Photocleavage of Aliphatic C–C Bonds in the Interstellar Medium

Ultraviolet (UV) processing in the interstellar medium (ISM) induces the dehydrogenation of hydrocarbons. Aliphatics, including alkanes, are present in different interstellar environments, being prevalently formed in evolved stars; thus, the dehydrogenation by UV photoprocessing of alkanes plays an important role in the chemistry of the ISM, leading to the formation of unsaturated hydrocarbons and eventually to aromatics, the latter ubiquitously detected in the ISM. Here, through combined experimental results and ab initio calculations, we show that UV absorption (mainly at the Lyα emission line of hydrogen at 121.6 nm) promotes an alkane to an excited Rydberg state from where it evolves toward fragmentation, inducing the formation of olefinic C=C bonds, which are necessary precursors of aromatic hydrocarbons. We show that the photochemistry of aliphatics in the ISM does not primarily produce direct hydrogen elimination but preferential C–C photocleavage. Our results provide an efficient synthetic route for the formation of unsaturated aliphatics, including propene and dienes, and suggest that aromatics could be formed in dark clouds by a bottom-up mechanism involving molecular fragments produced by UV photoprocessing of aliphatics.


INTRODUCTION
The molecular inventory of space is comprised of more than 250 identified molecular species (McGuire 2022) whose formation pathways are very diverse, from gasphase neutral-neutral reactions in evolved stars to solidstate radiochemistry in molecular clouds (Tielens 2013).In the interstellar medium (ISM), UV-induced chemistry Corresponding author: José Ángel Martín-Gago; Gonzalo Santoro gago@icmm.csic.es,gonzalo.santoro@csic.es is particularly relevant and it is considered as responsible for the dehydrogenation of carbonaceous cosmic dust (Jenniskens et al. 1993;Jones et al. 2013Jones et al. , 2017)).In addition, it provides plausible synthetic routes for prebiotic molecules in Dense Molecular Clouds (DMCs), including aminoacids and ribose (Bernstein et al. 2002;Caro et al. 2002;Ciesla & Sandford 2012;Meinert et al. 2016;Öberg 2016), with obvious implications in the emergence of life.
Hydrocarbons are widespread in space (Tielens 2005a;Chiar et al. 2013;Hansen et al. 2022) and, among them, polycyclic aromatic hydrocarbons (PAHs) account for the capture of up to 20% of the elemental carbon in the ISM (Peeters et al. 2021).In particular, the unidentified infrared emission (UIE) bands which fall in the spectral range from 3 to 20 µm (main bands at 3.3, 6.2, 7.7, 8.6, 11.2 and 12.7 µm) have generally been assigned to polyaromatic carriers that are small enough to be stochastically heated by the absorption of a single UV photon, which constitutes the polycyclic aromatic hydrocarbon (PAH) hypothesis (Leger & Puget 1984;Allamandola et al. 1985Allamandola et al. , 1989;;Puget & Leger 1989).The UIEs features are ubiquitously detected in a wide variety of astrophysical regions, including the ISM, star forming galaxies and extragalatic environments (Tielens 2008;Monfredini et al. 2019;Li 2020;García-Bernete et al. 2021).However, the formation mechanism of aromatics is not well constrained and the energetic processing of aliphatic hydrocarbons has been suggested as the driving force for an aliphatic-aromatic transition that leads to the aromatic enrichment of the ISM (Goto et al. 2003;Matrajt et al. 2005;Tielens 2005bTielens , 2013)).
On the other hand, the UIEs are accompanied by IR emission bands at 3.4, 6.85 and 7.25 µm that are due to aliphatic hydrocarbons (Pinho & Duley 1995;Yang et al. 2013;Jensen et al. 2022;Yang & Li 2023) and other carriers different from free PAHs have been proposed for the UIEs.These are usually comprised of a mixture of aromatic and aliphatic hydrocarbons and include mixed aromatic and aliphatic organic molecules (MAONs) (Kwok & Zhang 2011;Kwok & Zhang 2013) as well as hydrogenated amorphous carbon (HAC) nanoparticles (Duley & Williams 1988;Jones et al. 1990;Jones 2012;Jones & Habart 2015;Jones & Ysard 2022).Indeed, in the diffuse ISM, the 3.4 µm absorption band along with the weaker absorption features at 6.8 µm and 7.3 µm are attributed to the aliphatic component of carbonaceous dust, which is consistent with HAC grains (Pendleton & Allamandola 2002;Dartois et al. 2004).
Aliphatic hydrocarbons including alkanes are present in different interstellar environments where they are exposed to UV radiation.For instance: long chain aliphatics have been identified in cometary dust (Keller et al. 2006;Raponi et al. 2020); n-alkanes up to heptane (C 7 H 16 ) have been unequivocally detected in-situ by the Rosetta mission in comet 67P/Churyumov-Gerasimenko (Schuhmann, M. et al. 2019); linear alkanes have also been systematically identified in presolar grains in meteorites (Glavin et al. 2018); aliphatic hydrocarbons have been recently detected in the samples of the carbonaceous asteroid (162173) Ryugu returned to Earth with a CH 2 /CH 3 ratio pointing towards longer aliphatic chains than those of meterorites (Yabuta et al. 2023); to name a few.Nevertheless, it is worth noticing that the presence of alkanes in comets does not necessarily imply their presence in the ISM due to the reprocessing of the interstellar matter in the early solar nebula.Furthermore, C 4 -C 6 saturated hydrocarbon units are suggested to constitute the aliphatic portion of carbonaceous cosmic dust, weaving the aromatic backbone (Pendleton & Allamandola 2002;Dartois et al. 2005;Pino et al. 2008;Kwok & Zhang 2011;Kwok & Zhang 2013).
Importantly, aliphatic hydrocarbons, linear alkanes included, are prevalently formed at the conditions of the circumstellar envelopes (CSEs) of carbon-rich evolved stars by the interaction of atomic carbon and H 2 (Martínez et al. 2020) and these aliphatic molecules are incorporated into the carbonaceous cosmic dust that is expelled towards the interstellar medium.Dehydrogenation of the aliphatic portion of cosmic dust in the ISM increases the C/H ratio in dust grains, which is considered to be a UV-induced process and responsible for the transition from aliphatic-rich to aromatic-rich carbonaceous cosmic dust (Pino et al. 2008;Jones et al. 2013).
Spatial mapping of the aliphatic portion of interstellar dust towards the Galactic Centre has found a high variability in the aliphatic content, ranging from 4% to 25% of the total carbon abundance depending on the observed source (Godard et al. 2012;Günay et al. 2020).Thus, aliphatic hydrocarbons can lock as much elemental carbon as PAHs.The observed variability in the aliphatic fraction can be attributed to the different evolutionary stages of the aliphatic-to-aromatic transition (Jones et al. 2017).Nonetheless, the photon-induced destruction of aliphatics and aliphatic moieties in the ISM is yet to be fully unveiled and the implications of this process on the formation of aromatics is not yet ascertained.
Here, we report that vacuum UV radiation at a photon energy of mainly 10.2 eV (Ly-α emission line of hydrogen) and at low temperatures primarily induces the photocleavage of the C-C bonds in linear alkanes along with subsequent hydrogen transfer between the photofragments; thus, dehydrogenation in the ISM does not occur preferentially through direct hydrogen elimination by the UV radiation field but should substantially proceed as an effective process after aliphatic fragments (carrying hydrogen) are incorporated to the gas phase.In addition, our results can be generalized to several different astrochemical environments, providing a plausible route for the formation of molecular precursors of aromatics in cold environments, where gas-phase chemistry is restricted to barrierless and exoergic reactions.In particular, the mechanism of alkane photofragmentation that we present leads to the formation of propene (C 3 H 6 ) and dienes, whose chemical formation pathways at low temperatures are key for under-standing the recent detection of aromatics in dark clouds (McGuire et al. 2021;Cernicharo et al. 2021).

EXPERIMENTAL METHODS AND QUANTUM MECHANICAL CALCULATIONS
All the experiments have been carried out in the INFRA-ICE module (Santoro et al. 2020a) of the Stardust machine (Martínez et al. 2020;Santoro et al. 2020b;Accolla et al. 2021;Sobrado et al. 2023) in ultra-high vacuum (UHV) conditions (base pressure at room temperature: 3 × 10 −10 mbar).
We performed two different independent irradiation experiments.The first one consisted in the deposition of linear hexane (C 6 H 14 ; Sigma-Aldrich; purity > 99%) and subsequent UV-irradiation.The second was intended to generalize the results and consisted in the deposition and subsequent UV-irradiation of linear undecane (C 11 H 24 ; Sigma-Aldrich; purity > 99%).In both cases, alkane vapours were deposited on infrared transparent KBr substrates at 14 K. Prior to introducing vapours in the chamber, alkanes were further purified by three pump-thaw cycles.The column density (number of molecules per cm 2 ), N, of the deposited alkanes was calculated from the IR spectrum using the CH 2 /CH 3 stretching modes region (2800-3000 cm −1 ) according to where τ is the optical depth and A the band strength of the overall CH 2 /CH 3 stretching modes.We used A values of 7.2 × 10 −17 cm molecule −1 for C 6 H 14 (Matrajt et al. 2005) and 1.3 × 10 −16 cm molecule −1 for C 11 H 24 (Dartois et al. 2004) which lead to column densities of (4.6 ± 0.9) × 10 16 molecules cm −2 and (4.3 ± 0.9) × 10 16 molecules cm −2 , respectively, corresponding to about 45 monolayers (1 ML ≈ 10 15 molecules cm −2 ).After deposition, solid C 6 H 14 and C 11 H 24 were irradiated by UV photons using a H 2 -flowing discharge lamp (UVS 40A2, Prevac) operating at 60 W. At the selected working conditions, the spectrum of the lamp corresponds predominantly to the Lyman-α line of atomic hydrogen at 121.6 nm (10.2 eV) with contributions from the emission of molecular hydrogen at around 160 nm (7.8 eV).Hydrogen discharge lamps favouring Lyman-α emission have been shown to satisfactorily simulate the UV field of the ISM (Jenniskens et al. 1993) and have also been used to simulate the secondary UV field in DMCs (Alata et al. 2014) and Photon-Dominated Regions (PDRs) (Alata et al. 2015).As our lamp is windowless contributions from Lyman-β and Lyman-γ lines at 102.6 nm (12.1 eV) and 97.3 nm (12.7 eV) are also present.Therefore, windowless UV discharge lamps, more closely reproduce the ISM UV field as they cover the UV emission in the 91.2-115 nm range.Windowed lamps usually employ MgF 2 windows which shows a cutoff at wavelengths below 115 nm (Chen et al. 2013).
At the selected working conditions, the photon flux integrated over the whole spectral range is 6.2 × 10 14 ph s −1 cm −2 (Santoro et al. 2020a).Total UV fluences of ca. 10 19 ph cm −2 were employed.Considering a photon field in the diffuse ISM of ∼ 8 × 10 7 ph s −1 cm −2 (Mathis et al. 1983), the employed fluence corresponds to ∼ 10 3 -10 4 years in the diffuse ISM.In the case of DMCs, the secondary UV field is estimated as 10 4 ph s −1 cm −2 (Cecchi-Pestellini & Aiello 1992); thus, the total UV fluence used in the experiments corresponds to ∼ 3 × 10 7 years, a time similar to the lifetime of molecular clouds (Chevance et al. 2019).Nevertheless, it should be noted that the photon flux in our experiments is orders of magnitude higher than that in the ISM and DMCs, what might play a role as, e.g., relaxation between photon absorption events can be impeded.
During the complete UV irradiation, transmission IR spectra were concurrently acquired each 200 s using a vacuum VERTEX 70V spectrometer (Bruker) with a liquid-nitrogen cooled mercury-cadmium-telluride (MCT) detector.The complete optical path is kept under vacuum (10 −1 mbar).The spectral resolution was set to 2 cm −1 and 128 scans were co-added for each spectrum.The ZnSe windows that are used to isolate the vacuum of the spectrometer to the UHV of the sample chamber strongly decrease the sensitivity in the spectral range below 850 cm −1 .
From the IR spectra we have calculated the effective cross-sections of photodestruction, σ des , and photoformation, σ f orm , for several molecular moieties.The effective cross-sections are derived considering first-order reaction kinetics according to the following expressions (Cottin et al. 2003;Loeffler et al. 2005;Martín-Doménech et al. 2015): (2) where τ denotes the integrated optical depth of the selected IR band, τ ss the steady state optical depth, τ 0 the initial integrated optical depth, ϕ the photon flux and t the irradiation time.Effective cross-sections encompass all the possible formation/destruction pathways and therefore do not distinguish among different chemical routes.Thermal Programmed Desorption (TPD) measurements for C 6 H 14 were performed after the UV irradiation at a heating rate of 1 K min −1 using a Lakeshore 335 temperature controller.A PrismaPlus QMG 220 M2 (Pfeiffer) mass spectrometer continuously monitored the desorbed gaseous species from m/z = 1 to m/z = 200 and a complete mass spectrum was acquired every 1.2 K. From the mass spectra, TPD curves were derived at selected m/z values.TPD measurements of an identical sample without UV exposure was also acquired for comparison purposes.
Quantum mechanical calculations were performed at different levels of theory.To investigate the softening of the C-C bonds in C 6 H 14 , we fixed the C2-C3 bond length and relaxed the remaining degrees of freedom.This was performed for the neutral ground state (S 0 ; C 6 H 14 ), the cation ground state (S 0 ; C 6 H 14 + ) and neutral first excited state (S 1 ; C 6 H 14 ).
Energy barriers for neutral and cation states have been calculated using Density Functional Theory (DFT) (Lewis et al. 2011) using the BLYP exchange-correlation functional (Lee et al. 1988) with D3 corrections (Grimme et al. 2011) and norm conserving pseudopotentials.We employed a basis set of optimized numerical atomic-like orbitals (NAOs) (Basanta et al. 2007) with a 1s orbital for H and sp 3 orbitals for C atoms.The energy for the barriers were calculated by fixing the reactions coordinates and relaxing the geometries of the molecules.
For the analysis of the excited states DFT timedependent DFT (TD-DFT) quantum-chemical calculations were performed with Gaussian16 software (Frisch et al. 2016) employing CAM-B3LYP functional and 6-31++G* basis set.The neutral and cation structures of C 6 H 14 have been relaxed in their ground state and the first hundred vertical excited states of the neutral molecule have been calculated.Full relaxation of the neutral alkane on the first excited state surface was performed.For neutral ground state (S 0 ; C 6 H 14 ), cation ground state (S 0 ; C 6 H 14 + ) and neutral first excited state (S 1 ; C 6 H 14 ), the energy scan was performed constraining the C2-C3 bond length and relaxing all the other degrees of freedom.The calculations for the energy scan have also been performed using Gaussian16 software (Frisch et al. 2016).Finally, the ionization potential is estimated as the difference between the energy at the ground state neutral and that of the cation ground state, at a fixed geometry of the neutral state.

Formation of new chemical species during UV irradiation of linear alkanes
To investigate the UV photochemistry of linear alkanes at low temperature and at the conditions of the ISM we irradiated both linear hexane (C 6 H 14 ) and undecane (C 11 H 24 ) mainly with the Lyman-α emission of hydro-gen at 10.2 eV (λ=121.6 nm). Figure 1a shows the IR spectra of solid amorphous C 6 H 14 both as-deposited and after irradiation with a UV fluence of ca. 10 19 ph cm −2 .To highlight the changes in the spectra upon irradiation, Figure 1b shows the difference spectrum.
A clear reduction in the bands associated with C 6 H 14 is observed (see Appendix A for IR band assignment) along with the emergence of new absorption features that reveal the formation of olefinic moieties, both of vinyl (-CH=CH 2 ) and trans-vinylene (-CH=CH-) character.In particular, the bands at 1644 cm −1 and 3078 cm −1 are ascribed to the C=C and CH stretching modes of olefins whereas the doublet at 995 cm −1 and 911 cm −1 and the band at 969 cm −1 are very characteristic of vinyl (-CH=CH 2 ) and trans-vinylene (-CH=CH-) moieties, respectively (Socrates 2004).The band at 1437 cm −1 , which is observed to increase upon UV irradiation, can be attributed to methylene (CH 2 ) deformation in the presence of adjacent unsaturated groups (Socrates 2004).This assignment becomes clearer when considering the irradiation of crystalline C 6 H 14 (see Appendix D).
The IR spectra also shows the formation of methane (CH 4 ) as revealed by the IR bands at 1300 cm −1 and 3006 cm −1 (Gerakines et al. 1996), which might imply the formation of CH 3 radicals upon UV exposure.However, our ab initio calculations show that CH 4 can be directly formed as consequence of C-C photocleavage (see Section 3.3 and Appendix G) indicating that the photochemistry of C 6 H 14 may not be mediated by radical species.
A list of the new absorption features after UV irradiation along with its assignment is given in Table 1.Identical results were obtained for C 11 H 24 (see Appendix E) implying that the mechanism for olefin formation is not restricted to C 6 H 14 but general to mid-and long-chain linear alkanes.
From the evolution of the IR absorption features with UV fluence (Fig. 2), we have derived the effective destruction cross-sections, σ des , of CH 2 and CH 3 aliphatic moieties for C 6 H 14 , which show values of 3.2 × 10 −19 cm 2 ph −1 and 2.1-2.5 × 10 −19 cm 2 ph −1 , respectively.The higher value observed for CH 2 destruction indicates that the cleavage of C-C bonds is more likely to occur in the molecule backbone, a result that is further confirmed by the ab initio calculations (see Section 3.3).
erential formation of C=C at the molecule backbone.
However, this result should be taken with caution since the absorption coefficients for the IR bands analysed will depend on the particular unsaturated hydrocarbon and on its chemical environment, which makes it difficult to obtain definite results.The obtained values for the cross-sections are listed in Table 2.We note that due to the absence of isolated IR bands solely ascribed to C 6 H 14 , the photolysis rate of hexane cannot be derived from the IR spectra.Only the overall effective decrease in CH 2 /CH 3 saturated aliphatic moieties can be obtained, i.e., the reduction in the number of CH 2 /CH 3 moieties as a result of the formation of olefins.

Thermal desorption after UV irradiation
The formation of olefin moieties is further confirmed by Thermal Programmed Desorption (TPD) measurements.The results for some selected m/z values characteristic of aliphatic C n H m (1 ≤ n ≤ 6; m ≤ 14) molecular species are shown in Figure 3, where the TPD measurements of non-irradiated C 6 H 14 are also shown for comparison purposes.The desorption of CH 4 at around 45 K is evident by the increase in m/z =15 and 16 (Fig. 3a).As the temperature increases C 2 H x and C 3 H x species desorb with maximum desorption temperatures of 89 K and 105 K, respectively.We also detected C 2 H x and C 3 H x unsaturated species by their characteristic signals at m/z = 26, 27 and m/z = 39, 41.We note that m/z = 41 corresponds to the most intense signal from C 3 H 6 .
At higher temperatures, clear signatures of single and double C=C bonds in C 4 H x , C 5 H x and C 6 H x species are observed.The desorption temperature of olefins occurs at lower temperatures regarding their fully saturated counterparts and exhibit electron-impact dissociation patterns with mass peaks at ∆(m/z) = -2 with Wavenumber σ f orm Note- (a) For the abbreviation of the vibrational modes the reader is referred to Table 1.s: symmetric (b) The error of the cross sections is estimated at 20%. (c) The main contribution to this value comes from vinyl moieties since the C=C stretching mode of trans alkene moieties is weak if not absent.respect to the corresponding alkane (Gross 2017).shows maxima at a temperature of 126 K (Fig. 3b), once again verifying C=C bond formation.The peaks at m/z = 82 and at m/z = 84 are unambiguously ascribed to the parent molecules C 6 H 10 and C 6 H 12 .The former confirms that dienes are formed.Fig. 4 shows a comparison of the mass spectra of nonirradiated and UV-treated C 6 H 14 at selected desorption temperatures, corresponding to the most prominent desorption temperatures of C n H m (1 ≤ n ≤ 6) hydrocarbons.The gas phase spectrum of C 6 H 14 is also provided for comparison purposes.We note that we did not observe desorbed species at m/z > 86, which indicates that polymerization from C 6 H 14 fragments does not occur or at least is only a residual process.However, as our results are related to irradiated, isolated alkanes caution has to be taken to directly extrapolate them to the postirradiation behaviour of alkane moieties in hydrogenated amorphous carbon grains.

Mechanism of C-C photocleavage
In order to gain insight into the C-C photocleavage mechanism we performed quantum-mechanical calculations.The calculated ionization energy of C 6 H 14 is 10.8 eV, in agreement with experimental results for the gasphase ionization of C 6 H 14 , for which ionization energies of 10.1-10.6 eV have been reported (Hoogerbrugge et al. 1989;Steenvoorden et al. 1991).Therefore, the ionization potential lies on the same energy range of the main photon energy used in the experiments (10.2 eV).To account for the possible excited states of C 6 H 14 upon UV excitation we have simulated the absorption spectrum of C 6 H 14 using TD-DFT calculations (Appendix F).These show spectroscopically active excited states with an energy onset of ca. 8 eV (Morisawa et al. 2012;Mao et al. 2019) and intense absorption between 9.5-10.5 eV, where the main experimental photon energy lies.
In the case of alkanes, the low-lying excited states present a Rydberg character, displaying an electronic distribution far from the nuclei with a loosely bound excited electron (Morisawa et al. 2012;Mao et al. 2019).We verified this on C 6 H 14 by considering the lowest excited state (S 1 ) which is represented by the transition from the Highest Occupied Molecular Orbital (HOMO) to the Lowest Unoccupied Molecular Orbital (LUMO) (see Appendix F). S 1 presents an extended electron spatial distribution far from the nuclei due to the promotion of an electron from the HOMO to the LUMO orbital, having a Rydberg character.Because of this particular electronic structure, the electron can be considered as effectively detached and some characteristics of excited Rydberg states converge to those of the related cation, in particular when describing the reactivity of the excited states (Lipsky 1981).To verify this assumption, we compared the computed relaxed geometries of neutral C 6 H 14 in the ground state S 0 , cation C 6 H 14 + in the ground state S 0 and neutral C 6 H 14 in the S 1 excited state (for simplicity, in the following we will refer to them as S 0 -neutral, S 0 -cation and S 1 -neutral, respectively).
Interestingly, the relaxed geometry of S 1 -neutral shows marked similarities with that of S 0 -cation (Fig. 6a) and when comparing the relaxed geometries of S 0 -neutral to those of S 0 -cation and S 1 -neutral, we found that all C-C bonds lengthen.This indicates a softening of all C-C bonds (Table 3).In particular, the C2-C3 bond has a remarkable elongation: 1.60 Å for S 0 -cation/S 1 -neutral vs. 1.53 Å for S 0 -neutral.We also calculated the energy required to elongate the C2-C3 bond.We choose this bond as is the one exhibiting the higher elongation with respect to the neutral ground state but similar results are expected for the other C-C bonds.We found that the elongation energy is significantly lower for both S 0 -cation and S 1 -neutral with respect to S 0 -neutral, with that for S 1 -neutral even further reduced (see Fig. 5).From these calculations it is evident that both S 0 -cation and S 1 -neutral facilitate the C-C cleavage: the energy gain needed to elongate the C2-C3 distance is considerably lower than in the case of the neutral ground state.Both S 0 -cation and S 1 -neutral states behave similar at regions close to the energy minima, in accordance with the very similar relaxed geometries of both systems (Fig. 6a), showing that the softening of the C2-C3 bond upon excitation (S 1 -neutral) or removal of an electron from the HOMO valence orbital (S 0 -cation) is similar.Overall, the shown similarities in terms of structure and electronic properties between S 1 -neutral and S 0 -cation, indicate that the behaviour of the excited Rydberg states can be satisfactorily approximated by that of the cation and this is the approach adopted here.Thus, to model the fragmentation reactions occurring upon photoexcitation, the S 1 -neutral Rydberg state has been approximated by the S 0 -cation state.
Using this approach, we have calculated the reaction energy landscapes for C-C cleavage reactions of C 6 H 14 both in the S 0 -neutral and S 0 -cation states as well as for C-H breaking reactions leading to the elimination of H 2 (see Appendix G).In Figure 6c we show the results for the C2-C3 rupture and subsequent hydrogen relocation leading to the formation of C 2 H 6 and C 4 H 8 .
Both in the S 0 -neutral and S 0 -cation states there is a considerable energy barrier for the reaction but this is drastically reduced from 5.00 eV to 1.80 eV from the neutral to the cationic state.Thus, fragmentation in the S 0 -cation state becomes easier than in the S 0 -neutral state.Here, we recall that we are approximating the excited S 1 -neutral Rydberg state by the S 0 -cation state; thus, our results show that the fragmentation of C 6 H 14 is favoured when proceeding through electronically excited states.According to the energy landscape, the reaction for the formation of C 2 H 6 and C 4 H 8 in the S 0 -cation state proceeds by the elongation of the C2-C3 bond and the subsequent relocation of an H atom from C4 to C2, which induces the formation of a C=C bond between atoms C3 and C4, in agreement with the formation of vinyl groups observed by IR spectroscopy.
We have found a marked reduction in the energy barriers (by factors higher than 2) for all the calculated fragmentation reactions when considering S 0 -cation states with respect to the same reactions through S 0 -neutral states using BLYP-D3/NAO with Fireball DFT (see Appendix G).The reaction barriers are listed in Table 4.
Importantly, in the S 0 -cation states all the barrier energies for C-C bond breaking are lower than that for H 2 elimination and the reduction in energy barriers is more pronounced (both in absolute and relative values) for all the reactions involving C-C cleavage.These results might reflect the lower bond energies of aliphatic C-C bonds (∼ 3.8 eV) regarding that of aliphatic C-H bonds (∼ 4.3 eV) (Duley 2000;Jones 2012a) and prove that C-C cleavage is preferential over C-H bond breaking upon electronic excitation, in line with the experimental observation.In addition, the simulation of the energy landscape shown in Figure 17 (Appendix G) demonstrates that the radical formation through C 6 H 14 → C 6 H 13 + H (energy barrier of ∼ 4.5 eV) is less favourable than C-C bond cleavage from an excited Rydberg state.
Despite the aforementioned arguments towards a preferential C-C photocleavage and although the energy bar-  riers are considerably reduced when the reactions proceed through excited electronic states, the energy barriers are still high.Our calculations show that the geometrical relaxation from the S 0 -cation state provides 0.5 eV of thermal energy, which is still insufficient for the fragmentation of the molecule.Nevertheless, the dissociation barrier can be surpassed by the energy provided through the internal conversion from high energy excited states to low-lying Rydberg states.Part of the difference in energy from an excited state at the main photon experimental excitation (10.2 eV) to the S 1 -neutral excited state (calculated at ca. 8.1 eV) can be transferred to molecular phonon modes (Marciniak et al. 2021).In this way, the system can storage enough vibrational en-ergy to overcome the reaction barrier for fragmentation with selective C-C cleavage to multiple products (Los et al. 1991;de Koster & Beijersbergen 1995).
It is also to be noted that polymerization towards larger alkenes is not observed in our experiments (Sec.3.2).This points out to a non-radical mediated photochemistry, opposite to what has been observed for shorter alkanes (e.g., CH 4 , C 2 H 6 and C 3 H 8 ) (Carrascosa et al. 2020).In the latter case, it is well known that vacuum UV at low temperatures induces the formation of radicals by atomic hydrogen abstraction initiating a polymerization process.We did not observe this phenomenon and therefore the photochemistry of mid-to long alkanes seems different, supporting the photodissociation mechanism that we are proposing.

DISCUSSION
As we have previously demonstrated that at the conditions of the circumstellar envelopes (CSEs) of C-rich Asymptotic Giant Branch (AGB) stars mainly aliphatics are produced (Martínez et al. 2020), plausible scenarios for the formation of aromatics needs to be explored due to their presence in interstellar environments.Indeed, benzene has not been yet identified in AGBs but it has been reported in protoplanetary nebulae (PPNe) (Cernicharo et al. 2001), which suggest a UV-driven transition from aliphatics to aromatics in an evolutionary context.
On the other hand, the carbonaceous cosmic dust formed in AGBs consist in amorphous hydrocarbon particles comprised of sp 2 and sp 3 hybridizations (Andersen et al. 2003).The evolution from its formation in AGBs towards the PPNe and subsequent PNe phases increases the aromatic content to the detriment of the aliphatic portion (Joblin et al. 1996;Goto et al. 2007), which is consistent with the thermal annealing of the grains by the stochastic heating of energetic photons (Goto et al. 2000).
In the case of the diffuse ISM, dehydrogenation of the aliphatic portion of carbonaceous dust grains by the local interstellar UV radiation field is considered as the driving force towards aromatization (Jones 2012a), a process that will also occur in Photon-Dominated Regions (PDR) on estimated timescales of ∼ 10 3 yr or even lower (Jones 2012b).UV-induced dehydrogenation in the ISM has been proposed to proceed through direct photodissociation of C-H bonds in aliphatic molecular species and in hydrogenated carbon grains (Muñoz Caro et al. 2001;Mennella et al. 2001;Dartois et al. 2005).However, our results point towards a different scenario as we have shown that C-C photocleavage is favoured over C-H bond breaking.Therefore, hydrogen deple-tion in carbonaceous dust grains is primarily a result of the photolysis of aliphatic C-C bonds which produces small molecular fragments that might subsequently desorb through non-thermal processes (Fredon et al. 2021;Dartois et al. 2022;Del Fré et al. 2023) carrying hydrogen towards the gas phase and provoking an effective dehydrogenation of the carbon grains.
It is also likely that the cleavage of the aliphatic C-C bonds of carbonaceous grains induces a structural rearrangement of the bond network towards an olefinic-rich material (Smith 1984;Jones 2012a).According to our calculations the bond cleavage proceeds first by an elongation of the C-C bond and a subsequent H relocation.For aliphatic C-C bonds in a three dimensional carbon network such as HAC, the dangling C bond formed in the first step can interact with the network before H relocation occurs.This can liberate hydrogen from the carbonaceous material during the formation of the new bonding structure with the three dimensional network enabling the dissipation of any energy excess.This process can efficiently dehydrogenate the carbonaceous dust if, as suggested by our results, C-C cleavage is more prone than C-H bond breaking.Nevertheless, caution needs to be taken to directly extrapolate our results to complex carbon structures.
The mechanism of C-C photocleavage that we describe here agrees with previous experimental observations on the UV processing of carbonaceous interstellar dust analogues in which in addition to H 2 production, the formation of alipahtic hydrocarbons (including olefins) up to four carbon atoms has been reported (Muñoz Caro et al. 2001;Alata et al. 2015).The observation of C=C bonds of vinyl and trans-vinylene nature has also been experimentally observed during the UV irradiation of several aliphatic hydrocarbons at conditions of the diffuse ISM (Dartois et al. 2005).Nevertheless, a thorough description of the UV-induced chemistry has not been provided despite its importance for modelling the chemical evolution of hydrocarbons and carbonaceous dust grains in the ISM.
On the other hand, despite the fact that our results are general and not restricted to Dense Molecular Clouds (DMCs), they might contribute to explain the rich chemistry that has been recently identified in dark clouds.Small sized polycyclic aromatic hydrocarbons (PAHs) have been firmly identified in these environments (McGuire et al. 2021;Cernicharo et al. 2021;Burkhardt et al. 2021b), especifically in the Taurus Molecular Cloud (TMC-1), raising the question on how PAHs can be formed in these cold environments.
Chemical models have been able to satisfactorily reproduce the observed abundances of many of the more than 40 pure hydrocarbon species identified in TMC-1 but they have failed in explaining the abundances of cyclic molecules, systematically leading to lower values than those observed (Burkhardt et al. 2021b;McGuire et al. 2021;McCarthy et al. 2021).At present it is still not clear if aromatics are formed through top-down or bottom-up processes.Top-down approaches have been suggested to be responsible for the large abundance of aromatics in TMC-1 (Burkhardt et al. 2021a),which might be inherited from a previous diffuse phase.On the other hand, the recent discovery of 1-cyano-1,3butadiene (C 4 H 5 CN) (Cooke et al. 2023) and the spatial distribution of C 6 H 5 CN (Cernicharo et al. 2023) have been used to argue that the formation of aromatics proceeds through bottom-up chemical routes.
The most likely precursor for benzene formation in dark clouds from a bottom-up process is 1,3-butadiene (C 4 H 6 ), which is known to lead to benzene (C 6 H 6 ) through a barrierless and exoergic reaction with C 2 H (Jones et al. 2011).C 4 H 6 has no permanent dipole moment and is therefore invisible at radiowavelengths, but the identification of C 4 H 5 CN supports the presence of 1,3-butadiene in this environment (Morales et al. 2011).Thus, C 4 H 6 might be a key species in the chemistry of TMC-1 and bottom-up chemical routes for the formation of aromatics in cold interstellar environments are dependent on efficient synthetic pathways for C 4 H 6 .
A plausible formation of C 4 H 6 involves the reaction of propene (C 3 H 6 ) with the methylidyne (CH) radical (Daugey et al. 2005;Smith et al. 2006;Loison & Bergeat 2009).C 3 H 6 was detected more than 15 years ago in TMC-1 with fairly large abundance (Marcelino et al. 2007).However, to date no efficient reaction pathways towards C 3 H 6 at low temperatures have been identified, suggesting that its formation is not driven by gas-phase chemistry (Lin et al. 2013).
In the evolution of cosmic dust grains in DMCs, a carbonaceous mantle is considered to be formed around dust grains by the accretion of gas-phase carbon atoms.Indeed, the C-C photocleavage process that we describe in detail here can operate in the photon dominated regions (PDRs) of molecular clouds (Pety et al. 2005;Alata et al. 2014;Jones et al. 2017) and our results show that UV-induced photocleavage of mid-to long alkanes or aliphatic moieties of carbonaceous dust grains lead to C 3 H 6 .Likewise, it provides a pathway for the formation of dienes in these environments by the UV photoprocessing of carbonaceous mantles that accrete on dust particles in the early stages of DMCs evolution (Jones et al. 2013;Murga et al. 2023).These newly formed mantles consist of aliphatic-rich material (Ysard et al. 2015;Jones 2016;Murga et al. 2023) and we speculate that if C 3 H 6 and small dienes are photoformed on the surface of grains, they can desorb through non-thermal processes (Fredon et al. 2021;Dartois et al. 2022;Del Fré et al. 2023) to be incorporated into the gas-phase.The UVinduced formation of C 3 H 6 and dienes might be therefore essential to understand the synthesis of small PAHs in dark clouds.
Finally, the mechanism that we have presented is also efficient in the formation of vinyl moieties, with important implications on the chemical formation routes of vinyl-bearing molecules in the interstellar medium.Vinyl containing Complex Organic Molecules (COMs) constitute an important, prevalent molecular class among the molecules detected in the ISM and DMCs, many of which have been detected in the last few years (Gardner & Winnewisser 1975;Hollis et al. 2004;Agúndez et al. 2021;Cernicharo et al. 2021;Lee et al. 2021;Rivilla et al. 2022;Molpeceres & Rivilla 2022).

CONCLUSIONS
Our results explain in detail the photo-fragmentation mechanism of alkanes and alkane moieties in the ISM and show that C-C bond photocleavage is more likely than C-H bond breaking.The mechanism that we present here indicates that the dehydrogenation of dust towards an aromatic enrichment might proceed to a large extent as an effective process in which the fragments formed by the UV-photoprocessing of aliphaticrich dust grains carry hydrogen towards the gas-phase after non-thermal desorption processes.This is particularly relevant to understand and more precisely modelling the evolution of cosmic dust among the different astrophysical environments.Furthermore, we have observed an efficient formation of olefins, including propene and dienes.These have been suggested as plausible precursors for the gas-phase synthesis of aromatics in dark clouds, where gas-phase reactions are very much restricted.To explain the observed abundances of aromatics in these environments, propene and small dienes need to be incorporated in the very first steps of the chemical evolution models but, despite being detected, its formation mechanism is unknown which currently constitutes the main bottleneck for establishing consistent chemical routes towards the formation of aromatics in dark clouds.Although speculative, our findings might contribute to explain the high abundance of propene observed in TMC-1 and, by extension, of aromatics.Autónoma de Madrid and co-financed by European Structural Funds is also acknowledged.

APPENDIX
A. IR BAND ASSIGNMENT Table 5 lists the most prominent absorption features of amorphous C 6 H 14 , crystalline C 6 H 14 and C 11 H 24 at low temperature along with its vibrational assignment.
Note- (a) The vibrational modes are abbreviated as follows: ν: stretching; δ: deformation (b: bend; sc: scissor); γ: wagging; ρ: rocking; s: symmetric; as: asymmetric; ip: in-plane; oop: out-of-plane.To derive the effective cross-sections of photodestruction, σ des , and photoformation, σ f orm , for several molecular moieties, we performed a quantitative analysis of the IR spectra acquired during UV irradiation by fitting selected IR absorption features assuming Gaussian profiles.For the fitting of the CH 2 /CH 3 stretching mode region (2800-300 cm −1 ) we used eight Gaussian curves according to Snyder et al. (1978) and Jordanov et al. (2003) to account for the complex structure due to Fermi resonances.During the fitting process, peak positions were allowed to vary by ±1 cm −1 , whereas peak full widths at half maximum (FWHM) were restricted to values lower than 30 cm −1 , except for the case of the ν s,F CH 2 mode at 2898 cm −1 that was restricted to values lower than 50 cm −1 .The increase in FWHM of this last peak upon UV irradiation is the reason for relaxing the fitting constrain.This increase is related to the new molecules that are formed, whose spectral response changes the Fermi resonances of the CH 2 bending overtones with the fundamental CH 2 stretching modes.Figure 7 shows the fitting results of the IR spectra for as-deposited C 6 H 14 and after the complete UV treatment.The morphology of solid C 6 H 14 films depends on the substrate temperature during deposition.Whilst at 14 K an amorphous solid is obtained, a substrate temperature of 80 K produces a crystalline one, as revealed by the IR spectra (Fig. 9) Figure 9. IR spectra of crystalline (red; deposition temperature: 80 K) and amorphous (black; deposition temperature: 14 K) solid C6H14.The spectra are shifted vertically for clarity.
The UV irradiation of crystalline C 6 H 14 induces the same photochemistry as in the case of amorphous C 6 H 14 , i.e., the formation of olefinic moieties both of vinyl (-CH=CH 2 ) and trans-vinylene (-CH=CH-) nature along with the formation of CH 4 due to the photocleavage of the alkane C-C bonds, which, during the first stages of the UV irradiation, disrupts the crystalline morphology (Fig. 10).
On the other hand, as stated in Section 3.1, the relationship of the band at 1437 cm −1 with the formation of C=C bonds becomes more evident in the case of crystalline C 6 H 14 , since no overlapping of the CH 2 /CH 3 deformation modes at about 1480-1440 cm −1 occurs in the first IR spectra during UV exposure.This supports our assignment of this band to the deformation of methylene groups in the presence of adjacent unsaturated groups (Table 1).As mentioned in Section 3, the mechanism of formation of olefins through UV irradiation is not restricted to C 6 H 14 but general to mid-and long-chain linear alkanes.
To illustrate this, Figure 11 shows the IR spectra of amorphous undecane (C 11 H 24 ) at different irradiation fluences.The formation of -CH=CH-and -CH=CH 2 moieties and CH 4 is evident from the spectra.The band assignment of the new IR bands corresponds to those listed in Section 3.1 for C 6 H 14 (Table 1).Figure 12a shows the isosurface of the HOMO and LUMO molecular orbitals of C 6 H 14 .It can be shown that the LUMO is highly delocalized which is consistent with the Rydberg character of the orbital.In addition, in Table 6 we have listed the vertical excitation energy and weight of the HOMO → LUMO transition which is the dominant contribution to the S 1 excited state.The same excitation with Rydberg character also dominates the first triplet excited state (T1) at 8.01 eV.
We have also simulated the absorption spectrum of neutral C 6 H 14 from the calculations of the transitions from 8 eV to 11.5 eV.The results are shown in Figure 12b.For scanning the conformational space of the reactions we have chosen the BLYP functional given its capability for studying a larger conformational space.To validate its use in Figure 13 we show a comparison of the energy profiles around the minimum energy structure for the S 0 -neutral and S 0 -cation states using different DFT approximations, namely CAM-B3LYP/6-31++G* using Gaussian 16 and BLYP-D3/NAO using Fireball.The agreement is very good validating our selection of the BLYP functiontal.

Figure 1 .
Figure 1.a) IR spectra of C6H14 at 14 K both as-deposited and after a UV fluence of 9.8 × 10 18 ph cm −2 .The position of the most prominent absorption bands is indicated in the figure.b) Difference spectrum between the UV-treated and the as-deposited spectra.

Figure 2 .
Figure 2. Evolution of the integrated optical depth of selected IR absorption features with UV fluence.Solid lines correspond to the fitting of the experimental data to Eqs. 2 and 3

Figure 3 .
Figure 3. TPD results after the UV irradiation of C6H14 with a fluence of 9.8 × 10 18 ph cm −2 in the temperature ranges a) 30-120 K and b) 100-160 K.The results from nonirradiated C6H14 are also provided for comparison.

Figure 4 .
Figure 4. Mass spectra during the desorption of nonirradiated and UV-treated solid C6H14.The desorption temperature of each spectrum is indicated.The gas phase C6H14 spectrum is shown for comparison purposes.Peaks labelled with * at m/z = 18, 28 and 32 corresponds to residual H2O, CO and O2 gases in the chamber, respectively.

Figure 6 .
Figure 6.a) Comparison of the relaxed geometries of C6H14 in the neutral ground state S0 (ball and stick), the first excited state S1 (green) and the cation C6H14 + in the ground state S0 (red).b) Comparison of the relaxed geometries of ground state S0 of C6H14 (ball and stick) and ground state S0 of C6H14 + (red) highlighting the elongation of the C2-C3 and C3-C4 bonds.The distances are given in Å. c) Energy landscapes for the reaction C6H14 → C2H6 + C4H8 for neutral C6H14 and cation C6H14 + molecules, both in the ground state S0.The reaction coordinates are depicted in the bottom of the figure.

Figure 7 .
Figure 7. Spectral fitting of as-deposited C6H14 (top) and after a UV fluence of 9.8 × 10 18 ph cm −2 (bottom).Experimental spectra: black; Gaussian profiles: red; Fitting result: green.The experimental spectra and the fitting results are shifted vertically for clarity.

Figure
Figure IR spectra of amorphous C6H14 at different UV fluences.The new IR bands upon UV exposure are indicated as well as the corresponding UV fluence of each spectrum.The spectra are vertically shifted for clarity.

Figure
Figure IR spectra of crystalline C6H14 at different UV fluences.The new IR bands upon UV exposure are indicated as well as the corresponding UV fluence of each spectrum.The spectra are shifted vertically for clarity.

Figure
Figure IR spectra of amorphous C11H24 at different UV fluences.The new IR bands upon UV exposure are indicated as well as the corresponding UV fluence of each spectrum.The spectra are shifted vertically for clarity.

Figure 12 .
Figure 12. a) Isosurface of the molecular orbitals involved in the transition of the first excited state S1.b) Simulated absorption spectra of neutral C6H14 vertical excitation at the relaxed ground state structure (CAM-B3LYP/6-31++G*): transitions (bars) fitted with a Lorentzian lineshape (FWHM = 0.05eV; solid line).

Figure 13 .
Figure 13.Energy profiles for C2-C3 distance around the minimum energy structure for S0 and cation states with different DFT approximations

Figure 17 .
Figure 17.Energy landscapes for the photocleavage reaction C6H14 → C6H12 + H2.The reactions coordinates are the distance between the H and the C in C2 and the distance and between the H and the C in C3.

Table 1 .
New IR bands upon UV irradiation of C6H14 along with band assignment.

Table 2 .
Photodestruction and photoproduction effective cross-sections of C6H14

Table 3 .
C-C bond distances of C6H14 in the neutral S0, cation S0 and neutral S1 states.

Table 4 .
Energy barriers for the fragmentation of C6H14 for the formation of different products.