The Astrophysical Journal, 951:150 (7pp), 2023 July 10 https://doi.org/10.3847/1538-4357/acd192 © 2023. The Author(s). Published by the American Astronomical Society. Formation of Complex Organic Molecules on Interstellar CO Ices? Insights from Computational Chemistry Simulations Stefano Ferrero1 , Cecilia Ceccarelli2 , Piero Ugliengo3 , Mariona Sodupe1 , and Albert Rimola1 1 Departament de Química, Universitat Autònoma de Barcelona, Bellaterra, E-08193 Catalonia, Spain; mariona.sodupe@uab.cat, albert.rimola@uab.cat 2 Univ. Grenoble Alpes, CNRS, Institut de Planétologie et d’Astrophysique de Grenoble (IPAG), F-38000 Grenoble, France 3 Dipartimento di Chimica and Nanostructured Interfaces and Surfaces (NIS) Centre, Università degli Studi di Torino, via P. Giuria 7, I-10125 Torino, Italy Received 2023 February 25; revised 2023 April 16; accepted 2023 April 30; published 2023 July 12 Abstract The carbon ( P) atom is a reactive species that, according to laboratory experiments and theoretical calculations, 3 condensates with interstellar ice components. This fact is of uttermost importance for the chemistry in the interstellar medium (ISM) because the condensation reaction is barrierless, and the subsequent species formed are still reactive given their open-shell character. Carbon condensation on CO-rich ices forms the C=C=O (3Σ−) species, which can be easily hydrogenated twice to form ketene (H2CCO). Ketene is very reactive in terrestrial conditions, usually found as an intermediate that is difficult to isolate in chemical synthesis laboratories. These characteristics suggest that ketene can be a good candidate to form interstellar complex organic molecules via a two-step process, i.e., its activation followed by a radical–radical coupling. In this work, reactions between ketene and atomic H and the OH and NH2 radicals on a CO-rich ice model have been explored by means of quantum chemical calculations complemented by kinetic calculations to evaluate if they are favorable in the ISM. Results indicate that the addition of H to ketene (helped by tunneling) to form the acetyl radical (CH3CO) is the most preferred path as the reactions with OH and NH2 possess activation energies ( 9 kJ mol−1) hard to surmount in the ISM conditions unless external processes provide energy to the system. Thus, acetaldehyde (CH3CHO) and, probably, ethanol (CH3CH2OH) formation via further hydrogenations, are the possible unique operating synthetic routes. Moreover, from the computed, relatively large binding energies of OH and NH2 on CO ice, slow diffusion is expected, hampering possible radical–radical couplings with CH3CO. The astrophysical implications of these findings are discussed considering the incoming James Webb Space Telescope observations. Unified Astronomy Thesaurus concepts: Astrochemistry (75); Interstellar medium (847); Interstellar molecules (849); Interstellar dust (836); Surface ices (2117); Complex organic molecules (2256); Reaction rates (2081); Computational methods (1965) 1. Introduction through hydrogenation of CO followed by the formation of radicals via photodissociation on the ice surfaces and their Interstellar grains are submicron-sized solid particles made recombination (Garrod & Herbst 2006; Chuang et al. 2017, either of carbonaceous materials or of silicates. In cold (∼10 K) 2021; Simons et al. 2020). and dense (∼104 cm−3) molecular clouds, these grains are However, another promising route toward chemical com- covered predominantly by water icy mantles, but several other plexity in the interstellar medium (ISM), which has emerged in species have been detected in the infrared observations of the past few years, is the reactivity of atomic carbon with interstellar ices (e.g., Boogert et al. 2015; Yang et al. 2022; different components of the icy mantles. The condensation of McClure et al. 2023). Interstellar grains provide surfaces on which chemical reactions can occur, forming stable products of atomic carbon on water ice, both in its neutral (C) and in its astrochemical relevance (e.g., Tielens & Hagen 1982; Cuppen cationic (C+) forms, has been studied, and possible chemical et al. 2017; Potapov & McCoustra 2021). reactions with different icy components have been elucidated In addition to H2O, one of the most abundant icy species is (Krasnokutski et al. 2017; Shimonishi et al. 2018; Henning & carbon monoxide (CO), which is thought to form in the gas Krasnokutski 2019; Molpeceres et al. 2021; Potapov et al. phase and then freeze out onto the surface of the grains (Caselli 2021; Woon 2021). Moreover, some of these processes have et al. 1999; Bacmann et al. 2002; Favre et al. 2013). This CO been linked to chemical pathways to form amino acids freeze-out is supposed to happen after the formation of the (Krasnokutski et al. 2020, 2022). water ice, and therefore, interstellar ices are thought to present On the other hand, the condensation of atomic carbon on an (almost) onion-like structure, the innermost layers being pure CO ice is by far less studied. The C(3P) + CO reaction has formed by a polar phase dominated by water, whereas the outer been reported and studied in the gas phase by a recent layers are formed by a nonpolar phase, possibly dominated by computational work, which found the formation of the CO (Pontoppidan et al. 2008; Boogert et al. 2015). These C=C=O (3Σ−) species as the product in a barrierless way nonpolar outermost layers are thought to be crucial for the (Papakondylis & Mavridis 2019). In the ISM, these species can formation of interstellar complex organic molecules (iCOMs) be easily hydrogenated twice to form ketene (H2CCO), which can be hydrogenated even further to form acetaldehyde, as Original content from this work may be used under the terms found recently experimentally (Fedoseev et al. 2022), and of the Creative Commons Attribution 4.0 licence. Any further eventually, ethanol. The formation of the latter species is distribution of this work must maintain attribution to the author(s) and the title particularly interesting as, in the gas phase, it is the starting of the work, journal citation and DOI. point of a chain of reactions leading to glycolaldehyde 1 The Astrophysical Journal, 951:150 (7pp), 2023 July 10 Ferrero et al. (Skouteris et al. 2018). Recent observations by the James Webb cluster model is enough because, since the nature of the Space Telescope (JWST) toward background stars have shown interactions between CO molecules (namely, quadrupole– the possible presence of acetaldehyde and ethanol in the icy quadrupole and dispersion components; Zamirri et al. 2018), grain mantles in molecular clouds (Yang et al. 2022; McClure the generated cluster model exhibits all the possible CO et al. 2023). In these regions, the ices are composed of about orientations on the surface, that is, C/C, O/O, and C/O, hence 25% of frozen CO (i.e., the catastrophic freeze-out mentioned covering all the likely CO configurational variabilities. above has not occurred yet; e.g., Boogert et al. 2015), while As CO molecules in the cluster are not strongly bound some atomic carbon is still in the gas phase (Zmuidzinas et al. together, and to avoid large deformation of the cluster structure 1988; Kamegai et al. 2003). Although the identification of iced when studying adsorption and reactivity, constrained geometry acetaldehyde and ethanol needs to be confirmed, the possibility optimizations and TS searches were performed, as explained in that (either of) these species are already present in the the following. molecular cloud phase warrants a dedicated study on the Interaction energies for OH and NH2 on the CO cluster were chemistry triggered by the C condensation on CO ice. calculated by performing geometry optimizations, in which In this work, the formation and reactivity of ketene on a only the adsorbate (NH2 or OH) and the closest CO molecules model of CO ice are explored through quantum mechanical (3–5, depending on the site) were allowed to relax. Harmonic simulations. Its reactivity with abundant interstellar radicals, frequencies were calculated only for the adsorbates employing like H, O, N, NH, OH, and NH2, is studied in order to identify a partial hessian vibrational analysis (PHVA) scheme (Li & whether and which of these species can react with ketene, Jensen 2002). The ZPE-corrected interaction energies on the thereby opening up chemical pathways that form even more CO ice model were calculated as complex species. DH (0) = Ecomplex - E CO - Eadsorbate + DZPE, (1 ) 2. Computational Detail where Ecomplex is the absolute potential energy of the 2.1. Gas-phase Calculations adsorption complex, ECO is the absolute potential energy of A preliminary gas-phase screening of reaction barriers was the isolated optimized CO cluster, and Eadsorbate is the absolute made to determine which reactions are more probable at ISM potential energy of the isolated adsorbate and bearing in mind conditions. All the electronic structure calculations have been that at 0 K, the absolute ZPE energy is equal to the absolute carried out with the Orca 5.0 software (Neese et al. 2020). enthalpy, i.e., E0 = H(0). ΔZPE values have been calculated by Density functional theory (DFT) was used for geometry subtracting the ZPE corrections of the adsorbate optimized on optimizations adopting the ωB97x-D4 functional and the the CO cluster and the ZPE corrections of the isolated def2-TZVP basis set (hereafter referred to as ωB97x-D4/ TZVP; Weigend & Ahlrichs 2005; Najibi & Goerigk 2020). adsorbate. Geometry optimizations were carried out with the geometrical For the reactivity on the CO cluster, the structures of counterpoise correction method to remove the basis set reactants and products were first optimized, and then TS superposition error (Liu & McLean 1973; Kruse & Grimme structures were localized with the NEB-TS algorithm. Seven 2012). Electronic energies were refined with single-point CO molecules were included in the optimizations and TS calculations at the DFT-optimized geometries with the CCSD searches. A PHVA scheme was used to characterize the (T)-F12 and DLPNO-CCSD(T)-F12 methods (Adler et al. optimized structures as minimum or TS and to correct for ZPE. 2007; Hättig et al. 2012; Kong et al. 2012; Pavošević et al. Energy barriers were refined by single-point calculations at the 2017), which employ a cc-pVTZ-F12 basis set, the cc-pVTZ- DLPNO-CCSD(T)-F12 level of theory and calculated as F12-CABS near-complete basis set, and the aug-cc-pVTZ/C fitting basis set for the resolution of identity approximation DH ‡ (0) = E TS - Emin + DZPE, (2 ) (Weigend et al. 2002; Peterson et al. 2008). For simplicity, the CCSD(T)-F12/cc-pVTZ-F12//ωB97X-D4/TZVP and where ETS and Emin are the DLPNO-CCSD(T)-F12 absolute DLPNO-CCSD(T)-F12/cc-pVTZ-F12//ωB97x-D4/TZVP potential energies for the TS and the minimum structure of the levels of theory will be referred to as CCSD(T)-F12 and reaction, respectively. ΔZPE has been calculated by subtract- DLPNO-F12, respectively. For DLPNO calculations, a tight ing the ZPE of the fragment made by the adduct (ketene plus pair natural orbitals (PNO)setting was used. Transition state H/OH/NH2) optimized on the CO cluster and the ZPE of the (TS) structure searches have been conducted using the nudged adduct calculated in vacuum at the ωB97x-D4/TZVP level. At elastic band (NEB) algorithm implemented in Orca (Ãsgeirsson ISM temperatures (e.g., 10 K), thermal corrections are in et al. 2021). Harmonic vibrational frequencies were calculated practice negligible (Zamirri et al. 2017, 2019; Enrique-Romero to characterize the nature (e.g., minimum or TS) of the et al. 2019, 2021), so we assume that the calculated ΔH(0) and optimized structures and to correct the electronic energies for the zero-point energy (ZPE). ΔH‡(0) do not vary at the cryogenic temperatures. The visual molecular dynamics (VMD)software was used for rendering images (Humphrey et al. 1996). 2.2. Solid-phase Calculations In order to mimic a CO ice, a cluster approach was used in a 2.2.1. Instanton Rate Theory Calculations similar way to the inspiring paper by Lamberts et al. (2019). An initial cluster was generated with the Packmol software by To assess the effect of tunneling on the reaction rates, the randomly placing 20 CO molecules inside a sphere of semiclassical instanton theory (Chapman et al. 1975; Miller 12 Å radius, which was then optimized with the ωB97x-D4 1975) was employed for the H additions to ketene. A simple functional and a def2-SVP basis set (see Figure 1(A)). This estimation of the crossover temperature (Tc), at which tunneling 2 The Astrophysical Journal, 951:150 (7pp), 2023 July 10 Ferrero et al. Figure 1. Panel (A): optimized structure of the CO cluster model in Van der Waals representation. Panel (B), left: ketene molecule structure with the C1 and C2 labels. Panel (B), right (dashed box): atomistic structure of the species to react with ketene (H, N, O, NH, OH, and NH2) considering the two possible attacks (arrows toward C1 and C2). Atom color legend: H, white; C, cyan; N, blue; O, red. becomes important, was obtained as 3. Results and Discussion w 3.1. Gas-phase Calculations Tc = , (3 ) 2pkB The C( P) + CO reaction, already studied in the work of 3 where h is the Planckʼs constant, kB is the Boltzmann Papakondylis & Mavridis (2019), was here reproduced at the constant, and ω is the imaginary frequency of the TS. As ωB97x-D4/TZVP level of theory. Results are in agreement instanton theory tends to overestimate the reaction rates with the previous findings on the barrierless formation of around Tc (Andersson et al. 2009) and since our interest is on C=C=O (3Σ−), and thus, the formation of ketene (H2CCO) via rates at interstellar temperatures, instanton rate theory has a double hydrogenation reaction is viable. In order to study the been applied only in the deep tunneling regime, e.g., below reactivity of ketene and to assess the height of the reaction Tc. The instanton describes the most probable tunneling path barriers, gas-phase calculations were carried out for the reactions with abundant interstellar radicals, involving attacks from reactants to products at a given temperature and can be on the two carbon ketene atoms, that is, regarded as a saddle point on a ring-polymer potential energy surface constructed as a discretized Feynman path of N H2 CCO + X •  H2 C•CXO (4 ) segments, called beads (Kästner 2014; Beyer et al. 2016; H2 CCO + X•  H2 XCC •O, (5 ) Richardson 2018a, 2018b). Instantons for the hydrogenation • with X = H, O, N, NH, OH, and NH2 (see panel B of Figure reactions on the two ketene carbon atoms have been optimized 1). An important point is that, after the attack of X•, the newly employing a progressive cooling approach, starting from a formed molecule is still a radical due to its unpaired electron temperature just below Tc down to 50 K. As a first and therefore reactive to couple with other open-shell species to discretization of the path, 16 beads were employed, which possibly form iCOMs. In this work, the carbon bonded to the were then incremented up to 128 beads to obtain convergence oxygen atom is labeled as C1 and the other as C2 (see on the rates at 50 K. This was the lower limit temperature Figure 1B). because convergence at lower temperatures requires even All the studied reactions are exothermic but present energy more beads, making the calculation computationally imprac- barriers (see Table 1). tical. However, as it will be seen, the rate constants at 50 K do In all cases, except for H, the products arising from the C1 not depend on temperature, so they can be extrapolated to 10 attack are thermodynamically more stable than those from the K. A dual-level instanton approach (CCSD(T)-F12/cc-pVTZ- C2 attack due to the overstabilization gained by the π F12//ωB97X-D4/TZVP; Meisner & Kästner 2018) was then delocalization when forming, for instance, an amide bond or used to refine the energy of the beads. Finally, as these a carboxylic group. However, C2 is the most prominent site to reactions are supposed to happen on a CO-rich ice, the experience an attack due to the lower energy barriers in the implicit surface model approach was applied to include cases of H, N, O, and OH, whereas for NH and NH2, the attack surface effects (Meisner et al. 2017), which holds only if the to C1 is preferred. Moreover, the OH radical attack presents the lowest energy barrier among the species studied, whereas catalytic role of the surface is negligible. In this approx- N-bearing species are the most inert (see Table 1). imation, the rotational partition function is assumed to be Note that DFT energy barriers are, in most cases, in good constant during the reaction as the surface hampers rotations agreement with the results obtained with the CCSD(T)-F12 in both the reactant and TS structures. Instantons have been method. The barriers found for the hydrogenation reactions are optimized on the fly by interfacing the Orca software with a also in good agreement with Ibrahim et al. (2022). Python code developed in Jeremy Richardsonʼs group at ETH Furthermore, it can be noticed that the DLPNO-F12 results Zurich. agree fairly well with the CCSD(T)-F12 ones, which support 3 The Astrophysical Journal, 951:150 (7pp), 2023 July 10 Ferrero et al. Figure 2. Optimized structures of the reactant and product for the ketene + NH2 reaction taken as an example. Geometry-relaxed species are represented in a ball and stick mode, whereas transparent CO molecules are held fixed during geometry optimization. Table 1 ZPE-corrected Reaction Energies (ΔHreact) and Energy Barriers (ΔH‡(0)), in kJ mol−1 Reactions ΔHreact ΔH‡(0) Attack on C1 ωB97x CCSD(T) DLPNO ωB97x CCSD(T) DLPNO Ketene + H −147.0 −147.6 −150.1 36.3 30.6 31.1 Ketene + N −75.4 −46.4 −49.5 68.7 78.8 78.5 Ketene + NH −116.5 −99.9 −101.3 51.8 54.4 54.8 Ketene + NH2 −169.1 −163.8 −163.1 37.4 37.2 38.5 Ketene + O −183.9 −186.7 −185.4 14.9 15.7 17.1 Ketene + OH −209.9 −213.1 −213.6 5.2 3.3 4.6 Attack on C2 Ketene + H −173.9 −175.6 −175.7 19.3 16.8 17.1 Ketene + N −50.1 −27.4 −28.2 62.7 74.8 74.1 Ketene + NH −71.9 −60.7 −60.4 59.0 63.9 63.9 Ketene + NH2 −93.7 −91.5 −90.5 43.9 43.5 43.4 Ketene + O −121.3 −115.3 −115.7 5.8 9.8 9.2 Ketene + OH −117.5 −123.1 −122.9 4.1 3.1 3.4 Note. ωB97x-D4, CCSD(T), and DLPNO stand for ωB97x-D4/TZVP, CCSD(T)-F12//ωB97x-D4/TZVP, and DLPNO-F12//ωB97x-D4/TZVP level of theories, respectively. the use of DLPNO-F12 for the energetic refinement on CO The reactions have been modeled by adopting a Langmuir– ices, for which the CCSD(T)-F12 method cannot be employed Hinshelwood mechanism. Thus, to calculate the reaction due to the large size of the cluster. barriers, the reactant, TS, and product structures have been optimized on two adjacent adsorption sites, including seven unconstrained CO molecules (the rest remaining fixed; see 3.2. Solid-phase Calculations Figure 2). The calculated barriers are reported in Table 2. To confirm that the ketene formation is also doable in the According to the ISM conditions, they are all very high (the solid state, the C(3P) + CO(ice) condensation was also lowest one being 9.3 kJ mol−1) for the reactions to proceed. In investigated on the CO ice, resulting indeed in the formation the following, we discuss the differences with the barriers of the C=C=O (3Σ−) species. obtained in the gas phase. Based on the gas-phase results, the cases of H (the most Ketene + NH2. Comparison of the solid-state barriers abundant species) and OH and NH2 (the species presenting the with the gas-phase ones gives differences that are less than lowest barriers for the O- and N-containing radical family 4 kJ mol−1. Since the energy barriers for both cases are high species) have been selected to study their reactivity with ketene (insurmountable in the ISM conditions), these variations have on the CO ice. no practical effects so that CO ice behaves as an inert surface. 4 The Astrophysical Journal, 951:150 (7pp), 2023 July 10 Ferrero et al. Table 2 Table 3 Computed ZPE-corrected Energy Barriers (ΔH‡(0)) for the Reactions of Adsorption Energies on CO and H2O Ices of NH2 and OH Ketene with H, NH2, and OH on the CO Cluster Species CO ice Water ice Species Attack on C1 ΔH‡(0) Attack on C2 ΔH‡(0) (kJ mol−1) (kJ mol−1) (kJ mol−1) (kJ mol−1) OH −22.1 ± 3.9 −43.2 ± 16.1a DLPNO-F12 DLPNO-F12 −44.4b −32–41c H 29.1 13.5 −26.9 ± 11.5d NH2 41.0 42.9 NH2 −6.5 ± 3.2 −31–44e OH 10.2 9.3 −27.9 ± 3.1d Notes. Mean value and standard deviation of the adsorption energies ΔH(0) on Ketene + OH. For OH, there is an increase of 6 kJ mol−1 on the CO cluster calculated in this work (second column) and calculated data on CO, a nonnegligible variation since the gas-phase barriers are different water ice models from other works (third column) are reported. When very low. The increase can be attributed to the interactions more than two values were present in the literature, mean and standard between OH and the CO ice (absent in the gas phase), which deviation have been calculated. References. a Duflot et al. (2021), b Sameera et al. (2017), c Meisner et al. need to be partly overcome to proceed with the reaction on the (2017), d Ferrero et al. (2020), e Enrique-Romero et al. (2019). surface. Interestingly, for this case, the energy barrier is relatively small, which could be overcome classically by means of nonthermal mechanisms (as reported in works involving these two radicals on CO-rich ices will be larger than that on reactive OH radicals; see Garrod & Pauly 2011; Ishibashi et al. water ices due to the weaker radical/surface interactions in the 2021) or could even proceed with the help of heavy atom former. However, the binding energies of these species on CO quantum tunneling below the crossover temperature (see ice are still significantly large, so diffusion is expected to be very Meisner & Kästner 2016; Castro & Karney 2020). Moreover, slow at the ISM conditions, which in turn hampers diffusion- it is worth mentioning that the calculated barrier arises from limited reactions like radical–radical couplings. just one starting configuration of the reactants, which could well be lower in other surface reactive sites. 3.3. Kinetic Calculations and Tunneling Effects Ketene + H. Here, the C2 attack appears to be the most doable as it presents a barrier that is less than half of the barrier At the cryogenic temperatures of the ISM and for reactions of the C1 reaction, but it is still very high for the ISM involving light atoms, like H, quantum tunneling cannot be conditions. However, it is worth noticing that, with reactions overlooked. Thus, ketene hydrogenation considering both C1 happening at low temperatures and involving light species such and C2 attacks was also studied adopting the instanton theory as H, tunneling effects can dominate, and therefore, considering (outlined in Section 2.2.1), which calculates the most probable only classical barrier heights can lead to wrong conclusions. tunneling path connecting reactants and products. Regarding the This kinetic aspect is discussed in more detail in the Section 3.3. attack on C1, the calculated Tc is 230 K, whereas on C2, it is Another crucial point for the on-grain reactivity is the 157 K. Figure 3 reports the Arrhenius plots for these two H species/ice interactions that can be established at the surfaces. additions. As done in previous works (Meisner et al. 2017; On icy water grains, species like NH2 and OH can form strong Lamberts & Kästner 2017; Lamberts et al. 2016), the reactions hydrogen bond (H-bond) interactions, inhibiting their diffusiv- are considered as unimolecular because the diffusion of the two ity and hence, their reactivity via Langmuir–Hinshelwood. reactants is not considered and the rate constants measure the These interactions have been studied on water ice surfaces both rate starting from a reactive complex. The calculated rate experimentally and theoretically (Wakelam et al. 2017; constants refined at CCSD(T)-F12/cc-pVTZ-F12//ωB97X- Enrique-Romero et al. 2019, 2022, 2021; Ferrero et al. 2020; D4/TZVP at 50 K are 4.47 × 104 s−1 and 1.74 × 107 s−1 for Duflot et al. 2021; Tsuge & Watanabe 2021; Sameera et al. C1 and C2 hydrogenations, respectively. Remarkably, rate 2022, 2017). However, very little data are available in the constants keep almost invariant at these temperatures, so we can literature relative to CO ices. To have an idea of the interactions assume similar values for 10 K. Therefore, both attacks are of OH and NH2 with CO-rich ices (different in nature to water viable at ISM temperatures, being kinetically more favorable on ice), we used a similar strategy as in Lamberts et al. (2019) to C2. As stated in Section 2.2.1, these instanton calculations have calculate interaction energies: we sampled ten different binding been carried out in the gas phase, but the surface effects have sites around the CO cluster to calculate the corresponding ΔH been considered through the implicit surface approximation, (0) adsorption energies (results shown in Table 3). which in this case holds because the CO surface acts as an inert Although the binding site sampling is not exhaustive, we can substrate without affecting significantly the classical barriers. compare our values with other computational works that calculated the interaction of these two species on water ice 4. Astrophysical Implications and Concluding Remarks models. The difference in the adsorption energies is quite large, clearly indicating that the interaction on CO-rich ices is The calculations presented here show that, when atomic C significantly weaker than on H2O-rich ones (particularly for lands on an icy CO surface, C=C=O forms, as in the gas phase NH2), as already found in previous works for different radicals (Papakondylis & Mavridis 2019). However, once formed, it (Lamberts et al. 2019). This is due to the fewer and weaker could very likely be hydrogenated into ketene by the addition H-bonds that OH and NH2 form on CO ice with respect to water of H, also landing on the grain surfaces. We then explored if ice. This fact is also connected with the mobility of these species ketene (H2CCO) could grow into larger molecules by reacting on different icy surfaces. Even without calculating explicitly the with simple and abundant radicals on the grain surfaces, diffusion energy barriers, it is noticeable that the diffusivity of namely H, OH, and NH2. 5 The Astrophysical Journal, 951:150 (7pp), 2023 July 10 Ferrero et al. Figure 3. Arrhenius plots for the two different hydrogenation reactions on C1 (left panel) and C2 (right panel). Black dots correspond to values of the rate constant calculated with instanton theory at different temperatures, whereas the classical Eyring rate constant is depicted in red. Gray dashed vertical lines denote the crossover temperature. These reactions on the (nonpolar) CO ice present relatively surface chemistry, other than the formation of H2, is efficient. high energy barriers ( 9 kJ mol−1) for the ISM conditions. It Indirect proof that frozen CO is present and that it is seems clear that ketene hydrogenation can occur through the H hydrogenated is provided by the observed relatively abundant tunneling. Thus, acetyl radical (CH3CO) formation (through H gaseous methanol in PDRs (e.g., Bouvier et al. 2020). In this addition to ketene) is the most kinetically favorable path. respect, therefore, the C=C=O hydrogenation probably occurs Formation of radical precursors for acetamide (CH3CONH2) in these regions and can lead to frozen acetaldehyde and, and acetic acid (CH3COOH) (through NH2 and OH addition, perhaps, ethanol. respectively), according to our results, are not expected to form. Another possibility is that ketene formed in the gas phase is However, local surface heating and other nonthermal also frozen into the grain icy mantles. Astrochemical models mechanisms could be operative, this way helping the predict a gaseous ketene abundance of about 10−10–10−8, while occurrence of these reactions so that these synthetic paths observations indicate 10−10–10−9 abundances (e.g., Bacmann cannot be discarded. et al. 2012; Jaber et al. 2014; Vastel et al. 2014). Assuming that CH3CO can be in turn hydrogenated and produce the frozen ketene is all converted into acetaldehyde and/or acetaldehyde (CH3CHO) and/or CO + CH4. Recent experi- ethanol, it provides upper limits to their abundance of 10−10– ments indicate that the latter channel competes with the first 10−9 with respect to H, namely about 10−6–10−5 with respect to one, with a branching ratio up to 80%, against naive frozen water. expectations (Ibrahim et al. 2022). One can speculate that In summary, it is possible that (frozen) ketene is formed in acetaldehyde can successively be hydrogenated into ethanol by the PDR skins of the molecular clouds, where C and frozen CO other H additions. One could also suppose that the CH3CO may coexist, which then would trigger the formation of iced radical could react with the OH and NH2 radicals. However, acetaldehyde and ethanol, possibly explaining the new our computed, relatively large binding energies of OH and NH2 observations of JWST (which need confirmation). on CO ice point toward a very low diffusivity, inhibiting, once again, the formation of CH3COOH and CH3CONH2 via Acknowledgments radical–radical coupling with CH3CO. S.F. is deeply indebted to Gabriel Laude and Jeremy As mentioned in the Section 1, gaseous atomic C and frozen Richardson for their help and advice on instanton calculations. CO are simultaneously found in molecular clouds, probably in This project has received funding from the Marie Sklodowska- the photodissociated region (PDR) skins. The bottleneck to form Curie for the project “Astro-Chemical Origins” (ACO), grant ketene is probably the quantity of gaseous atomic C since a agreement No. 811312 and within the European Unionʼs substantial fraction of frozen CO, about 25% of the ice, is Horizon 2020 research and innovation program from the observed in diffuse clouds (with visual extinction ∼3 mag; European Research Council (ERC) for the projects “The Dawn Boogert et al. 2015). Estimating the amount of gaseous atomic of Organic Chemistry” (DOC), grant agreement No. 741002 C, where also frozen CO exists, is not easy, but observations and “Quantum Chemistry on Interstellar Grains” (QUANTUM- show that it is up to the same abundance of gaseous CO in giant GRAIN), grant agreement No. 865657. MINECO (project molecular clouds (e.g., Plume et al. 1999), making the C=C=O PID2021-126427NB-I00) is also acknowledged. hydrogenation a possible important source of frozen acetalde- hyde and, perhaps, ethanol. It is worth emphasizing that in the ORCID iDs interiors of the molecular clouds, where carbon is almost entirely trapped in CO, neutral carbon is not expected to be as Stefano Ferrero https://orcid.org/0000-0001-7819-7657 abundant as in their skins. Therefore, the mechanism described Cecilia Ceccarelli https://orcid.org/0000-0001-9664-6292 in this work is mostly viable in the latter, which represents only Piero Ugliengo https://orcid.org/0000-0001-8886-9832 a small fraction of the molecular clouds. That said, little is Mariona Sodupe https://orcid.org/0000-0003-0276-0524 known of the composition of ices in the PDRs and whether Albert Rimola https://orcid.org/0000-0002-9637-4554 6 The Astrophysical Journal, 951:150 (7pp), 2023 July 10 Ferrero et al. References Lamberts, T., & Kästner, J. 2017, ApJ, 846, 43 Lamberts, T., Kumar Samanta, P., Köhn, A., & Kästner, J. 2016, PCCP, 18, Adler, T. B., Knizia, G., & Werner, H.-J. 2007, JChPh, 127, 221106 33021 Andersson, S., Nyman, G., Arnaldsson, A., Manthe, U., & Jónsson, H. 2009, Lamberts, T., Markmeyer, M. N., Kolb, F. J., & Kästner, J. 2019, ESC, 3, 958 JPCA, 113, 4468 Li, H., & Jensen, J. H. 2002, Theoretical Chemistry Accounts, 107, 211 Ãsgeirsson, V., Birgisson, B. O., Bjornsson, R., et al. 2021, J. Chem. Theory Liu, B., & McLean, A. D. 1973, JChPh, 59, 4557 Comput., 17, 4929 McClure, M. K., Rocha, W. R. M., Pontoppidan, K. M., et al. 2023, NatAs, Bacmann, A., Lefloch, B., Ceccarelli, C., et al. 2002, A&A, 389, L6 7, 431 Bacmann, A., Taquet, V., Faure, A., Kahane, C., & Ceccarelli, C. 2012, A&A, Meisner, J., & Kästner, J. 2016, Angewandte Chemie International Edition, 541, L12 55, 5400 Beyer, A. N., Richardson, J. O., Knowles, P. J., Rommel, J., & Althorpe, S. C. Meisner, J., & Kästner, J. 2018, J. Chem. Theory Comput., 14, 1865 2016, J. Phys. Chem. Lett., 7, 4374 Meisner, J., Lamberts, T., & Kästner, J. 2017, ESC, 1, 399 Boogert, A. A., Gerakines, P. A., & Whittet, D. C. 2015, ARA&A, 53, 541 Miller, W. H. 1975, JChPh, 62, 1899 Bouvier, M., López-Sepulcre, A., Ceccarelli, C., et al. 2020, A&A, 636, Molpeceres, G., Kästner, J., Fedoseev, G., et al. 2021, J. Phys. Chem. Lett., 12, A19 10854 Caselli, P., Walmsley, C. M., Tafalla, M., Dore, L., & Myers, P. C. 1999, Najibi, A., & Goerigk, L. 2020, J. Comput. Chem., 41, 2562 ApJL, 523, L165 Neese, F., Wennmohs, F., Becker, U., & Riplinger, C. 2020, JChPh, 152, Castro, C., & Karney, W. L. 2020, Angewandte Chemie International Edition, 224108 59, 8355 Papakondylis, A., & Mavridis, A. 2019, JPCA, 123, 10290 Chapman, S., Garrett, B. C., & Miller, W. H. 1975, JChPh, 63, 2710 Pavošević, F., Peng, C., Pinski, P., et al. 2017, JChPh, 146, 174108 Chuang, K. J., Fedoseev, G., Qasim, D., et al. 2017, MNRAS, 467, 2552 Peterson, K. A., Adler, T. B., & Werner, H.-J. 2008, JChPh, 128, 084102 Chuang, K. J., Fedoseev, G., Scirè, C., et al. 2021, A&A, 650, A85 Plume, R., Jaffe, D. T., Tatematsu, K., Evans, N. J. I., & Keene, J. 1999, ApJ, Cuppen, H. M., Walsh, C., Lamberts, T., et al. 2017, SSRv, 212, 1 512, 768 Duflot, D., Toubin, C., & Monnerville, M. 2021, FrASS, 8, 24 Pontoppidan, K. M., Boogert, A. C. A., Fraser, H. J., et al. 2008, ApJ, Enrique-Romero, J., Ceccarelli, C., Rimola, A., et al. 2021, A&A, 655, A9 678, 1005 Enrique-Romero, J., Rimola, A., Ceccarelli, C., et al. 2019, ESC, 3, 2158 Potapov, A., Krasnokutski, S. A., Jäger, C., & Henning, T. 2021, ApJ, 920, 111 Enrique-Romero, J., Rimola, A., Ceccarelli, C., et al. 2022, ApJS, 259, 39 Potapov, A., & McCoustra, M. 2021, IRPC, 40, 299 Favre, C., Cleeves, L. I., Bergin, E. A., Qi, C., & Blake, G. A. 2013, ApJL, Richardson, J. O. 2018a, JChPh, 148, 200901 776, L38 Richardson, J. O. 2018b, IRPC, 37, 171 Fedoseev, G., Qasim, D., Chuang, K.-J., et al. 2022, ApJ, 924, 110 Sameera, W. M. C., Senevirathne, B., Andersson, S., Maseras, F., & Ferrero, S., Zamirri, L., Ceccarelli, C., et al. 2020, ApJ, 904, 11 Nyman, G. 2017, J. Phys. Chem. C, 121, 15223 Garrod, R. T., & Herbst, E. 2006, A&A, 457, 927 Sameera, W. M. C., Senevirathne, B., Nguyen, T., et al. 2022, FrASS, 9, Garrod, R. T., & Pauly, T. 2011, ApJ, 735, 15 890161 Hättig, C., Klopper, W., Köhn, A., & Tew, D. P. 2012, ChRv, 112, 4 Shimonishi, T., Nakatani, N., Furuya, K., & Hama, T. 2018, ApJ, 855, 27 Henning, T. K., & Krasnokutski, S. A. 2019, NatAs, 3, 568 Simons, M. a. J., Lamberts, T., & Cuppen, H. M. 2020, A&A, 634, A52 Humphrey, W., Dalke, A., & Schulten, K. 1996, J. Mol. Graph., 14, 33 Skouteris, D., Balucani, N., Ceccarelli, C., et al. 2018, ApJ, 854, 135 Ibrahim, M., Guillemin, J.-C., Chaquin, P., Markovits, A., & Krim, L. 2022, Tielens, A. G. G. M., & Hagen, W. 1982, A&A, 114, 245 PCCP, 24, 23245 Tsuge, M., & Watanabe, N. 2021, Acc. Chem. Res., 54, 471 Ishibashi, A., Hidaka, H., Oba, Y., Kouchi, A., & Watanabe, N. 2021, ApJL, Vastel, C., Ceccarelli, C., Lefloch, B., & Bachiller, R. 2014, ApJL, 795, L2 921, L13 Wakelam, V., Loison, J. C., Mereau, R., & Ruaud, M. 2017, MolAs, 6, 22 Jaber, A. A., Ceccarelli, C., Kahane, C., & Caux, E. 2014, ApJ, 791, 29 Weigend, F., & Ahlrichs, R. 2005, PCCP, 7, 3297 Kamegai, K., Ikeda, M., Maezawa, H., et al. 2003, ApJ, 589, 378 Weigend, F., Köhn, A., & Hättig, C. 2002, JChPh, 116, 3175 Kästner, J. 2014, WIREs Comput. Mol. Sci., 4, 158 Woon, D. E. 2021, Acc. Chem. Res., 54, 490 Kong, L., Bischoff, F. A., & Valeev, E. F. 2012, Chem. Rev., 112, 75 Yang, Y.-L., Green, J. D., Pontoppidan, K. M., et al. 2022, ApJL, 941, L13 Krasnokutski, S. A., Chuang, K.-J., Jäger, C., Ueberschaar, N., & Henning, T. Zamirri, L., Casassa, S., Rimola, A., et al. 2018, MNRAS, 480, 1427 2022, NatAs, 6, 381 Zamirri, L., Corno, M., Rimola, A., & Ugliengo, P. 2017, ESC, 1, 384 Krasnokutski, S. A., Goulart, M., Gordon, E. B., et al. 2017, ApJ, 847, 89 Zamirri, L., Ugliengo, P., Ceccarelli, C., & Rimola, A. 2019, ESC, 3, 1499 Krasnokutski, S. A., Jäger, C., & Henning, T. 2020, ApJ, 889, 67 Zmuidzinas, J., Betz, A. L., Boreiko, R. T., & Goldhaber, D. M. 1988, ApJ, Kruse, H., & Grimme, S. 2012, JChPh, 136, 154101 335, 774 7