.01}$ and coming from the singlet PES via efficient triplet to singlet ISC. The product BFs for both O(³P) and O(¹D) reactions are summarized in Table 3. The fact that the BF of the pyrrole + CO channel from O(³P) is 0.98 indicates that the extent of ISC is very large (98%).

Table 3 Product branching fractions (BFs) for the distinct O(³P) + pyridine and O(¹D) + pyridine reactions at E_c = 37.2 kJ/mol (adapted from Recio et al. 2022)

The O(³P) + pyridine reaction is the system with the largest contribution of triplet to singlet ISC observed up to now, a feature that theory explained with the occurrence of ISC in the entrance channel for the ipso addition of O(³P) to the lone pair of the nitrogen atom, before the entrance barrier (Recio et al. 2022). Instead, for the O(¹D) reaction the main product channel is the three-body channel leading to pyrrolyl + H + CO (BF = 0.78 ± 0.10), which, while nearly thermoneutral for O(³P) [channel (14)], is strongly exothermic for O(¹D) ($\Delta {H}_{0}^{0}=-188$ kJ/mol). Notably, the 3-body channel was also observed to be the main channel, leading to cyclopentadiene + H + CO, in the corresponding reaction of O(¹D) with benzene (Vanuzzo et al 2021). For O(¹D), the pyrrole + CO channel has BF = 0.16 ± 0.08, the H channel has BF = 0.014 ± 0.007, and the HCO-forming channel has BF = 0.04 ± 0.02 (see Table 3). This indicates that, overall, also for the O(¹D) reaction the H-displacement channel is minor.

The experimentally estimated extent of ISC of about 98% is in sharp contrast to what was derived for the reaction of O(³P) with the isoelectronic benzene molecule, where the extent of ISC was determined to be ca. 50% at a comparable E_c (Vanuzzo et al. 2021). In O(³P) + benzene, the H + C₆H₅O (phenoxy) production is controlled by the spin-allowed triplet pathway with BF = 0.66 (0.48 from the adiabatic contribution on the triplet PES and 0.18 from the nonadiabatic contribution on singlet PES via ISC), that is, it is much larger than in O(³P) + pyridine. The ring contraction channel leading to CO + C₅H₆ (cyclopentadiene) occurs via triplet to singlet ISC with a BF of 0.32, that is, the ring contraction channel in O(³P) + benzene is much smaller (50%) than in O(³P) + pyridine (98%).

The observation that ISC to the singlet PES, leading to formation of pyrrole + CO [both 1H- and 2H-pyrrole (channel (12a) and (12b), respectively) in a 1/2.4 ratio, according to statistical calculations (Recio et al. 2022)], occurs in the entrance channel before the barrier, following the ipso interaction between the O atom and the lone pair of the N atom of pyridine on the triplet PES, represented the first detection of ISC in the entrance channel of a bimolecular reaction in the absence of heavy atoms (e.g., iodine). Interestingly, this reaction mechanism involving ISC before the entrance barrier may play a role in bimolecular reactions with the same characteristics. In particular, considering that important biomolecules, including nucleobases, are N-heterocyclic aromatic compounds, the finding of our study on the reaction of O(³P) with pyridine about the key role of the electrophilic attack of the oxygen atom on the nitrogen atom accompanied by efficient ISC, as opposed to the usual attack on the carbons of the aromatic ring, may have significant implications in biological processes. Finally, in relation to the potential astrochemical relevance of the O(³P) + pyridine reaction, our results suggested a possible destruction pathway of pyridine hitherto unobserved in different environments of the interstellar medium.

3 Conclusions and outlook

We have reviewed the recent contributions made by CMB studies in our laboratory of the bimolecular reactions of ground and excited state oxygen atoms, O(³P) and O(¹D), with the two simplest unsaturated nitriles, cyanoacetylene and cyanoethylene, and the simplest N-heterocyclic molecule, pyridine. The primary products of these multichannel reactions and their product branching fractions have been determined under well-defined energetic conditions. With the assistance of synergistic calculations of the relevant PESs and of statistical (RRKM/ME) calculations, in the conditions of the experiments, of product BFs, the detailed reaction mechanisms have been elucidated and the central role played by intersystem-crossing in the O(³P) reactions has been assessed. Specifically, ISC has been found to control the product distributions in all these reactions. The contributions of these studies to the modelling, and therefore to a deeper comprehension of combustion and astrochemical processes have been discussed.

We emphasize that the characterization of multichannel bimolecular neutral–neutral reactions of relevance in combustion and astrochemistry is a challenging endeavour that certainly call for a multidisciplinary approach. Because kinetic experiments are seldom able to determine all the primary products and their branching fractions at the temperatures of relevance in these environments, dynamical experiments using the CMB technique with universal MS detection, empowered with soft electron-ionization (or soft VUV photo-ionization), have revealed to be a powerful and most valuable tool, complementary to kinetic experiments, to provide product BFs in specific energy (temperature) conditions. The CMB experiments, combined with high-level theoretical calculations of the relevant reaction PESs and statistical (RRKM/ME) computations of product BFs with inclusion of nonadiabatic effects (ISC) can ultimately lead to the predictions of both the global reaction rate constant and the channel-specific rate constants as a function of temperature (and also pressure), that are the needed quantities for improving combustion and astrochemical models. We emphasize that the detailed comparison between theoretical predictions of product BFs of a multichannel reaction, using coupled potential energy surfaces with inclusion of ISC effects, with the experimental product BFs provides a sensitive test of the reliability of the underlying reaction PESs. When the PESs are validated by a good accord between experiment and theory, then theory can be used to predict product BFs as well as channel-specific rate constants as a function of temperature, which may range from the low values of cold molecular clouds to those of the atmosphere of Titan, up to those of combustion processes. Up to now this approach has been very rewarding when applied to a large variety of reactions of oxygen atoms with unsaturated aliphatic and aromatic hydrocarbons as well as the present N-bearing molecules.

We are currently extending in our laboratory similar combined CMB and theoretical investigations to also reaction involving, besides oxygen atoms also sulfur atoms, alkyl radicals and other molecular species of relevance in combustion and astrochemistry, such as alkyl-benzenes, other heterocyclic molecules, and the prototype of polycyclic aromatic hydrocarbons, such as naphthalene. Combustion and astrochemical models are expected to benefit from the additional, detailed information that can be obtained from studies such as those described here. New insights and better understanding of these complex environments can ultimately be expected.