Carbon Atom Reactivity with Amorphous SolidWater: H2O Catalyzed Formation of H2CO

German Molpeceres,∗,† Johannes Kastner,† Gleb Fedoseev,‡,¶ Danna Qasim,‡,§

Richard Schomig,† Harold Linnartz,‡ and Thanja Lamberts∗,‖,‡

†Institute for Theoretical Chemistry, University of Stuttgart, 70569, Stuttgart, Germany‡Laboratory for Astrophysics, Leiden Observatory, Leiden University, PO Box 9513, 2300

RA Leiden, The Netherlands¶Research Laboratory for Astrochemistry, Ural Federal University, Kuibysheva St. 48,

620026 Ekaterinburg, Russia§Current address: Astrochemistry Laboratory, NASA Goddard Space Flight Center,

Greenbelt, MD 20771, USA‖Leiden Institute of Chemistry, Gorlaeus Laboratories, Leiden University, PO Box 9502,

2300 RA Leiden, The Netherlands

E-mail: molpeceres@theochem.uni-stuttgart.de; a.l.m.lamberts@lic.leidenuniv.nl

Abstract

We report new computational and experimentalevidence of an efficient and astrochemically rel-evant formation route to formaldehyde (H2CO).This simplest carbonylic compound is centralto the formation of complex organics in cold in-terstellar clouds, and is generally regarded tobe formed by the hydrogenation of solid-statecarbon monoxide. We demonstrate H2CO for-mation via the reaction of carbon atoms withamorphous solid water. Crucial to our proposedmechanism is a concerted proton transfer cat-alyzed by the water hydrogen bonding network.Consequently, the reactions 3C + H2O −−→3HCOH and 1HCOH −−→ 1H2CO can takeplace with low or without barriers, contraryto the high-barrier traditional internal hydro-gen migration. These low barriers or absencethereof explain the very small kinetic isotopeeffect in our experiments when comparing theformation of H2CO to D2CO. Our results rec-oncile the disagreement found in the literatureon the reaction route: C + H2O −−→ H2CO.

TOC graphic

In molecular clouds where stars are born,temperatures can be as low as 10 K, and wateris the main component of the ice mantles coat-ing micron-sized dust grains.1 On the surface ofthese grains, a rich chemistry accounts for muchof the chemical complexity of the known inter-stellar complex organic molecules (COMs).2–5

At the core of COM synthesis in space is theformation of bonds to carbon atoms, which,in turn, depends upon the main reservoir ofcarbon. In the translucent stage of a molec-ular cloud or under influence of cosmic-ray ir-radiation, carbon is predominantly present inits atomic form C(3P0).

6–12 The interaction

1

arX

iv:2

110.

1588

7v1

[as

tro-

ph.G

A]

29

Oct

202

1

of atomic carbon with water has been exten-sively studied, both experimentally and the-oretically.13–20 The particularly puzzling con-trast between the gas phase and the condensedphase is illustrated by the reduction of the C−Odistance of 1.89 A in the gas-phase C−H2Ocomplex,15 to 1.50 A on average in water clus-ters.19,20 In these clusters, the C−O distancedepends on the local geometry of the adsorp-tion site, the formation of a 3C−OH2 complex19

and/or the formation of a [COH– +H3O+] com-

plex.20

Formaldehyde (H2CO) is the lowest energyisomer of C-H2O and is a key species in as-trochemical reaction networks as a precursor ofCOMs.21,22 At present, chemical models con-sider the sequential hydrogenation of solid COas its main surface formation route.23,24 Thislinks H2CO formation to later stages of a molec-ular cloud, where CO ice is typically abun-dant.1,25 The formation of H2CO may proceedalso in the gas phase.26

The 3C + H2O −−→ products reaction in thegas phase proceeds through quantum tunneling,indicated by a lower reactivity with D2O.16 Inmatrix-isolation studies, however, no productsfor the same reaction have been detected.13,15

Furthermore, C atoms generated in an arc dis-charge react with water at 77 K.14 A subsequentGC/MS analysis showed a variety of aldoses,and, among them, formaldehyde. Very recentexperimental work indicates the formation ofH2CO in the solid state on water ice.27

In this letter we provide a detailed expla-nation of the formation of formaldehyde fromthe reaction of carbon atoms with amorphoussolid water within a theoretical framework, sup-ported by tailored experiments. Our work sim-ulates the earliest stages in the star forma-tion process, before the catastrophic CO freeze-out stage,28 i.e., before the CO hydrogena-tion chain dominates H2CO formation. Ourexperiments probe the kinetic isotope effect,comparing the products of the reactions C +H2O/HDO/D2O. We resolve the apparent dis-crepancy between earlier studies by introduc-ing our proposed mechanism in which watermolecules collectively catalyze the reaction viaa proton transfer. We show that this transfer

operates throughout the entire 3C + H2O −−→H2CO reaction sequence. This implies thatformaldehyde ice abundances could be higherthan previously anticipated, especially in theearly stages of star formation, which is soonexpected to be probed by James Webb SpaceTelescope (JWST) observations of interstellarices.1,29

We modeled the formation of H2CO on amor-phous water ice clusters considering two dis-tinct mechanisms: 1. Traditional internal iso-merization, and 2. Concerted water-assistedisomerization, and compare this to the internalisomerization in the gas phase. Below, we listthe relevant reaction steps for which we wish topoint out that in both mechanisms all speciesare at all times adsorbed on a surface. Thus,mechanism 1 may resemble the process in thegas phase, but with adsorbed reactants. Wepresent transition states for both mechanismsand the analogs in the gas phase in Fig. 1 toguide the reader through the discussion. Forboth mechanisms the reaction sequence startswith:

3C + H2O −−→ 3C−OH2 (1)3C−OH2 −−→ 3HCOH . (2)

Reaction (1) represents the complex forma-tion and reaction (2) an insertion or isomeriza-tion via internal hydrogen migration.

Subsequently, also for both mechanisms, anintersystem crossing (ISC) to the singlet sur-face needs to take place followed by a hydrogenmigration to finally yield formaldehyde, eitherthrough

3HCOH −−→ 3H2CO (3)3H2CO −−→ 1H2CO (4)

or

3HCOH −−→ 1HCOH (5)1HCOH −−→ 1H2CO . (6)

Reactions (2), (3) and (6) proceed with high ac-tivation barriers in the gas phase.16,17,30 We test

2

Figure 1: Transition state geometries for reaction (2) in the upper row, and reaction (6), in the lowerrow, in the gas phase (left column) and for both the traditional internal isomerization mechanism1 (middle column) and the concerted water-assisted mechanism 2 (right column). Atoms directlyinvolved in the reaction mechanism are depicted in full color, water molecules not participating inthe reaction are transparent in the background. Color code: Teal – Carbon, Red – Oxygen, White– Hydrogen

Figure 2: Schematic reaction profiles for reactions (2) and (6) within the internal isomerizationmechanism (Mechanism 1, red) and the water assisted mechanism (Mechanism 2, green). Notethat green dashed lines here represent paths with a small barrier and barrierless paths. The minimaare taken as the average binding energies listed in Table 1 and the saddle points as the highestactivation energies in Table 2. For 1H2CO, the binding energy is taken as the average of theendpoints of intrinsic reaction coordinate calculations. The reader is referred to the text for anextensive discussion of the energetics of the reaction.

3

here to what extent this holds in mechanism 1.Mechanism 2 follows the same reaction steps,but a surrounding ice water molecule acts as aproton donor (e.g., δ+) for reactions (2) and (6)(depicted by the arrows in Fig. 1), greatly low-ering the reaction barriers, see also Fig. 2. Notethat reaction (3) does not proceed in mecha-nism 2 and is not included in either Fig. 1 orFig. 2.

Reactions are computationally studied indi-vidually by placing the respective reactants on awater ice cluster, (H2O)14. We randomly placedthe adsorbates: 3C , 3HCOH and 1HCOH ondifferent pre-optimized (H2O)14 clusters, con-structed via molecular dynamics heating fol-lowed by a geometry optimization using ab-initio methods and determining 48, 42 and44 binding sites, respectively. The ranges ofbinding energies along with their average valuefor each adsorbate are gathered in Table 1.From these binding sites both mechanism 1 andmechanism 2 were explored, in search of tran-sition states for reactions (2), (3) and (6). Weutilized a single shallow binding site for mecha-nism 1, whereas we sampled three binding sitesfor mechanism 2, because the latter is moredependent on the binding site geometry. Thelevel of theory used is MPWB1K/def2-TZVPon optimized structures at the MPWB1K/def2-SVP level, and benchmarked against gas-phasevalues using the UCCSD(T)-F12a/cc-pVTZ-F12//MPWB1K/def2-TZVP level of theory(also included in Table 2). This yields verygood results within the variability provided bythe different binding sites. Note that our valuesin the triplet channel are also in good agree-ment with the results of Hickson et al. 16 and Liet al. 17

To describe the striking difference betweenboth mechanisms, that ultimately renders acompletely different physicochemical picture,we first focus on the reaction sequence in mech-anism 1 followed by mechanism 2. Figure 2gives a schematic representation of the energyprofile of both mechanisms. Table 1 and 2 buildon Fig. 2, providing detailed values of the bind-ing energies and activation barriers. All bindingenergies are computed from the difference be-tween the energy of the adsorbate+cluster mi-

nus the sum of the separate components whileactivation energies are obtained relative to thereactant on the surface obtained from intrin-sic reaction coordinate (IRC) calculations. Allbinding energies (Table 1) and activation ener-gies (Table 2) are reported including zero-pointenergy (ZPE). More details on the protocol fol-lowed for the ab-initio calculations can be foundin the Supporting Information.Mechanism 1. The onset of the reaction is the

binding of a carbon atom atop amorphous solidwater (ASW). Three regimes can be clearly dis-tinguished, see also Table 1. The first regime,3C + (H2O)14 −−→ 3C− (H2O)14 is the one rel-evant for mechanism 1, since there is no spon-taneous chemical conversion after interactingwith H2O ice.19,20 Within this regime we ex-pect other reactions with carbon to be possible,e.g, to form CH4 after hydrogenation.32 In or-der to form 3HCOH via internal isomerization,the hydrogen atom of C−OH2 that migratesshould not take part in any hydrogen bondswith the backbone of the cluster, see middletop panel in Fig. 1. Such a site (Ubind = 72.8kJ/mol) rendered an activation energy of ∆Ua=73.3 kJ/mol. After formation of 3HCOH, chem-ical conversion to 3H2CO is studied through atransition state found for a weak binding site(Ubind = 10.3 kJ/mol) with a resulting barrierof 123.9 kJ/mol. Both activation energies areeither higher than or very similar to the cor-responding gas-phase reaction,17 see Table 2.Such barriers are too high to be overcome at10 K, even on interstellar timescales.

The formation of formaldehyde would requireintersystem crossing (ISC) from 3HCOH to1HCOH via reaction (5). The singlet state ismore stable than the triplet one by 88.9 kJ/mol.The explicit simulation of ISC is beyond thescope of the present work, but studies in thegas phase showed that the spin-orbit couplingis strong,33 which determines the intersystemcrossing rate. We expect that the ‘heavy’ Oatoms in the ice further enhance spin-orbit cou-pling. Experimental ISC timescales for reac-tion (4) report fast conversion (ns to µs)34,35

and we expect the same for reaction (5).Finally, a relatively shallow binding site

(Ubind = 22 kJ/mol, with the OH moiety not

4

Table 1: Binding energy ranges and averages for the different species and complexes on waterclusters considered in this work. Average binding energies are obtained as the mean of the individualbinding energies in each binding site. Please note that the binding energies provided here apply toboth mechanisms 1 and 2.

Adsorbate - cluster system Binding energy range Average binding energy(kJ/mol) (kJ/mol)

−−→ 3C− (H2O)14i 60 - 133 96

3C + (H2O)14ii −−→ 3[COH–H3O

+]− (H2O)13 88 - 143 116−−→ 3HCOH− (H2O)13 231 - 268 244

3HCOH + (H2O)14 −−→ 3HCOH− (H2O)14 7.5 - 52.5 29.41HCOH + (H2O)14 −−→ 1HCOH− (H2O)14 7.8 - 95.6 52.4ia) The values provided in the literature for 3C on ASW (average binding) are 116 kJ/mol and 79 kJ/mol.19,20

b) 3C adsorption on ice Ih reported binding energies are 153 and 127 kJ/mol depending on the binding site.31

iiAccording to our sampling: 71% 3C− (H2O)14; 19%3[COH–H3O

+]− (H2O)13; and 10% 3HCOH− (H2O)13

Table 2: Activation energies without (∆Ea) and with ZPE (∆Ua) in kJ/mol for reactions (2),(3) and (6) in the gas phase, via the internal isomerisation mechanism (1) and the water-assistedisomerisation mechanism (2) at the MPWB1K/def2-TZVP//MPWB1K/def2-SVP level of theory.In parentheses UCCSD(T)-F12a/cc-pVTZ-F12//MPWB1K/def2-TZVP reference values.

Mechanism Reaction Binding site i ∆Ea ∆Ua

Gas phasereaction (2) – 69.7 (72.5) 57.3 (60.1)reaction (3) – 139.6 (139.7) 126.6 (126.6)reaction (6) – 151.3 (144.4) 134.9 (127.9)

Mechanism 1reaction (2) shallow 86.9 73.3reaction (3) shallow 138.3 123.9reaction (6) shallow 163.8 146.6

Mechanism 2ii

reaction (2)medium / deep 30.3–36.9 9.7–11.5[COH– + H3O

+] 17.5 -0.31reaction (6) medium / 2 deep 12.8–31.7 -1.0–12.0

iSee textiiReaction (3) does not proceed in Mechanism 2.

interacting with the ice) gives an activationenergy of ∆Ua=146.6 kJ/mol for reaction (6),1HCOH −−→ 1H2CO, depicted in the middlebottom panel in Fig. 1. The barrier is signif-icantly higher than in the gas phase. We at-tribute this to a stabilizing effect of the surfaceon the reactant state, again rendering this path-way unlikely to be relevant at low temperatures.

In short, as a result of the high barriers in-volved, also on the surface, the traditional in-ternal isomerization mechanism 1 is unlikely tolead to formaldehyde on water ices.Mechanism 2. During the adsorption of 3C on

ASW, not only the 3C−H2O structure is found,but also two other outcomes are observed: a

[COH–+H3O+] complex20 and a direct and bar-

rierless to 3HCOH, also hinted at by Shimonishiet al. 19 Closer inspection of our optimizationtrajectories shows that the migrating H is trans-ferred from a polarized water molecule close tothe 3C−OH2 adduct, resembling an acid-baseprocess. Whether or not this chemical conver-sion takes place thus greatly depends on thelocal geometry of the water into which the car-bon inserts. The Supporting Information de-tails that the barrierless pathway remains bar-rierless, also for other exchange and correlationfunctionals.

In addition to this barrierless pathway, con-certed transition states leading to 3HCOH for-

5

mation are explored. We depict one in the topright panel in Fig. 1. We found no transitionstate for the weakly bound C atom, furthersuggesting that reactions involving free carbonhappen in these binding sites. For the stronglyand medium bound atoms, the activation ener-gies are ∆Ua=9.7 and 11.5 kJ/mol, respectively,a factor ∼ 7 lower than for the mechanism 1,see also Fig. 2.

Finally, the transition state found for the[COH– + H3O

+] complex leads to a negativebarrier (i.e., barrierless) upon the inclusion ofthe ZPE correction. In general, including ZPEcorrections reduces the barrier more for thewater-assisted mechanism, see Table 2, whichcan be rationalized by the imaginary transitionmode involving collective high-frequency waterstretching modes.

Since 3HCOH can thus readily be formed,we also examined the conversion to 3H2COthrough the water-assisted mechanism. How-ever, all three binding sites lead to endother-mic reaction paths and thus reaction (3) and,therefore, also reaction (4) are unlikely to takeplace. For the intersystem crossing reaction (5),3HCOH −−→ 1HCOH, the same reasoning asmentioned above holds.

Finally, transition states for reaction (6) arefound for three different binding sites. Two ofthem lead to activation energies of ∆Ua=9.0and 12.0 kJ/mol, in agreement with a previ-ous work.36 For the third binding site, the reac-tion is barrierless upon inclusion of ZPE. This isclearly depicted also in the bottom right panelof Fig. 1: the water ice acts as a catalyst for thereaction transferring a proton from the surfacehydrogen bond network to the CH moiety.

Summarizing, the 1HCOH −−→ 1H2CO re-action sequence can take place via the water-assisted mechanism. For some binding sites, wefind no barrier for the reaction. Combining thisfinding with the results for reaction (1) and re-action (2), opens the possibility of an effectivelybarrierless reaction pathway for the formationof H2CO from the C + H2O reaction assumingthat intersystem crossing is fast, see Fig. 2.Deuteration and Experiments. An overall

low-barrier or barrierless pathway implies that,on ice, a small isotope effect is expected for

Table 3: Activation energies (∆UA in kJ/mol)for water assisted reactions (2) and (6) andthree deuteration cases, see text, compared toCase 0 without deuteration.

Reaction3C−OH2 −−→ 3HCOH 1HCOH −−→ 1H2COCase ∆UA Case ∆UA

Highest Activation EnergyCase 0 11.5 Case 0 12.0Case 1 12.3 Case 1 12.2Case 2 18.3 Case 2 17.4Case 3 17.8 Case 3 17.9

Lowest Activation EnergyCase 0 -0.3 Case 0 -1.0Case 1 3.4 Case 1 2.3Case 2 5.6 Case 2 3.7Case 3 5.4 Case 3 3.9

reactions in which the hydrogen atoms are re-placed by deuterium. This is in stark contrastwith results obtained in the gas phase findinga significant kinetic isotope effect.16 Further-more, the so-called deuterium fractionation ofmolecules is often used as a probe to understandwhether formation has taken place in early orlate molecular cloud stages in the ISM.37 Thebarrierless nature of the reaction could have aconsequential effect on formaldehyde deuteriumfractionation in interstellar clouds.

The influence of deuterium substitution onthe activation energies is evaluated by recom-puting barrier heights for three different deu-terium substitution cases for the key reac-tions (2) and (6) in mechanism 2:

1. Substitution of the H atom that transfersto the C atom.

2. Substitution of all H atoms taking part inthe water-assisted mechanism (e.g., thosein the concerted transition state)

3. Substitution of all H atoms in the watercluster.

We selected the binding sites associated withthe highest and lowest activation energy. Ourresults are summarized in Table 3.

The transition states concern a concerted mo-tion, involving many molecules which is nicely

6

illustrated by the fact that the activation energyis only mildly affected in Case 1, but clearly in-creases for Cases 2 and 3. The increase of thebarrier height finds its origin in the delocalizednature of the transition state, significantly re-ducing the magnitude of the imaginary transi-tion mode. Despite the increase, the activationenergies remain low enough to expect that areaction between the carbon atom and deuter-ated water can take place, for instance via tun-neling.27 Note that the barrierless pathways for3HCOH formation in reaction (2) are also stillrelevant for reactions with HDO or D2O.

Our theoretical results were tested againstfour tailored experiments that we per-formed following H2CO detection in previ-ous works.27,32,38 Table 4 outlines the four co-depositions of a mixture of C atoms, H2O,and/or D2O at 10 K using the SURFRESIDE3

setup.38 Experiments 1 and 2 test formationof H2CO and D2CO. Experiment 3 probes thesimultaneous formation of H2CO, HDCO andD2CO isotopologues from reaction of carbonwith H2O, HDO, and D2O. In particular, HDOis formed as the result of D/H exchange onthe metal walls of the dosing line capillariesupon simultaneous injection of H2O and D2Ovapours which thus leads to the deposition of aH2O:HDO:D2O mixture. Note that while thereis no production of C atoms in our control ex-periment 4, the source was operated at a lowtemperature of 1200 ◦C in order to account forthe possible effects of contaminants from thedegasing C-atom source, such as H2O and CO2.The growing ice was monitored with ReflectionAbsorption InfraRed Spectroscopy (RAIRS).More details on the experiments are given inthe Supporting Information.

The RAIR spectra corresponding to exper-iments 1 − 4 are depicted in Fig. 3 for thewavenumber range specific to the CO stretch,CH2 scissoring and rocking modes of the threeformaldehyde isotopes of interest here: 1750 −980 cm−1. The peak positions for all relevantmolecules are indicated with a line at the cor-responding wavenumber in the figure and arelisted in the Supporting Information. Exper-iments 1 and 2 show clear H2O / H2CO andD2O / D2CO detections, respectively. This

Table 4: Summary of the four key experiments,including the atomic and molecular effectivefluxes and the total time of the experiment.

# C H2O D2O Timecm−2 s−1 cm−2 s−1 cm−2 s−1 min

1 5× 1011 1.2× 1013 – 502 5× 1011 – 1.4× 1013 503 5× 1011 1.4× 1013 1.6× 1013 1254 – 1.4× 1013 1.6× 1013 65

also points towards the fact that indeed in-tersystem crossing has occurred on a shorttimescale. Integrating the CO stretch area un-der the peak, and taking into account the cor-responding absorption band strength,39,40 thisleads to a H2CO/D2CO ratio of about 1.2. Inother words, we find a very small kinetic isotopeeffect, that we attribute to the overall isotopeeffect of the various reactions in the reactionsequence C + H2OASW −−→ H2COASW. Ex-periment 3 shows that all three formaldehydeisotopologues have been formed simultaneouslyfrom a co-deposition of C + H2O/HDO/D2O.No kinetic isotope effect can be calculated in ex-periment 3 because the band strength of the COstretch mode for HDCO is not reported and thethree carbonyl stretch modes overlap with eachother and with the H2O bending mode. Controlexperiment 4 indicates that no formaldehyde isformed without impacting C-atoms and also noeffect is seen due to potential contaminants inthe C-atom source. These experimental resultsare fully in line with the computational out-comes presented above in Table 3.

We conclude this work emphasizing that wehave found an intricate reaction mechanismfor the formation of formaldehyde from carbonatoms on top of amorphous solid water, mod-elled via an ice cluster approach. This worknow reconciles the mismatch between gas-phaseand solid-state experimental and theoretical ef-forts through the finding that the water assistedmechanism (2) relies on the collective inter-play of the water ice network leading to a con-certed proton transfer. Therefore, it operatesmuch more efficiently than the internal hydro-gen migration mechanism (1), and explains the

7

10001100120013001400150016001700Wavenumber (cm 1)

0.000

0.005

0.010

0.015

0.020

0.025

Abso

rban

ce (a

rb. u

nits

)

} } }

Exp. 1: C + H2OExp. 2: C + D2OExp. 3: C + H2O + D2OExp. 4: H2O + D2O

H2COD2COHDCO

H2O HDO D2O

Figure 3: RAIR spectra of the four experiments outlined in Table 4. The vertical lines indicatepeak positions of the formaldehyde and water isotopologues.

lack of detection of reaction products in matrix-isolation experiments.13,15 Furthermore, we ex-pand on previous computational work19,20 dis-cussing the initial stages of 3C+H2O interactionwith ASW.

The low or null activation barriers for thewater-assisted mechanism explain the rapidsolid-state formation of H2CO on amorphoussolid water seen experimentally. They are alsoconsistent with the minute kinetic isotope ef-fect observed for the formation of D2CO, incontrast with gas-phase studies16 and in agree-ment with a recent laboratory effort.27 The in-ternal isomerization mechanism (Mechanism 1)in turn, analogous to the gas-phase process,16,17

presents barriers too high to be relevant un-der interstellar cold conditions as those foundin molecular clouds.

From an astrochemical point-of-view our re-sults serve to highlight the following implica-tions:

• The evidence of a catalytic effect of water– and potentially other hydrogen bondednetworks – suggests that proton transferreactions may operate in interstellar icesin the presence of highly polarizing adsor-bates;

• The formation of H2CO in early stagesof a molecular cloud lifetime points to an

early formation of carbon bearing (com-plex) organic molecules;

• The deuterium fractionation of observedformaldehyde in cold regions will be in-fluenced by the reaction route presentedhere, and should be taken into accountalong with the deuterium fractionationexpected from the main formation routeof formaldehyde in later states of a molec-ular cloud CO + 2 H/D;

• The formation route proposed here opensnew avenues for several astrochemicalreaction networks, e.g., formation ofmethanol via subsequent hydrogen ad-ditions, and reactions with other radicalsthat are abundant in the water ice phase,worth to be explored by astrochemicalmodels

Acknowledgement Ko-Ju Chuang andMelissa McClure are greatly thanked for stim-ulating discussions. We also thank the anony-mous reviewers for their suggestions. Com-puter time was granted by the state of Baden-Wurttemberg through bwHPC and the GermanResearch Foundation (DFG) through grantno. INST 40/467-1FUGG is greatly acknowl-edged. G.M thanks the Alexander von Hum-boldt Foundation for a post-doctoral research

8

grant. We thank the Deutsche Forschungsge-meinschaft (DFG, German Research Founda-tion) for supporting this work by funding EXC2075 - 390740016 under Germany’s ExcellenceStrategy. T.L. is grateful for support from theDutch Research Council (NWO) via a VENIfellowship (grant no. 722.017.008). Astro-chemistry in Leiden is supported by the DutchAstrochemistry Network (DAN) II (ProjectNo. 648.000.029). G.F. acknowledges financialsupport from the Russian Ministry of Scienceand Higher Education via the State Assign-ment Contract. FEUZ-2020-0038. This workhas been supported in part by the Danish Na-tional Research Foundation through the Centerof Excellence ‘InterCat’ (Grant agreement no.:DNRF150)

Supporting Information Avail-

able

Details on the theoretical framework as well asthe experimental set-up can be found in theSupporting Information. Furthermore, intrin-sic reaction coordinate calculations, consistencytests for the barrierless conversion and transi-tion state geometries are included as well.

References

(1) Boogert, A. C. A.; Gerakines, P. A.; Whit-tet, D. C. Observations of the icy universe.Annu. Rev. Astron. Astrophys. 2015, 53,541–581.

(2) Herbst, E.; van Dishoeck, E. F. Com-plex Organic Interstellar Molecules. Annu.Rev. Astron. Astrophys. 2009, 47, 427–480.

(3) Jin, M.; Garrod, R. T. Formation ofComplex Organic Molecules in Cold In-terstellar Environments through Nondiffu-sive Grain-surface and Ice-mantle Chem-istry. Astrophys. J. Suppl. Ser. 2020, 249,26.

(4) Manigand, S.; Jørgensen, J. K.; Cal-cutt, H.; Muller, H. S. P.; Lig-terink, N. F. W.; Coutens, A.; Droz-dovskaya, M. N.; van Dishoeck, E. F.;Wampfler, S. F. The ALMA-PILS survey:inventory of complex organic molecules to-wards IRAS 16293-2422 A. Astron. Astro-phys. 2020, 635, A48.

(5) Yang, Y.-L. et al. The Perseus ALMAChemistry Survey (PEACHES). I. TheComplex Organic Molecules in PerseusEmbedded Protostars. Astrophys. J.2021, 910, 20.

(6) van Dishoeck, E. F.; Black, J. H. The Pho-todissociation and Chemistry of Interstel-lar CO. Astrophys. J. 1988, 334, 771.

(7) Snow, T. P.; McCall, B. J. Diffuse Atomicand Molecular Clouds. Annu. Rev. Astron.Astrophys. 2006, 44, 367–414.

(8) Langer, W. The carbon monoxide abun-dance in interstellar clouds. Astrophys. J.1976, 206, 699–712.

(9) Keene, J.; Blake, G. A.; Phillips, T. G.;Huggins, P. J.; Beichman, C. A. The abun-dance of atomic carbon near the ioniza-tion fronts in M17 and S140. Astrophys.J. 1985, 299, 967–980.

(10) Papadopoulos, P. P.; Thi, W. F.; Viti, S.CI lines as tracers of molecular gas, andtheir prospects at high redshifts. Month.Not. R. Astron. Soc. 2004, 351, 147–160.

(11) Burton, M. G.; Ashley, M. C. B.;Braiding, C.; Freeman, M.; Kulesa, C.;Wolfire, M. G.; Hollenbach, D. J.; Row-ell, G.; Lau, J. Extended Carbon LineEmission in the Galaxy: Searching forDark Molecular Gas along the G328 Sight-line. Astrophys. J. 2015, 811, 13.

(12) Bisbas, T. G.; Schruba, A.; vanDishoeck, E. F. Simulating the atomicand molecular content of molecular cloudsusing probability distributions of physicalparameters. Mon. Not. R. Astron. Soc.2019, 485, 3097–3111.

9

(13) Ortman, B. J.; Hauge, R. H.; Mar-grave, J. L.; Kafafi, Z. H. Reactions ofsmall carbon clusters with water in cryo-genic matrices: The FTIR spectrum ofhydroxyethynylcarbene. J. Phys. Chem.1990, 94, 7973–7977.

(14) Flanagan, G.; Ahmed, S. N.; Shevlin, P. B.Formation of Carbohydrates in the Reac-tion of Atomic Carbon with Water. J. Am.Chem. Soc. 1992, 114, 3892–3896.

(15) Schreiner, P. R.; Reisenauer, H. P. The“Non-Reaction” of Ground-State TripletCarbon Atoms with Water Revisited.ChemPhysChem 2006, 7, 880–885.

(16) Hickson, K. M.; Loison, J.-C.; Nunez-Reyes, D.; Mereau, R. Quantum Tunnel-ing Enhancement of the C + H2O and C+ D2O Reactions at Low Temperature. J.Phys. Chem. Lett. 2016, 7, 3641–3646.

(17) Li, J.; Xie, C.; Guo, H. Kinetics and dy-namics of the C(3P) + H2O reaction on afull-dimensional accurate triplet state po-tential energy surface. Phys. Chem. Chem.Phys. 2017, 19, 23280–23288.

(18) Wakelam, V.; Loison, J. C.; Mereau, R.;Ruaud, M. Binding energies: New valuesand impact on the efficiency of chemicaldesorption. Mol. Astrophys. 2017, 6, 22–35.

(19) Shimonishi, T.; Nakatani, N.; Furuya, K.;Hama, T. Adsorption energies of carbon,nitrogen, and oxygen atoms on the low-temperature amorphous water ice: A sys-tematic estimation from quantum chem-istry calculations. Astrophys. J. 2018,855, 27.

(20) Duflot, D.; Toubin, C.; Monnerville, M.Theoretical Determination of Binding En-ergies of Small Molecules on InterstellarIce Surfaces. Front. Astron. Space Sci.2021, 8, 24.

(21) Chuang, K. J.; Fedoseev, G.; Ioppolo, S.;van Dishoeck, E. F.; Linnartz, H. H-atom addition and abstraction reactions

in mixed CO, H2CO and CH3OH ices- an extended view on complex organicmolecule formation. Mon. Not. R. Astron.Soc. Lett. 2020, 455, 1702–1712.

(22) Butscher, T.; Duvernay, F.; Rimola, A.;Segado-Centellas, M.; Chiavassa, T. Radi-cal recombination in interstellar ices, a notso simple mechanism. Phys. Chem. Chem.Phys. 2017, 19, 2857–2866.

(23) Watanabe, N.; Kouchi, A. Efficient For-mation of Formaldehyde and Methanol bythe Addition of Hydrogen Atoms to CO inH2O-CO Ice at 10 K. Astrophys. J. 2002,571, L173–L176.

(24) Fuchs, G. W.; Cuppen, H. M.; Iop-polo, S.; Romanzin, C.; Bisschop, S. E.;Andersson, S.; Van Dishoeck, E. F.; Lin-nartz, H. Hydrogenation reactions in in-terstellar CO ice analogues. Astron. As-trophys. 2009, 505, 629–639.

(25) Caselli, P.; Walmsley, C. M.; Tafalla, M.;Dore, L.; Myers, P. C. CO Depletion in theStarless Cloud Core L1544. Astrophys. J.1999, 523, L165–L169.

(26) Terwisscha van Scheltinga, J. et al. TheTW Hya Rosetta Stone Project. II. Spa-tially Resolved Emission of FormaldehydeHints at Low-temperature Gas-phase For-mation. The Astrophysical Journal 2021,906, 111.

(27) Potapov, A.; Krasnokutski, S.; Jager, C.;Henning, T. A New “Non-energetic”Route to Complex Organic Molecules inAstrophysical Environments: The C +H2O → H2CO Solid-state Reaction. 2021.

(28) Pontoppidan, K. M.; Fraser, H. J.; Dar-tois, E.; Thi, W. F.; van Dishoeck, E. F.;Boogert, A. C. A.; d’Hendecourt, L.; Tie-lens, A. G. G. M.; Bisschop, S. E. A 3-5µ m VLT spectroscopic survey of embed-ded young low mass stars I. Structure ofthe CO ice. Astron. Astrophys. 2003, 408,981–1007.

10

(29) Keane, J. V.; Tielens, A. G. G. M.;Boogert, A. C. A.; Schutte, W. A.; Whit-tet, D. C. B. Ice absorption features in the5-8 µm region toward embedded proto-stars. Astron. Astrophys. 2001, 376, 254–270.

(30) Schreiner, P. R.; Reisenauer, H. P.;Pickard Iv, F. C.; Simmonett, A. C.;Allen, W. D.; Matyus, E.; Csaszar, A. G.Capture of hydroxymethylene and its fastdisappearance through tunnelling. Nature2008, 453, 906–909.

(31) Ferrero, S.; Martınez-Bachs, B.; Enrique-Romero, J.; Rimola, A. Adsorption ofAtoms on a Crystalline Ice Surface Model:Results from Periodic ab Initio Simula-tions. Computational Science and Its Ap-plications – ICCSA 2020. Cham, 2020; pp553–560.

(32) Qasim, D.; Fedoseev, G.; Chuang, K. J.;He, J.; Ioppolo, S.; van Dishoeck, E. F.;Linnartz, H. An experimental study of thesurface formation of methane in interstel-lar molecular clouds. Nat. Astron. 2020,4, 781–785.

(33) Zhang, P.; Maeda, S.; Morokuma, K.;Braams, B. J. Photochemical reactions ofthe low-lying excited states of formalde-hyde: T1 / S0 intersystem crossings, char-acteristics of the S1 and T1 potential en-ergy surfaces, and a global T1 potentialenergy surface. J. Chem. Phys. 2009, 130,114304.

(34) Hwang, D.-Y.; Mebel, A. M.; Wang, B.-C.Ab initio study of the addition of atomiccarbon with water. Chem. Phys. 1999,244, 143–149.

(35) Kruger, B. C.; Schafer, T.; Wodtke, A. M.;Park, G. B. Quantum-state resolved life-time of triplet (a3A2) formaldehyde. J.Mol. Spectrosc. 2019, 362, 61–68.

(36) Peters, P. S.; Duflot, D.; Faure, A.; Ka-hane, C.; Ceccarelli, C.; Wiesenfeld, L.;Toubin, C. Theoretical investigation of the

isomerization of trans-HCOH to H2CO:An example of a water-catalyzed reaction.J. Phys. Chem. A 2011, 115, 8983–8989.

(37) Ceccarelli, C.; Caselli, P.; Bockelee-Morvan, D.; Mousis, O.; Pizzarello, S.;Robert, F.; Semenov, D. Deuterium Frac-tionation: The Ariadne’s Thread fromthe Precollapse Phase to Meteorites andComets Today. Protostars and Planets VI.2014; p 859.

(38) Qasim, D.; Witlox, M. J. A.; Fedoseev, G.;Chuang, K. J.; Banu, T.; Krasnokut-ski, S. A.; Ioppolo, S.; Kastner, J.; vanDishoeck, E. F.; Linnartz, H. A cryogenicice setup to simulate carbon atom reac-tions in interstellar ices. Rev. Sci. Instrum.2020, 91, 054501.

(39) Hidaka, H.; Watanabe, M.; Kouchi, A.;Watanabe, N. Reaction Routes in the CO-H2CO-dn -CH3OH-dm System Clarifiedfrom H(D) Exposure of Solid Formalde-hyde at Low Temperatures. Astrophys. J.2009, 702, 291–300.

(40) Strickler, S. J.; Barnhardt, R. J. Abso-lute vibronic intensities in the 1A2 ← 1A1

absorption spectrum of formaldehyde. J.Phys. Chem. 1982, 86, 448–455.

11

Graphical TOC Entry

12