MNRAS 516, 2586–2596 (2022) https://doi.org/10.1093/mnras/stac2358 Advance Access publication 2022 August 29 Acetaldehyde binding energies: a coupled experimental and theoretical study S. Ferrero ,1 F. Grieco,2,3 A.-S. Ibrahim Mohamed,2 F. Dulieu ,2 ‹ A. Rimola ,1 ‹ C. Ceccarelli ,4 ‹ C. Nervi,5 M. Minissale6 and P. Ugliengo 5 1 Departament de Quı́mica, Universitat Autònoma de Barcelona, E-08193 Bellaterra, Catalonia, Spain 2 CY Cergy Paris Université, Observatoire de Paris, PSL University, Sorbonne Université, CNRS, LERMA, F-95000 Cergy, France 3 University of Ghent, Department of Physics and Astronomy, Ghent, B-9000, Belgium 4 Univ. Grenoble Alpes, CNRS, IPAG, F-38000 Grenoble, France Downloaded from https://academic.oup.com/mnras/article/516/2/2586/6678576 by 81533202 user on 20 March 2023 5 Dipartimento di Chimica, Università degli Studi di Torino, via P. Giuria 7, I-10125, Torino, Italy 6 Aix-Marseille University, CNRS, PIIM, 13013 Marseille, France Accepted 2022 August 17. Received 2022 August 16; in original form 2022 June 20 ABSTRACT Acetaldehyde is one of the most common and abundant gaseous interstellar complex organic molecules found in cold and hot regions of the molecular interstellar medium. Its presence in the gas-phase depends on the chemical formation and destruction routes, and its binding energy (BE) governs whether acetaldehyde remains frozen on to the interstellar dust grains or not. In this work, we report a combined study of the acetaldehyde BE obtained via laboratory temperature programmed desorption (TPD) experiments and theoretical quantum chemical computations. BEs have been measured and computed as a pure acetaldehyde ice and mixed with both polycrystalline and amorphous water ice. Both calculations and experiments found a BE distribution on amorphous solid water that covers the 4000–6000 K range when a pre-exponential factor of 1.1 × 1018 s−1 is used for the interpretation of the experiments. We discuss in detail the importance of using a consistent couple of BE and pre-exponential factor values when comparing experiments and computations, as well as when introducing them in astrochemical models. Based on the comparison of the acetaldehyde BEs measured and computed in the present work with those of other species, we predict that acetaldehyde is less volatile than formaldehyde, but much more than water, methanol, ethanol, and formamide. We discuss the astrochemical implications of our findings and how recent astronomical high spatial resolution observations show a chemical differentiation involving acetaldehyde, which can easily explained due to the different BEs of the observed molecules. Key words: astrochemistry – molecular data – molecular processes – ISM: abundances. Several experimental and theoretical studies have focused on 1 I N T RO D U C T I O N the acetaldehyde formation routes both in the gas-phase and on the Acetaldehyde (CH3 CHO) was one of the first polyatomic molecules grain-surfaces. Charnley (2004) first proposed that acetaldehyde is discovered in the interstellar medium (ISM; Gottlieb 1973; Fourikis synthesized in the gas-phase via the reaction of the ethyl radical et al. 1974). It is a rather common interstellar molecule, found in (CH3 CH2 ) with atomic oxygen (O). Subsequently, Skouteris et al. both warm and cold environments. For example, it is abundant in hot (2018) found that acetaldehyde could be a daughter of ethanol cores (e.g. Blake et al. 1987; Csengeri et al. 2019; Law et al. 2021), (CH3 CH2 OH), while Vasyunin et al. (2017) proposed a synthesis hot corinos (e.g. Cazaux et al. 2003; Manigand et al. 2020; Yang from methanol (CH3 OH) reacting with CH. Finally, publicly et al. 2021; Chahine et al. 2022), protostellar molecular shocks (e.g. available astrochemical reaction networks (KIDA and UMIST: Lefloch et al. 2017; Codella et al. 2020; De Simone et al. 2020), and Wakelam et al. 2012; McElroy et al. 2013, respectively) also report young discs (e.g. Codella et al. 2018; Lee et al. 2019). Against the an ionic route via the protonated acetaldehyde (CH3 CHOH+ ), theoretical expectations, acetaldehyde is also present in cold prestel- which is in turn formed from dimethyl ether (CH3 OCH3 ) reacting lar cores (e.g. Bacmann et al. 2012; Vastel et al. 2014; Scibelli & with H+ . Vazart et al. (2020) has reviewed all these reactions from Shirley 2020). Finally, acetaldehyde is detected in comets (Crovisier the theoretical and experimental point of view and concluded that et al. 2004; Le Roy et al. 2015; Biver et al. 2021), where the measured only the first two routes are viable (i.e. involving CH3 CH2 and relative abundance with respect to methanol is similar to that found CH3 CH2 OH, respectively), while the last two (i.e. those involving in hot corinos (e.g. Bianchi et al. 2019; Drozdovskaya et al. 2019). CH3 OH and CH3 CHOH+ ) are inefficient in the ISM conditions. Finally, the observed correlation between the derived abundances of acetaldehyde and ethanol in warm and hot sources is in favour of acetaldehyde being the ethanol daughter (Vazart et al. 2020). E-mail: francois.dulieu@cyu.fr (FD); albert.rimola@uab.cat (AR); For the grain-surface formation routes, the situation is more cecilia.ceccarelli@univ-grenoble-alpes.fr (CE) debated. Since the end of last millennium, experiments have found © 2022 The Author(s) Published by Oxford University Press on behalf of Royal Astronomical Society Acetaldehyde BE 2587 that acetaldehyde is formed in UV-illuminated ices consisting of interstellar water ice. These authors found a BE equal to ∼5400 K. water, CO, methanol, and methane (e.g. Hudson & Moore 1997; Recently, Corazzi et al. (2021) reported an experimental study of the Bennett et al. 2005; Öberg et al. 2010; Martı́n-Doménech, Öberg & BE on olivine surfaces covered by an ice of pure acetaldehyde or Rajappan 2020) or other components (e.g. Chuang et al. 2021). mixed with water. They found that the acetaldehyde BE on a mixture Triggered by these experiments, Garrod & Herbst (2006) proposed of iced water-acetaldehyde is 3079 K, substantially different from the formation of acetaldehyde from the combination of the radicals the Wakelam et al.’s estimate. In a very recent combined theoretical HCO and CH3 on the dust grain icy surfaces, when the grain and experimental study, Molpeceres et al. (2022) obtained, for the temperature increases enough to make the two radicals mobile. desorption of acetaldehyde on non-porous amorphous solid water However, a first study by Enrique-Romero et al. (2016) suggested that surfaces, the values of 3624 and 3774 K, respectively. These values the reaction of the two radicals on an amorphous water ice surface are relatively similar to the Corazzi et al.’s value, but still far from actually does not end up into the formation of acetaldehyde but that of Wakelam et al.’s one. There is, therefore, a need to clarify the rather into carbon monoxide (CO) and methane (CH4 ). Subsequent origin of this difference as well as to improve both experimental and Downloaded from https://academic.oup.com/mnras/article/516/2/2586/6678576 by 81533202 user on 20 March 2023 and more accurate studies confirmed that the acetaldehyde formation theoretical estimates. on amorphous water surfaces may or may not occur and is always In this work, we combine state-of-the-art experimental (tem- in competition with the CO + CH4 formation (Rimola et al. perature programmed desorption, TPD) and theoretical (quantum 2018; Enrique-Romero et al. 2019, 2020). Similar conclusions were chemical computations) techniques with the aim to provide possibly reached when considering the reaction on CO-rich ices (Lamberts the most accurate estimate of the acetaldehyde BE on water ice, et al. 2019). Finally, the most recent theoretical study by Enrique- simulating at best the ISM conditions. The article is organized as Romero et al. (2021) found that the efficiency of the acetaldehyde follows. Section 2 reports the experiments, Section 3 the theoretical formation by the coupling of HCO and CH3 is a strong function of computations, and Section 4 discusses the results and comparison of the grain temperature and the diffusion energies of the two radicals, the two methods, as well as the astrophysical implications. with the possibility to being practically zero if their mobility is relatively large (i.e. a diffusion/BE ratio less than 0.3). In agreement with these theoretical predictions, a very recent and sophisticated 2 L A B O R AT O RY E X P E R I M E N T S study by Gutiérrez-Quintanilla et al. (2021) found no formation The experiments were performed with the FORmation of MOLecules of acetaldehyde on an amorphous water ice enriched with HCO in the InterStellar Medium (FORMOLISM) setup. The apparatus is and CH3 . In their experiment, Gutiérrez-Quintanilla et al. (2021) meant to study the reactivity of atoms and molecules on surfaces trapped radicals, formed via photolysis of methanol ice, in an argon of astrophysical interest, under the conditions of temperatures and matrix at 14 K and identified them via electronic paramagnetic pressures similar to those of the ISM. The practicalities of the resonance spectroscopy. Once liberated from the matrix, the various experimental setup are described here but more details are given radicals combined giving rise to several species, but not acetalde- in previous works (Amiaud et al. 2007; Congiu et al. 2012). hyde. More recently, non-energetic processes on the grain surfaces have been found to possibly produce acetaldehyde in CO-rich ices (e.g. Fedoseev et al. 2022). Adding confusion to this situation, 2.1 Methods Hudson & Ferrante (2020) noticed that the infrared (IR) bands, on which the identification (frequencies) and quantification (band FORMOLISM is composed of an ultra-high vacuum (UHV) stainless strengths) of acetaldehyde in the various experiments relied, were steel chamber with a pressure of a few 10−11 mbar. The sample holder often incorrect. Finally, this work also emphasized the almost im- is located in the centre of the main chamber, and it is thermally possibility for astronomical observations to be able to identify frozen connected to a closed cycle Helium cryostat. The temperature of acetaldehyde. the sample is measured in the range of 10–800 K by a calibrated Whatever the formation mechanism, either in the gas-phase or on titanium diode connected to the sample holder. This one is made of the grain surfaces, acetaldehyde would freeze on to the cold dust a 1-cm diameter copper block (18-mm long), which is covered with grain mantles and would be removed from them only when the a highly oriented pyrolytic graphite (HOPG, ZYA-grade) sample. dust temperature reaches the acetaldehyde sublimation temperature, The HOPG is a model of an ordered carbonaceous material, carbon which is governed by its binding energy (BE: sometimes it is also atoms in a hexagonal lattice, mimicking some aspects of interstellar called desorption energy) to the water ice, and its pre-exponential dust grains analogues. The HOPG grade (10 mm diameter–2 mm factor (see Section 4). Please note that the sublimation temperature thickness) was firstly dried in an oven at about 100◦ C for 2 h, and can be reached due to both thermal and non-thermal processes, then cleaved several times in air using the ‘Scotch tape’ method at such as the desorption caused by the interaction of the cosmic- room temperature to yield several limited defects and step edges rays, which permeate the Milky Way, with the dust grains. In the (Chaabouni et al. 2020). The HOPG was glued directly on to the latter case, for example, only a fraction of the grain is heated up and copper finger and dried in ambient conditions for about an hour acetaldehyde would be desorbed if the reached temperature is larger before inserting it in the chamber and closing the set-up. Once high than its sublimation temperature. Therefore, in practice, whether vacuum is reached (<10−7 mbar) in few minutes, the sample is dried acetaldehyde is in the gas-phase (where it can be observed via its under vacuum during 3 h at 350 K. Then the HOPG sample and the rotational lines) or frozen on to the grain mantles (where it has not whole system is baked out at 100◦ C for few days in order to remove been detected so far) is completely governed by its BE. Moreover, BE any adsorbed contaminants and reach the UHV base pressure. Before always enters in an exponential form, so that its accurate estimation starting any experiment, the system is slowly heated up to 720 K to is essential to properly assess whether and how much acetaldehyde maximize the degassing rate and ensure a very clean sample surface. would be gaseous. FORMOLISM is equipped with a quadrupole mass spectrometer So far, there are very few specific studies on the acetaldehyde BE. (QMS). It allows both the detection of species desorbing from the To our best knowledge, Wakelam et al. (2017) is the first theoretical sample during a TPD and, when placed in front of the beam-line, the study, but it only considered one water molecule to simulate the characterization and the calibration of molecular beams. MNRAS 516, 2586–2596 (2022) 2588 S. Ferrero et al. The CH3 CHO (Sigma Aldrich, purity higher than 99.5 per cent) molecular beam is prepared in a triply differentially pumped beam- line aimed at the sample holder. It is composed of three vacuum chambers connected together by tight diaphragms of 3 mm diameters. Molecular pressure injected in the gas line is of about 1.4 mbar. In the expansion chamber (first stage), the pressure is in the 10−5 mbar range, while in the main chamber, the rise of the pressure is not measurable (<10−11 mbar). The CH3 CHO on HOPG experiments have been used as reference and preliminary studies before investigating the desorbing behaviour of the molecule on two different D2 O ice substrates formed on the sample. In this study, we have deposited CH3 CHO molecules on Downloaded from https://academic.oup.com/mnras/article/516/2/2586/6678576 by 81533202 user on 20 March 2023 HOPG by using only one beam-line oriented at 57◦ with respect to the surface of the sample, while D2 O ice films have been deposited on HOPG by using a separated channel and a micro-capillary array doser moved close to the sample holder. It allows to control the ice deposition rates and to avoid to raise the pressure in the whole chamber. The two different D2 O ices are, respectively, compact non- Figure 1. Set of four smoothed TPDs (m/z = 44) for different time porous amorphous deuterated water (ASW-c) and (poly)crystalline depositions of acetaldehyde on HOPG: 1.5 min (0.4 ML), 3 min (0.8 ML), deuterated ice (PCI). ASW-c is formed when the sample is held 6 min (1.7 ML), and 12 min (3.3 ML). at 110 K, while PCI layers have been formed by depositing D2 O at 110 K, heating until 150 K with a ramp rate of 5 K min−1 and waiting corresponding to 1.5 and 3 min). It initially increases in intensity for few minutes to be sure that the crystallization has happened, as with coverage but saturate for longer depositions. After 6 min of detected by the change in the desorption rate (Speedy et al. 1996). deposition, the low-temperature peak clearly appears. The TPD after For both ices, about 30 monolayers (ML) have been deposited on 12-min deposition shows both peaks equally intense, although the HOPG, while the pressure in the main chamber never exceeded low-temperature peak is slightly broader. The high-temperature peak 5 × 10−8 mbar. After the CH3 CHO deposition phase at 45 K, we is attributed to sub-ML regime and is first populated, and, when the used the TPD technique by warming-up the sample and detecting ML is saturated and the multilayer regime starts, the second peak the desorbing species from the surface through mass spectrometry. at lower temperature appears. By following the TPD experiments TPD were obtained from 45 to 220 K for the CH3 CHO on HOPG on HOPG, an ML of molecules on the substrate is believed to form experiments and only to 140 K for the CH3 CHO on water ice ones, after ca. 4 min of deposition. We define this time to calibrate the rest in order to avoid D2 O desorption after 150 K. In all cases, a linear of the analyses. As HOPG and ASW-c have about the same surface heating rate of 12 K min−1 has been programmed to heat up the density of sites, the same time deposition is used to characterize the sample. Because of the poor thermal contact between the sample coverage on ice, and to derive the BEs of CH3 CHO as a function of holder and the copper block, the temperatures registered during the the coverage. TPD were recalibrated by depositing different gases of known BEs In order to derive the BEs on HOPG and on the two different ices, (O2 , Kr, D2 O). The accuracy is about 1 K for the values of temperature the three TPDs for the 4-min depositions on HOPG, ASW-c, and PCI below 125 and 0.2 K for the ones above. The actual new ramps have been selected. By proceeding this way, the interaction between obtained after calibration are taken into account in the simulations a ML of deposited CH3 CHO and each substrate can be studied. and analyses. Fig. 2 shows the results for one of them, i.e. the experiment of 4-min With the QMS, the same molecule can be detected in different m/z deposition of CH3 CHO on PCI, and the simulation obtained by fitting corresponding to the fragmentation of the parent molecules after the the data. The procedure is detailed in Chaabouni et al. (2018). The electron impact (here at 30 eV). The main fragments under study here calculation uses 11 independent TPDs, starting from 4500 K, with an are m/z = 44 (molecular mass of CH3 CHO+ ) and m/z = 29 (HCO+ , energy gap between each TPD of 150 K (i.e. 4500, 4650, 4800 K...). the main fragment of acetaldehyde). The desorption and degree of Each individual TPD is calculated using simple Arrhenius kinetics, crystallization of D2 O ice films have been followed through the signal named the Polanyi–Wigner equation: of m/z = 20.   dN Edes r(T ) = − = AN n exp − (1) dT T 2.2 Results where r(T) is the desorption rate (in ML s−1 ), N is the number density Experiments on HOPG have been used mainly as reference, in order of molecules adsorbed on the surface (ML), and n is the order of the to understand the behaviour of the acetaldehyde adsorption and the desorption equal to 1 in our case. A is the pre-exponential factor (s−1 ), deposition time needed to have a ML on the substrate. Seven measure- T is the temperature of the surface (in K), and Edes is the activation ments corresponding to increasing deposition times (1, 1.5, 3, 4, 5, energy for desorption (in K). If there is no reorganization of the 6, and 12 min) have been carried out on HOPG to study the evolution surface,1 Edes is equal to BE. First-order desorption TPD profiles are of the TPDs from a sub-ML regime to a multilayer one. Fig. 1 shows independent of the surface population, so that each independent BE only four TPD curves belonging to four different deposition times can be weighted, and finally a distribution of BEs can be found. We using the same first stage pressure to introduce a constant flux of have set the pre-exponential factor at 1.1 × 1018 s−1 (see Section 4). CH3 CHO in the system. There are two main features appearing in the graph: a low-temperature peak between 105 and 110 K, and a second peak centred around 117.5 K. At low deposition times, 1 This is 100 per cent true on graphite, but this is an assumption/approximation only the peak at higher temperature appears (black and red curves in case of acetaldehyde on water. MNRAS 516, 2586–2596 (2022) Acetaldehyde BE 2589 Downloaded from https://academic.oup.com/mnras/article/516/2/2586/6678576 by 81533202 user on 20 March 2023 Figure 2. Simulation of 11 TPDs (red line) used to fit the experimental TPD (black line, raw data not smoothed) obtained from 4 min (1 ML) of CH3 CHO deposition on PCI (pre-exponential factor = 1.1 × 1018 s−1 , Emin = 4500 K, Emax = 6000 K by steps of 150 K). Table 1. Desorption energy values of the 11 simulated TPDs obtained for three sets of 4 min CH3 CHO depositions: on HOPG, ASW-c, and PCI (pre- exponential factor: 1.1 × 1018 s−1 ). HOPG c-ASW PCI Edes (K) Population Population Population (per cent) (per cent) (per cent) 4500 0 0 <4 4650 10 16 28 4800 0 32 27 4950 0 10 12 5100 34 12 12 5250 66 7 9 5400 0 8 6 5550 0 9 6 5700 <4 6 8 5850 <4 7 <4 6000 <4 <4 <4 Figure 3. Energy distributions of 11 simulated TPDs for the 4 min CH3 CHO depositions on HOPG, ASW-c, and PCI taken from Table 1. The arbitrary choice of binning the energy distribution (basis of 11 BEs) have been optimized with the aim to have a good fitting of all the three experiments under analysis. Fig. 2 gives a qualitative idea of the simulation for the case of the PCI substrate. The best values water ice (even for the PCI case). There is a larger distribution of found for the 11 TPDs, also known as energy distribution set for each population covering almost all the possible energy values proposed substrate, are reported in Table 1. The values are here expressed as by the simulation. The ASW-c substrate presents the largest distri- a function of the population, in percentage, having that specific BE. bution of BEs. The energy distribution histogram associated with We estimate the accuracy of the method to be around a few per cents, the desorption of CH3 CHO on PCI presents, instead, a more peaked mostly due to the noise-to-signal ratio and the presence of a small distribution towards lower values of energy (mainly below 4900 K). background noise. This is due to the reduced number of combination of water molecules The three energy distributions are displayed in Fig. 3. We omit the on the surface. The disorder here is provoked by the possibility of populations below 4 per cent. One can observe the main differences having an O or an H in the vicinity of the adsorption sites, even if the because of the different substrate. For HOPG, there is not an actual collective structure is cubic. On the contrary, ASW-c have a more distribution, but it rather shows almost only two main energy values, disordered nature and, accordingly, the adsorption sites may have a which are associated with the sub-ML regime (at 5100–5250 K). As more variable number of water molecules strongly interacting with can be noticed by the 10 per cent population at 4650 K, for the 4 min CH3 CHO (see below). Finally, we calculate the weighted average for depositions the multilayer regime slightly started already. This is in each case to estimate the difference in BEs for the different substrates. agreement with the accuracy of the determination of the flux, which We obtain the following result: 5150 K for HOPG (range: 4650– is usually estimated around 15 per cent. 5250 K), 5080 K for ASW-c (range: 4650–5850 K), and 4990 K for The distributions of BEs of CH3 CHO on ices present different PCI (range: 4650–5700K). This corresponds to what one can expect characteristics and are a signature of the more disordered nature of by looking at the trend in the desorption profiles. MNRAS 516, 2586–2596 (2022) 2590 S. Ferrero et al. 3 Q UA N T U M C H E M I S T RY C A L C U L AT I O N S optimized structure was used as input parameter to derive the pure acetaldehyde periodic solid-state structures, which was achieved by Simulations of the adsorption of an acetaldehyde molecule on using the programs of DFTB+ (version 20.2.1; Hourahine et al. crystalline and amorphous ice models, and as a component of a 2020), including Self Consistent Charge (Elstner et al. 1998) and pure acetaldehyde surface have been carried out in order to (i) gain the D4 dispersion model by Grimme (Caldeweyher, Bannwarth & insights of the adsorption process at an atomic level and (ii) calculate Grimme 2017; Caldeweyher et al. 2019), and VASP 5.4.4 (Kresse & BEs of the species on ices with the purpose to make comparisons Hafner 1993, 1994; Kresse & Furthmüller 1996a, b) using PAW PBE with TPD experiments. pseudopotentials (Kresse & Joubert 1999). Solid-state structure prediction was performed using the USPEX 3.1 Methods (version 10.4.1) evolutionary algorithm (Oganov & Glass 2006; Oganov, Lyakhov & Valle 2011; Lyakhov et al. 2013), with embed- 3.1.1 Computational details ded the topology crystal structure generator (Bushlanov, Blatov & Downloaded from https://academic.oup.com/mnras/article/516/2/2586/6678576 by 81533202 user on 20 March 2023 Oganov 2019), employing the DFTB+ code for geometry opti- All the calculations were performed with the ab initio CRYSTAL17 mizations. The first generation of crystal structures were randomly code (Dovesi et al. 2018). The code can simulate any kind of system, generated, while subsequent generations were obtained by applying from non-periodic molecules to full periodic 3D crystalline systems, the genetic algorithm embedded in USPEX coupled with different employing atom centred Gaussian basis sets for the description of variation operators, in particular rot-mutation and lattice mutation. the electronic structure. In this work, we carried out DFT-based static The search space was limited from 1 to 4 molecules per cell. Cal- optimizations using the BFGS optimization algorithm, relaxing both culations proceeded for 23 generations and 570 structures, in which the cell parameters and the atomic positions. These calculations were USPEX identified the best structure, which contains 2 molecules performed with the PBEsol0-3c functional (Doná, Brandenburg & per cell and a final volume of 118.77 Å3 . Since periodic DFTB+ Civalleri 2019), which makes use of an Ahlrichs’ polarized valence- calculations tend to excessively shrink the unit cell, subsequent single zeta basis set in order to reduce the computational cost. calculations with USPEX but this time employing VASP calculations Then, on the PBEsol-3c optimized geometries, single-point energy were performed using as initial seeds the best structures found by calculations were carried out with the B3LYP functional (Lee, DFTB+. This proceeded for 28 generations and 723 structures, Yang & Parr 1988; Becke 1993) added with the Grimme’s D3(BJ) resulting in the final optimized structure characterized by a = 4.960 correction (to account for dispersive forces (Grimme et al. 2010; Å, b = 5.655 Å, c = 4.536 Å, and α = 89.55◦ , β = 89.70◦ , γ = Grimme, Ehrlich & Goerigk 2011) with the aim to refine the BEs 127.17◦ (with a final volume of 127.17 Å3 ) at a higher level of theory. These refinement calculations have been The resulting bulk structure has been optimized at the PBEsol0-3c performed using an Ahlrichs’ VTZ basis set added with a double set level and it is depicted in Fig. 5(a). Once we obtained the bulk, we of polarization functions. For the sake of accuracy, the calculations cut it along the (010) direction to obtain a 2D periodic slab model, carried out with the B3LYP-D3(BJ) method have been corrected which does not posses a large dipole along the non-periodic direction for the basis set superposition error, adopting the a posteriori (z axis (Figs 5(b) and (c)). counterpoise correction. Vibrational harmonic frequencies were calculated at the PBEsol0- 3c level using the finite differences method. A partial Hessian approach was used to reduce the computational cost of the calcu- 3.2 Results lations. Thus, the vibrational frequencies were calculated on the optimized geometries only for a fragment of the entire system, which 3.2.1 Adsorption of acetaldehyde on ice structures included the acetaldehyde molecule and the closest water molecules We first simulate the acetaldehyde adsorption process by placing the interacting with it. From this vibrational frequency calculations, molecule on the 1x1 crystalline surface, in which a single hydrogen the zero-point energy (ZPE) correction terms were included in the bond (H-bond) between a dangling hydrogen (dH) of the ice surface calculations of the BEs, hereafter referred to as BE(0). and the O atom of acetaldehyde is established. The size of the 1x1 unit cell is very small (it only contains one dH) and accordingly adsorption of acetaldehyde ends up with a structure resembling a 3.1.2 Water ice and pure acetaldehyde periodic structures mono-layer coverage adsorption situation (see Fig. 6(a)), in which The crystalline and amorphous water ice surface models used in this lateral interactions between acetaldehydes of adjacent unit cells take article have already been described in Ferrero et al. (2020) and are place due to the periodic boundary conditions. By using the 2x1 reported in Fig. 4. Briefly, the crystalline slab model derives from unit cell, such lateral interactions are removed, giving rise to a the (010) surface of the proton-ordered P-ice. For this surface, two low coverage regime, in which a single acetaldehyde is adsorbed different unit cells have been used, the 1x1 and the 2x1 (Figs 4a and b, on the surface (see Fig. 6(b)). The obtained ZPE-corrected binding respectively), this way covering both high and low coverage regimes. energies, BE(0), are 7253 and 5194 K for the 1x1 and 2x1 unit cell The amorphous water ice surface was constructed by joining different cases, respectively (reported in Table 2). Since the amorphous ice water clusters from Shimonishi et al. (2018), in which, upon imposing is larger (60 water molecules) and presents a disordered surface periodicity and optimizing the structure, the resulting surface model morphology, this surface model presents more adsorption sites. presents a cavity (Fig. 4c). We sampled some of them by placing acetaldehyde molecule in Since the crystalline structure of pure acetaldehyde is not available different starting positions and then optimizing the structures. Out yet, in order to simulate a pure acetaldehyde slab we resorted to of a total of 14 optimizations, we obtained 9 different minima, in the crystal structure prediction algorithm. First, the acetaldehyde which in all the cases acetaldehyde establishes H-bond interactions molecule was optimized in gas phase with the B3LYP-D3(BJ) with dH of the amorphous surface. The different calculated BE(0) method combined with the optimized def2-SVP basis set with values are reported in Table 2, which spans the ca. 2800–6000 K the Gaussian 09 (Frisch et al. 2010) program. This gas-phase range. Figs 6(c) and (d) show the optimized structures for the MNRAS 516, 2586–2596 (2022) Acetaldehyde BE 2591 Downloaded from https://academic.oup.com/mnras/article/516/2/2586/6678576 by 81533202 user on 20 March 2023 Figure 4. Panel(a) : top view of the 1x1 unit cell of the P-ice (010) surface. Panel (b): top view of the 2x1 supercell of the P-ice (010) surface. Panel (c): side view of the amorphous ice surface. The employed 1x1 cell and 2x1 supercell of panels A and B are highlighted in blue. complexes presenting the highest and the lowest BE(0) values, 4 DISCUSSION respectively. 4.1 The importance of the pre-exponential factor Before discussing the chemical physics, or any methodological bias, it is important to be able to compare the values derived by 3.2.2 Adsorption of acetaldehyde on pure acetaldehyde surface the experimental and theoretical methods, which in the case of A single acetaldehyde molecule was adsorbed on the 2x2 super cell desorption is not straightforward. To understand this point, we will of the (010) pure acetaldehyde slab model. Its BE(0) was computed first recall the framework in which the BE values are obtained in similarly as described for the adsorption on the water ice structures. each approach. Unfortunately, the three communities (laboratory, The calculated BE(0) for this case is 2650 K. This value is a lower quantum chemistry, and astrochemistry) have their own underlying limit of the BE(0), as the isolated molecule is engaged in very definitions, which we will discuss first. few interactions with the companion surface. The upper limit of the BE is the cost needed to extract a single acetaldeyde molecule 4.1.1 The experimental point of view from the external layer of the (010) surface. In this case, the BE(0) is a compromise between the rupture of a consistent number of Experiments do not directly measure a desorption energy, but a intermolecular interactions holding acetaldehyde within the slab and desorbing flux, and, by applying equation 1 (see Section 2), a BE is the geometry relaxation of the cavity created in the surface. The derived (which is supposed to be equal to the desorption energy in BE(0) was, therefore, calculated by subtracting the energy of the most of the cases). However, to do so, one needs to set a value for the relaxed surface with the cavity plus the energy of the extracted pre-exponential factor, since the equation has two parameters: the acetaldehyde from the energy of the pristine perfect surface. The pre-exponential factor A and the BE (if the desorption order is set result is a BE(0) = 5692 K, that is, almost twice as large as the to n = 1). Here, following the recommendation by Minissale et al. acetaldehyde BE(0) at the clean (010) surface (vide supra). (2022), we adopt the derivation of the pre-exponential factor based MNRAS 516, 2586–2596 (2022) 2592 S. Ferrero et al. Downloaded from https://academic.oup.com/mnras/article/516/2/2586/6678576 by 81533202 user on 20 March 2023 Figure 6. Panel (a): top view of the optimized geometry of one acetaldehyde molecule adsorbed on the 1x1 cell of the P-ice (010) surface model. Panel (b): top view of the optimized geometry of one acetaldehyde molecule adsorbed on the 2x1 supercell of the P-ice (010) surface model. The 1x1 cell and the 2x1 supercell are highlighted in blue. Side view of the optimized geometries of acetaldehyde adsorbed on the amorphous ice surface model presenting the highest (Panel (c)) and the lowest (Panel (d)) BEs. Table 2. Calculated (ZPE corrected) BE(0) (in K) of acetaldehyde on the crystalline and the amorphous ice surface models. Crystalline ice BE(0) 1x1 cell 7253 2x1 cell 5194 Amorphous ice Site1 5208 Site2 3366 Site3 5700 Figure 5. Panel (a): PBEsol0-3c optimized structure of the bulk model for Site4 4708 pure acetaldehyde. Panel (b): top view of the 1x1 unit cell of the (010) pure Site5 2809 acetaldehyde slab surface model. Panel (c): top view of the 2x2 supercell of Site6 4028 the (010) pure acetaldehyde slab surface model. The employed unit cells are Site7 3450 highlighted in blue. Site8 4597 Site9 6038 on the transition state theory (TST). This theory takes into account both the rotational and translation partition functions of the desorbing molecules and, therefore, allows including in the calculation of the We used the following formula: pre-exponential factor, A TST , the entropic effect associated with the √ kinetic desorption rate. We briefly describe the salient points of the kb T q ‡ kb T A π  2 3/2  calculation of A TST (more details can be found in Tait et al. 2005). ATST = = 8 π kb Tpeak Ix Iy Iz (2) h qads h 2 σ h3 MNRAS 516, 2586–2596 (2022) Acetaldehyde BE 2593 where qads and q‡ are single-particle partition functions for the ad- T ST Therefore, the two equations coincides when the qvib ratio is equal sorbed (initial) state and the transition state, respectively, calculated to unity. The application of the full TST treatment to acetaldehyde, at the temperature Tpeak . A is the surface area per adsorbed molecule. limited to one representative case related to the P-ice model (with Here, like in other experiments or computational studies, it is fixed a BE(0) of 60.3 kJ mol−1 , corresponding to the complex shown in to 10−19 m2 (the inverse of the number of surface sites per unit area). Fig.6(a)), gave the ratio between the full TST treatment and the Ix , Iy , and Iz are the principal moments of inertia for the rotation Tait et al. (2005) pre-factor close to unity for temperatures ≤50 K, of the particle, obtained by diagonalizing the inertia tensor of the decreasing by ∼1 and 2 orders of magnitude at 100 and 200 K, free acetaldehyde. They are fixed to Ix = 52.46, Iy = 46.86, Iz respectively. Therefore, considering the desorption temperature of = 8.78 amu Å2 . The symmetry factor, σ , is the number of different 107 K for acetaldehyde, the Tait et al. (2005) approximation is but indistinguishable rotational configurations of the particle and it relatively good within around one order of magnitude of the value of is fixed to 1 in the case of acetaldehyde (as it belongs to the Cs ≈ 1018 s−1 while being much simpler to apply. symmetry point group). The thermal wavelength of the molecule, Another advantage of adopting equation (2) over the full QM Downloaded from https://academic.oup.com/mnras/article/516/2/2586/6678576 by 81533202 user on 20 March 2023 , depends on its atomic mass (44 amu) and on the peak of the TST one is its explicit dependence on Tpeak . For these reasons, in desorption energy (Tpeak = 115 K), and it is 24.5 pm. By using these the following, we will adopt equation (2) for the calculation of the parameters, ATST = 1.07 × 1018 s−1 . We emphasize, however, that pre-exponential factor. ATST is a function of Tpeak . As shown in Fig. 1, the temperature of the peak depends on the surface coverage. For example, when using Tpeak = 108 K, ATST is equal to 8.58 × 1017 s−1 , while fixing Tpeak 4.1.4 Comparison with previous experimental estimates = 117 K then ATST = 1.13 × 1018 s−1 . In this study, we choose ATST Now we have a tool to compare BEs provided by different values = 1.1 × 1018 s−1 , which is the average value. of pre-exponential factors. Corazzi et al. (2021) recently reported an experimental value of BE = 3100 K of acetaldehyde co-deposited with water on a substrate made of micrometer-sized olivine grains. 4.1.2 The astrochemical modelling point of view It obviously differs from the measured values here and it is lesser than any calculated value. However, these authors used a fixed value Often, the pre-exponential factor is calculated using the harmonic of A = 1012 s−1 . To be able to compare with our results, we use oscillator approximation introduced by Hasegawa, Herbst & Leung equation (4) and find that the couple BE = 3079 K and A = 1012 (1992) and here called AHH : s−1 corresponds to the couple BE = 4680 K and A = 1.1 × 1018 s−1 ,   that is exactly what has been found here for the multilayer energy 2 ns BE 1/2 AHH = (3) of acetaldehyde. This example shows how important is to take into π2 m account, not only the BEs, but also the pre-exponential factor that where ns is the surface density of sites and m is the mass of the allows us to calculate a desorption flux, to reproduce experiments or adsorbed species. Using this approximation, which does not take into to simulate the desorption in ISM conditions. account the fact that the adsorbed molecule does not rotate whereas the desorbed one does, we find AHH = 1.3 × 1012 s−1 if we choose an arbitrary value of BE = 4500 K. In addition, in the Hasegawa 4.2 Comparison of experimental and theoretical results et al. (1992) formulae, the value of AHH depends exclusively on the The aim of this section is to compare the experimental and theoretical BE of the adsorbate and is calculated directly from it, so that, in the approaches. We will do that for the pure acetaldehyde, crystalline, models, there is a unique parameter, BE, and not the A and BE pair, as and amorphous water ices, separately. determined in experiments. We stress that the chosen pre-exponential factor A can strongly affect the BE value determined using the TPD technique, since they are actually coupled (see equation 1) and are 4.2.1 Pure acetaldehyde ice somewhat degenerated (monotonic). To compare the values obtained By comparing theory with experiments, in general, we compare a with different pre-exponential factors A, we can rescale the BE values static and ‘uni-molecular’ calculation with a measurement that is using AHH and the following formula: both dynamic and averaged over a large population and a large BEHH = BETST − Tpeak ln(ATST /AHH ) (4) number of situations. We will first compare the case of the pure acetaldehyde ice. Here, experiments provide a BE of 4650 K, while where BEHH and AHH (BETST and ATST ) are the pair BE and pre- calculations dealing with the ideal model of a molecule above the exponential factor calculated with the harmonic oscillator approxi- (010) acetaldehyde surface gives a BE(0) of 2650 K. Given the mation of Hasegawa et al. (1992) or the TST proposed by Tait et al. robustness of the measurement (confirmed by Corazzi et al. 2021), (2005). one would conclude that what is measured does not correspond to what is calculated. In reality, when a molecular film is heated, it will constantly reorganize itself, and the situation of a molecule on top 4.1.3 The theoretical point of view of the surface is not the most energetically stable configuration. The other extreme case corresponds to the calculation of the BE(0) of A full quantum mechanical (QM) evaluation of the pre-exponential one acetaldehyde molecule extracted from the surface slab, which factor can be derived by the statistical mechanical treatment as in the is 5692 K (see above). We think that this last computed value is TST, considering the availability of the harmonic frequencies for all probably overestimated with respect to the experimental BE, as we involved systems. The QM pre-exponential factor ν TST (T) is related only focused on the process occurring at the rather stable (010) to the ATST of equation (2) by: surface without considering more defective surfaces, from which the ‡ desorption can occur from kink and edges leaving the molecule less qvib (M)‡ qvib (S) νT ST (T ) = AT ST . (5) engaged with the underneath layers. qvib (C)   T ST qvib MNRAS 516, 2586–2596 (2022) 2594 S. Ferrero et al. As a first conclusion, therefore, it can be said that the present 4.3 Comparison of the acetaldehyde BE with those of other calculations provide good high and low boundary values to the important interstellar molecules experimental results, but that the dynamic (in the case of experiments) The detailed calculations and measurements of the acetaldehyde versus the static (in the case of calculations) nature can make any BEs on water ices show a consistent but complex picture. First, direct comparison somewhat misleading. as already reported in other works (e.g. Dulieu et al. 2005; Ferrero et al. 2020), BE on an icy surface is not a single value, but there is rather a distribution of BEs caused by the different sites to which 4.2.2 CH3 CHO desorbing from crystalline water ice acetaldehyde can be adsorbed. Second, using the BE values without Although we provided two BE values because of the use of two paying attention to its associated pre-exponential factor can lead different unit cell sizes (i.e. 1x1 and 2x1 supercell), here, for the to inconsistencies, if not mistakes. That being said, most of the sake of comparison with experiments, we take the value of BE(0) astrochemical models use a single value of BE and the AHH pre- = 5194 K obtained using the 2x1 supercell, as it is more realistic. exponential factor, recalculated from it following equation (3), which Downloaded from https://academic.oup.com/mnras/article/516/2/2586/6678576 by 81533202 user on 20 March 2023 In experiments, we find a distribution of BEs. It is not a bias of the is not a guarantee of correctness. method, as it was possible to derive a unique value of the BE for the Another question is: which single value of BE to choose from the HOPG surface. Actually, in experiments, even for a PCI, the surface observed distribution? To choose a single value in the distribution, is disordered. Indeed, the water molecules at the outermost positions one needs to know the nature of the ice, for instance, if the surface of the surface almost randomly alternate dH pointing outside or inside will be fully or partly covered. Acetaldehyde is a frequently observed the solid. Therefore, even for crystalline water ice, distributions of BE molecule in the gas phase, but it has not been detected yet in the ice are usually measured whatever is the adsorbate (Amiaud et al. (2007), mantles because its concentration is probably low (but see also the Noble et al. (2012a), Nguyen et al. (2018)). The calculations, made discussion in the ‘Introduction’). Therefore, the coverage of this on a perfectly regular or periodic crystal, are not able to reproduce molecule on the surface of icy grains should remain low and the low this disorder. coverage side of the distribution shall be preferred. Moreover, the In experiments, the largest population is found for BE = 4650 K interval of values that we propose overlaps the BE extrapolated by and it corresponds to the BE of an acetaldehyde multilayer. This Wakelam et al. (2017; 5400 K), which turns out to be correct if one means that CH3 CHO molecules prefer to rearrange themselves, adopts the pre-exponential factor of A = 1.1 × 1018 s−1 . However, if probably creating some sort of small clusters, rather than spreading the astrochemical model uses the harmonic oscillator approximation over the surface. In other words, acetaldehyde do not wet properly and a value of AHH ∼ 1012 s−1 , a value around 3800 K shall be the water ice surface. The low coverage values, which correspond preferred. more to the calculations made for a unique molecule, lay between The advantage to use the couple BE = 5400 K and A = 1.1 × 1018 4950 and 5700 K. This is in excellent agreement with the calculated s−1 is that one can then compare this BE with that derived by value of BE(0) = 5194 K. We note that population at 5700 K could the theoretical calculations and to the already published BE val- also correspond to defaults or steps in the poly-crystalline assembly. ues of other molecules (Minissale et al. 2022). As an example, CH3 CHO can be compared to H2 CO, which seems to have a similar desorption behaviour, exhibiting a non co-desorption with water 4.2.3 CH3 CHO desorbing from amorphous water ice (Noble et al. 2012b). The values for H2 CO are: BE = 4117 K (A = 8.29 × 1016 s−1 ). On the contrary, both ethanol and methanol Both experiments and calculations show a broad distribution of BEs. exhibit co-desorption with water and have larger BEs, BE = 7000 K Calculations demonstrates that, compared to the crystalline case, (A = 3.89 × 1018 s−1 ; Dulieu, Vitorino & Minissale, private some sites are more energetically favourable (two over nine), one is communication) and BE = 6621 K (A = 3.18 × 1017 s−1 ; Bahr, about equal, and the other six are less favourable sites for adsorption. Toubin & Kempter 2008; Minissale et al. 2022), respectively. Finally, There is no doubt that such less bounded sites exist, but in exper- formamide (NH2 CHO) has a refractory behaviour with respect to iments they will not be populated, as molecules reorganize during water. It desorbs at higher temperatures from bare surfaces because the heating phase and tend to occupy the available most favourable of its high BE = 9561 K (A = 3.69 × 1018 s−1 ; Chaabouni et al. adsorption sites. This has been well experimented and documented 2018; Minissale et al. 2022). One obtains the same substantial result as the ‘filling behaviour’, for relatively volatile substances (Kimmel when considering the theoretical BEs calculated by Ferrero et al. et al. 2001; Dulieu et al. 2005). In the experiments, for all adsorbates, (2020), although they are BE distributions. the minimum BE measurable is set by the limit of the BE of the multilayer. This is why we observe that half of the population of acetaldehyde desorbs from sites with a BE very close to that of pure 4.4 Astrochemical implications CH3 CHO films. Indeed, this accumulation at the lower part of the BE distribution corresponds, in the calculations, to the six (out of From an astrochemical point of view, two points are particularly nine) sites that have a BE lower than 4650 K (the multilayer limit). relevant. The first point regards the presence of acetaldehyde in Once again, we find here that the theoretical calculations are in very cold (∼10 K) objects (e.g. Bacmann et al. 2012; Vastel et al. 2014; good agreement with what is obtained experimentally. They show Scibelli & Shirley 2020; Zhou et al. 2022). The relatively large BE perfectly the extent of the distribution (up to 6000 K), and propose (4800–6000 K) makes it difficult to explain the presence of gaseous a good sampling of sites. Finally, we point out here that we do acetaldehyde if formed on the grain-surfaces, and would rather favour not observe any co-desorption of acetaldehyde with water. All the a gas-phase formation. Of course, this may just move the problem of acetaldehyde desorbs prior to the water. This is consistent with the the presence of the reactants needed to synthesize acetaldehyde from rather low values of BE calculated for the interaction with water, them, namely ethyl radical and/or ethanol. Non-thermal mechanisms lower than the multilayer BE measurements of the acetaldehyde could be at play such as cosmic-ray bombardment that would not be film alone. This molecule would somehow have a low hydrophilic chemically selective (Dartois et al. (2019)), whereas the chemical character. desorption initiated by hydrogenation on the grain surface may have MNRAS 516, 2586–2596 (2022) Acetaldehyde BE 2595 different efficiencies depending on the molecule (Minissale et al. 4600–4800 K. In amorphous water ice, the BE ranges between about 2016). 4600 and 5900 K, with a peak around 4800 K. The weighted average As written in the ‘Introduction’ section, the BE of a given species in the three cases are 5150, 5080, and 4990 K, respectively. determines at what temperature the species remains adsorbed or goes (ii) The theoretical calculations give BE values on the different to the gas-phase. In hot cores/corinos, several species, including sites of the used ice model. BEs spread over 5194 and 7253 K in the acetaldehyde, have a jump in the abundance in the region where case of crystalline ice, and 2809–6038 K in the case of amorphous the dust temperature reaches about 100 K. This is believed to ice. In the pure acetaldehyde ice, two extreme values of 2650 and be caused by the sublimation of the frozen water (e.g. Charnley, 5692 K are obtained. Tielens & Millar 1992; Ceccarelli et al. 2000; Jaber et al. 2014), (iii) We discussed and showed the importance of the pre- which is the major constituent of the icy grain mantles (e.g. Boogert, exponential factor when deriving, comparing, and using BEs derived Gerakines & Whittet 2015). However, the analysis of the emission from experiments and theoretical calculations. Remarkably, when lines of different species and their spatial distributions sometimes the correct pre-exponential factor is used with the derived BE, Downloaded from https://academic.oup.com/mnras/article/516/2/2586/6678576 by 81533202 user on 20 March 2023 suggest that there is a differentiation in the sublimation of different experiments and theory are in fair good agreement. species (e.g. Manigand et al. 2020; Bianchi et al. 2022). (iv) A comparison of the the derived acetaldehyde BEs with those The recent study by Bianchi et al. (2022) definitively shows a of other important species shows that acetaldehyde would desorb at chemical differentiation between the two hot corinos of the SVS13 temperatures lower than those at which water desorbs and even lower protobinary system. Specifically, the analysis of the high-spatial temperatures with respect to formamide. resolution ALMA observations leads to the conclusion that gaseous (v) The large acetaldehyde BE challenges the explanation for its acetaldehyde and formamide are distributed in an onion-like structure gas-phase presence in cold (∼10 K) astronomical objects, especially of the hot corino, with formamide becoming abundant in a warmer if it is formed on the grain surfaces. The problem may be alleviated region with respect to the acetaldehyde region. A natural explanation if it is formed by gas-phase reactions. of this behaviour is that species with larger BE emit lines (mostly) (vi) In hot cores/corinos, the measured and computed BE is in more compact regions, i.e. the region corresponding to their in agreement with the observations of acetaldehyde originating in sublimation temperature, rather than the water sublimation front, regions colder than formamide, whose BE is larger. because the large densities (≥108 cm−3 ) make those species freeze- Finally, our study shows the importance to extend the methodology out back very quickly. Therefore, given their respective BEs (see adopted here to other molecules of astrochemical interest, such as Section 4.3), acetaldehyde is emitted in a region more extended and ethanol or formamide, which are nowadays routinely observed in colder than that of formamide. astronomical objects and that show a spatial segregation probably More in general, by considering their respective BEs, formalde- due to their different BE. hyde (H2 CO) and acetaldehyde would desorb before water, ethanol, and methanol with water, but formamide has to wait for higher average temperatures, given its much higher BE. Our new estimates AC K N OW L E D G E M E N T S of the acetaldehyde BE, slightly lower than water, would suggest that This project has received funding within the European Union’s acetaldehyde should be found approximately in the regions where Horizon 2020 research and innovation programme from the Eu- water and methanol are present too, which is what has been so ropean Research Council (ERC) for the projects ‘The Dawn of far found (Bianchi et al. 2022). On the contrary, formamide has Organic Chemistry’ (DOC), grant agreement no. 741002 and ‘Quan- definitively a larger BE range, so that we predict formamide to tum Chemistry on Interstellar Grains’ (QUANTUMGRAIN), grant originate in a hotter region than that of acetaldehyde, which is indeed agreement no. 865657, and from the Marie Sklodowska-Curie for what is observed in the few cases where a similar analysis has been the project ‘Astro-Chemical Origins’ (ACO), grant agreement no. carried out (e.g. Csengeri et al. 2019; Okoda et al. 2021; Bianchi 811312. This work was supported by the Agence Nationale de et al. 2022). la recherche (ANR) SIRC project (grant ANR-SPV202448 2020- In summary, with the high sensitivity of the new present facilities, 2024). CC wishes to thank Eleonora Bianchi and Lorenzo Tinacci such as ALMA and NOEMA, very likely studies revealing a differ- for fruitful discussions on the chemical segregation observed in entiation in the chemical species in hot cores/corinos will become hot corinos and the evaluation of the theoretical binding energy, more an more available. We need to be prepared with studies similar respectively. This work has also been funded by the European to the one reported here, where the BEs of more complex organic Research Council (ERC) for the Starting Grant ‘DustOrigin’, hold molecules are estimated, so that we can appropriately interpret the by Prof. Ilse De Looze, University of Ghent, grant agreement ID: astronomical observations. In turn, those observations can help to 851622. AR is indebted to the Ramón y Cajal programme. validate our methods, for which so much uncertainty still persist. DATA AVA I L A B I L I T Y 5 CONCLUSIONS The data underlying this article are freely available in Zenodo We presented TPD experiments and quantum chemical computations at https://zenodo.org/communities/aco-astro-chemical-origins/?pag of the acetaldehyde BE on a pure acetaldehyde ice, and on crystalline e=1&size = 20 and amorphous water ices. The main conclusions of this work are the following: REFERENCES (i) The experiments indicate that the acetaldehyde BE has a Amiaud L., Dulieu F., Fillion J. H., Momeni A., Lamaire J. L., 2007, J. Chem. distribution of values. In the acetaldehyde ice, the BE has two main Phys., 127, 12 energy values, at about 4650 and 5250 K, with the peak around Bacmann A., Taquet V., Faure A., Kahane C., Ceccarelli C., 2012, A&A, 541, 5250 K being about 10 times more frequent. In crystalline water ice, L12 the BE ranges between about 4600 and 5700 K, with a peak around Bahr S., Toubin C., Kempter V., 2008, J. Chem. Phys., 128, 134712 MNRAS 516, 2586–2596 (2022) 2596 S. Ferrero et al. Becke A. D., 1993, J. Chem. Phys., 98, 1372 Grimme S., Ehrlich S., Goerigk L., 2011, J. Comput. Chem, 32, 1456 Bennett C. J., Jamieson C. S., Osamura Y., Kaiser R. I., 2005, ApJ, 624, 1097 Gutiérrez-Quintanilla A. et al., 2021, MNRAS, 506, 3734 Bianchi E. et al., 2019, MNRAS, 483, 1850 Hasegawa T. I., Herbst E., Leung C. M., 1992, ApJS, 82, 167 Bianchi E., López-Sepulcre A., Ceccarelli C., Codella C., Podio L., Bouvier Hourahine B. et al., 2020, J. Chem. Phys., 152, 124101 M., Enrique-Romero J., 2022, ApJL, 928, L3 Hudson R. L., Ferrante R. F., 2020, MNRAS, 492, 283 Biver N. et al., 2021, A&A, 648, A49 Hudson R. L., Moore M. H., 1997, Icarus, 126, 233 Blake G. A., Sutton E. C., Masson C. R., Phillips T. G., 1987, ApJ, 315, 621 Jaber A. A., Ceccarelli C., Kahane C., Caux E., 2014, ApJ, 791, 29 Boogert A. C. A., Gerakines P. A., Whittet D. C. B., 2015, ARA&A, 53, 541 Kimmel G. a., Stevenson K. P., Dohn’alek Z., Smith R. S., Kay B. D., 2001, Bushlanov P. V., Blatov V. A., Oganov A. R., 2019, Comput. Phys. Commun., J. Chem. Phys., 114, 5284 236, 1 Kresse G., Furthmüller J., 1996a, Comput. Mater. Sci., 6, 15 Caldeweyher E., Bannwarth C., Grimme S., 2017, J. Chem. Phys., 147, Kresse G., Furthmüller J., 1996b, Phys. Rev. B, 54, 11169 034112 Kresse G., Hafner J., 1993, Phys. Rev. B, 47, 558 Caldeweyher E., Ehlert S., Hansen A., Neugebauer H., Spicher S., Bannwarth Kresse G., Hafner J., 1994, Phys. Rev. B, 49, 14251 Downloaded from https://academic.oup.com/mnras/article/516/2/2586/6678576 by 81533202 user on 20 March 2023 C., Grimme S., 2019, J. Chem. Phys., 150, 154122 Kresse G., Joubert D., 1999, Phys. Rev. B, 59, 1758 Cazaux S., Tielens A. G. G. M., Ceccarelli C., Castets A., Wakelam V., Caux Lamberts T., Markmeyer M. N., Kolb F. J., Kästner J., 2019, ACS Earth Space E., Parise B., Teyssier D., 2003, ApJ, 593, L51 Chem., 3, 958 Ceccarelli C., Loinard L., Castets A., Tielens A. G. G. M., Caux E., 2000, Law C. J., Zhang Q., Öberg K. I., Galván-Madrid R., Keto E., Liu H. B., Ho A&A, 357, L9 P. T. P., 2021, ApJ, 909, 214 Chaabouni H., Diana S., Nguyen T., Dulieu F., 2018, A&A, 612, A47 Le Roy L. et al., 2015, A&A, 583, A1 Chaabouni H., Minissale M., Baouche S., Dulieu F., 2020, HAL, hal- Lee C.-F., Codella C., Li Z.-Y., Liu S.-Y., 2019, ApJ, 876, 63 02982177 Lee C., Yang W., Parr R. G., 1988, Phys. Rev. B, 37, 785 Chahine L. et al., 2022, A&A, 657, A78 Lefloch B., Ceccarelli C., Codella C., Favre C., Podio L., Vastel C., Viti S., Charnley S. B., 2004, Adv. Space Res., 33, 23 Bachiller R., 2017, MNRAS, 469, L73 Charnley S. B., Tielens A. G. G. M., Millar T. J., 1992, ApJ, 399, L71 Lyakhov A. O., Oganov A. R., Stokes H. T., Zhu Q., 2013, Comput. Phys. Chuang K. J., Fedoseev G., Scirè C., Baratta G. A., Jäger C., Henning T., Commun., 184, 1172 Linnartz H., Palumbo M. E., 2021, A&A, 650, A85 Manigand S. et al., 2020, A&A, 635, A48 Codella C. et al., 2018, A&A, 617, A10 Martı́n-Doménech R., Öberg K. I., Rajappan M., 2020, ApJ, 894, 98 Codella C. et al., 2020, A&A, 635, A17 McElroy D., Walsh C., Markwick A., Cordiner M., Smith K., Millar T., 2013, Congiu E., Chaabouni H., Laffon C., Parent P., Baouche S., Dulieu F., 2012, A&A, 550, A36 J. Chem. Phys., 137, 054713 Minissale M. et al., 2022, ACS Earth Space Chem., 6, 597 Corazzi M. A., Brucato J. R., Poggiali G., Podio L., Fedele D., Codella C., Minissale M., Dulieu F., Cazaux S., Hocuk S., 2016, A&A, 585, A24 2021, ApJ, 913, 128 Molpeceres G., Kaestner J., Herrero V., Peláez R., Maté B., 2022, A&A, 664, Crovisier J., Bockelée-Morvan D., Colom P., Biver N., Despois D., Lis D. A169 C., Teamtarget-of-opportunity radio observations of comets, 2004, A&A, Nguyen T., Baouche S., Congiu E., Diana S., Pagani L., Dulieu F., 2018, 418, 1141 A&A, 619, 111 Csengeri T., Belloche A., Bontemps S., Wyrowski F., Menten K. M., Noble J. A., Congiu E., Dulieu F., Fraser H. J., 2012a, MNRAS, 421, 768 Bouscasse L., 2019, A&A, 632, A57 Noble J. A., Theule P., Mispelaer F., Duvernay F., Danger G., Congiu E., Dartois E., Chabot M., Id Barkach T., Rothard H., Augé B., Agnihotri A. N., Dulieu F., Chiavassa T., 2012b, A&A, 543, A5 Domaracka A., Boduch P., 2019, A&A, 627, A55 Öberg K. I., van Dishoeck E. F., Linnartz H., Andersson S., 2010, ApJ, 718, De Simone M. et al., 2020, A&A, 640, A75 832 Doná L., Brandenburg J. G., Civalleri B., 2019, J. Chem. Phys., 151, 121101 Oganov A. R., Glass C. W., 2006, J. Chem. Phys., 124, 244704 Dovesi R. et al., 2018, WIREs Comput. Mol. Sci., 8, e1360 Oganov A. R., Lyakhov A. O., Valle M., 2011, Acc. Chem. Res., 44, 227 Drozdovskaya M. N., van Dishoeck E. F., Rubin M., Jørgensen J. K., Altwegg Okoda Y., Oya Y., Abe S., Komaki A., Watanabe Y., Yamamoto S., 2021, K., 2019, MNRAS, 490, 50 ApJ, 923, 168 Dulieu F., Amiaud L., Baouche S., Momeni A., Fillion J.-H., Lemaire J., Rimola A., Skouteris D., Balucani N., Ceccarelli C., Enrique-Romero J., 2005, Chem. Phys. Lett., 404, 187 Taquet V., Ugliengo P., 2018, ACS Earth Space Chem., 2, 720 Elstner M., Porezag D., Jungnickel G., Elsner J., Haugk M., Frauenheim T., Scibelli S., Shirley Y., 2020, ApJ, 891, 73 Suhai S., Seifert G., 1998, Phys. Rev. B, 58, 7260 Shimonishi T., Nakatani N., Furuya K., Hama T., 2018, ApJ, 855, 27 Enrique-Romero J. et al., 2020, MNRAS, 493, 2523 Skouteris D., Balucani N., Ceccarelli C., Vazart F., Puzzarini C., Barone V., Enrique-Romero J., Ceccarelli C., Rimola A., Skouteris D., Balucani N., Codella C., Lefloch B., 2018, ApJ, 854, 135 Ugliengo P., 2021, A&A, 655, A9 Speedy R. J., Debenedetti P. G., Smith R. S., Huang C., Kay B. D., 1996, J. Enrique-Romero J., Rimola A., Ceccarelli C., Balucani N., 2016, MNRAS, Chem. Phys., 105, 240 459, L6 Tait S. L., Dohnálek Z., Campbell C. T., Kay B. D., 2005, J. Chem. Phys., Enrique-Romero J., Rimola A., Ceccarelli C., Ugliengo P., Balucani N., 122, 164708 Skouteris D., 2019, ACS Earth Space Chem., 3, 2158 Vastel C., Ceccarelli C., Lefloch B., Bachiller R., 2014, ApJ, 795, L2 Fedoseev G., Qasim D., Chuang K.-J., Ioppolo S., Lamberts T., van Dishoeck Vasyunin A. I., Caselli P., Dulieu F., Jiménez-Serra I., 2017, ApJ, 842, 33 E. F., Linnartz H., 2022, ApJ, 924, 110 Vazart F., Ceccarelli C., Balucani N., Bianchi E., Skouteris D., 2020, MNRAS, Ferrero S., Zamirri L., Ceccarelli C., Witzel A., Rimola A., Ugliengo P., 2020, 499, 5547 ApJ, 904, 11 Wakelam V. et al., 2012, ApJS, 199, 21 Fourikis N., Sinclair M. W., Robinson B. J., Godfrey P. D., Brown R. D., Wakelam V., Loison J. C., Mereau R., Ruaud M., 2017, Mol. Astrophys., 6, 1974, Aust. J. Phys., 27, 425 22 Frisch M. J. et al., 2010, Gaussian09 , version D.01. Gaussian Inc., Yang Y.-L. et al., 2021, ApJ, 910, 20 Wallingford, CT Zhou C. et al., 2022, A&A, 658, A131 Garrod R. T., Herbst E., 2006, A&A, 457, 927 Gottlieb C. A., 1973, in Gordon M. A., Snyder L. E., eds, Molecules in the Galactic Environment. Harvard University, Cambridge, Massachusetts, p. 181 Grimme S., Antony J., Ehrlich S., Krieg H., 2010, J. Chem. Phys., 132, This paper has been typeset from a TEX/LATEX file prepared by the author. 154104 MNRAS 516, 2586–2596 (2022)