PHYSICAL REVIEW LETTERS 131, 238001 (2023)

Editors' Suggestion Featured in Physics

Mechanism of Ultraviolet-Induced CO Desorption from CO Ice:

Role of Vibrational Relaxation Highlighted
Samuel Del Fré ,1,* Alejandro Rivero Santamaría ,1,† Denis Duflot ,1,‡ Romain Basalgète ,2,§ Géraldine Féraud ,2,∥
Mathieu Bertin ,2,¶ Jean-Hugues Fillion ,2,** and Maurice Monnerville 1,††
1
Univ. Lille, CNRS, UMR 8523 - PhLAM - Physique des Lasers Atomes et Molécules, F-59000 Lille, France
2
Sorbonne Université, Observatoire de Paris, Université PSL, CNRS, LERMA, F-75005 Paris, France
(Received 23 May 2023; accepted 13 October 2023; published 6 December 2023)

Although UV photon-induced CO ice desorption is clearly observed in many cold regions of the
Universe as well as in the laboratory, the fundamental question of the mechanisms involved at
the molecular scale remains debated. In particular, the exact nature of the involved energy transfers in
the indirect desorption pathway highlighted in previous experiments is not explained. Using ab initio
molecular dynamics simulations, we explore a new indirect desorption mechanism in which a highly
vibrationally excited CO (v ¼ 40) within an aggregate of 50 CO molecules triggers the desorption of
molecules at the surface. The desorption originates first from a mutual attraction between the excited
molecule and the surrounding molecule(s), followed by a cascade of energy transfers, ultimately resulting
in the desorption of vibrationally cold CO (∼95% in v ¼ 0). The theoretical vibrational distribution, along
with the kinetic energy one, which peaks around 25 meV for CO with low rotational levels (v ¼ 0, J < 7),
is in excellent agreement with the results obtained from VUV laser induced desorption (157 nm) of CO
(v ¼ 0, 1) probed using REMPI.

DOI: 10.1103/PhysRevLett.131.238001

Carbon monoxide (CO) is known to be a primordial transition (DIET) mechanism [13] ] and to involve the 2 to
tracer of the gas phase density and temperature of many 5 outmost layers of the ice only (∼7 to 18 Å) [12,14,18].
regions of the interstellar medium (ISM). In star and planet- The exact mechanism at the origin of the desorption is,
forming cold regions, gaseous CO undergoes a freeze-out however, still largely unknown and its dependence on many
onto (sub)micron-sized dust grains, resulting in the for- parameters is still debated. Probably the most striking
mation of solid CO layers [1,2]. This condensation is, property of the mechanism is that it is mostly indirect:
however, counterbalanced by nonthermal desorption proc- synchrotron-based experiments have revealed that the
esses, maintaining a certain budget of rotationally cold desorbing molecule is not necessarily the one which
carbon monoxide into the gas phase (e.g., [3]). The absorbed the photon, but instead the dominating process
desorption subsequent to the irradiation of the icy grains involves an energy transfer between the excited molecule
by vacuum UV (VUV) photons (5–13.6 eV), has been for a and a surface molecule [14]. This indirect desorption has
long time proposed as a dominant mechanism, for instance, even been shown to be operative between an excited CO
in protoplanetary disks [4]. Constraining the efficiency and and other coadsorbed molecules, such as N2 or CH4
mechanisms of the CO UV-photodesorption became of [18,19], making this mechanism particularly interesting
paramount importance since it may impact both the to predict photodesorption yields from composite, and thus
location of the CO snowlines, thereby affecting the for- more astrophysically relevant, ices. However, none of the
mation dynamics and composition of exoplanets [5–7], and above-mentioned experimental studies were able to pro-
the chemical richness in these cold media as CO ices are a pose a detailed desorption mechanism that would explain
potential starting point for a complex chemistry leading to all the observed properties, and, in particular, none could
methanol formation, and to a rich subsequent organic bring a hint on the exact nature of the involved energy
chemistry [8,9]. For these reasons, the VUV-photodesorp- transfer, mainly because none could access the total energy
tion of solid CO has been for decades the subject of a large balance of the process by probing the amount of transla-
panel of experimental studies aiming to provide absolute tional and internal energy left in the desorbates. There is
desorption yields to the astrochemical community [10–17]. only one detailed theoretical study concerning the photo-
Below 10 eV, the yield is of about 1 desorbed molecule per desorption of pure CO ices performed by van Hemert et al.
absorbed photon, although it depends on the temperature at [20]. They simulated the dynamics of large amorphous and
which the ice was grown [15]. The desorption has been crystalline CO clusters after the UV excitation of a single
shown to be triggered by the A1Π − X1Σþ transition in the CO molecule by using molecular dynamics (MD) simu-
7–10 eV energy range [desorption induced by electronic lations based on a classical force field. The proposed

0031-9007=23=131(23)=238001(6) 238001-1 © 2023 American Physical Society

 PHYSICAL REVIEW LETTERS 131, 238001 (2023)

mechanism involves the excited molecule which, after event), referred to as the time-of-flight (TOF) of the
internal conversion, lands back to its fundamental elec- photodesorbing molecules. TOF spectra are then converted
tronic state in a very high vibrational state, thus leading to to give one kinetic energy distribution following the
its direct desorption or to an indirect desorption of a Zimmermann et al. approach [29]. More details about
neighboring molecule. Their simulation estimates the the experimental technique can be found in the
probability of photodesorption to be a factor of 3 to 11 Supplemental Material [30], and in [57].
lower than the experimental one [13], and the indirect The AIMD simulations focused on the description of the
pathway is rather minor compared to the direct one. This is vibrational energy redistribution when an excited CO
in contradiction with the experimental results described molecule transfers its electronic energy to a high vibrational
above. In addition, their model also predicts an absence of state of its ground state. This phase corresponds to the part
desorption when a CO molecule is directly vibrationally of the DIET mechanism after the internal conversion.
excited in its electronic ground state. Particularly, the first three bands A1Πðv0 ¼ 0; 1; 2Þ ←
Here, we make a new proposition for the microscopic X1Σþðv00 ¼ 0Þ of the absorption spectrum correspond to
mechanism involved in the solid CO photodesorption, with an internal energy ranging between 8 and 8.4 eV, which can
the specific aim to finally characterize the indirect desorp- promote the molecule to the electronic ground vibrational
tion and the nature of the energy transfer between the levels v ¼ 39–41. Aggregates of 50 CO molecules were
excited molecule and its neighbors. The methodology we generated, optimized, and finally thermalized (15 K) using
employ is based on a coupled experimental and theoretical canonical AIMD calculations. The aggregate phase space
strategy. New VUV photodesorption experiments, that configurations obtained after the thermalization process
provide vibrational and translational energy distribution were modified to vibrationally excite a single CO molecule
of the photodesorbed CO molecules, are performed. The (v ¼ 40 in this work) located in the central core of the
obtained results are compared with ab initio molecular aggregate (further details in [30]). Following a standard
dynamics (AIMD) simulations performed with the Vienna procedure in gas-surface dynamics simulations [37–45] a
ab initio simulation package (VASP) based on density random aggregate prepared as mentioned above, was
functional theory (DFT) [21,22]. For these calculations selected for each trajectory to run constant energy
we use the Perdew-Burke-Ernzerhof (PBE) [23] exchange- AIMD calculations. Finally, under these conditions, 100
correlation functional, together with the DFT-D3(BJ) AIMD trajectories were propagated using VASP. The maxi-
dispersion correction [24,25]. mum propagation time was 5 ps with a time step of 0.5 fs.
The experiments were performed using an ultrahigh Two possible exit channels were taken into account, called
vacuum setup (base pressure of ∼10−10 Torr). A pure desorption and nondesorption: a molecule was considered
CO ice (∼100 ML) was deposited onto a polycrystalline desorbed when the distance from the CO center of mass and
oxygen-free high-conductivity copper substrate cooled the aggregate surface exceeded 3 Å and not desorbed when
down at 15 K by a closed-cycle Helium cryostat. VUV after 5 ps no molecule fulfilled the above condition. The
photons were generated by four-wave frequency mixing in total energy was well conserved for each trajectory with a
Xe [26,27], resulting in a pulsed VUV beam (10 Hz, standard deviation of ∼30 meV. The energy distributions
duration of 10 ns, 106 photons per pulse) that was tuned to of the molecules along the dynamics were obtained using
the A-X (0,0) transition of CO (157 nm) and focused at the the standard semiclassical determination method for the
CO ice. The individual rovibrational states of photode- translational, vibrational, and rotational energies [58].
sorbed neutral CO (v, J) were probed by a resonance- Among the 100 trajectories, 88% resulted in the desorp-
enhanced multi-photon ionization (REMPI) technique tion of CO molecule(s). The average desorption time was
using a tunable pulsed laser (OPO, 200–700 nm) focused 2 ps and interestingly, the desorption was not observed
a few millimeters in front of the ice surface. After resonant before 700 fs, suggesting that more complex mechanisms
ionization of 1Σþ CO (v, J) states, the ions are detected by than a fast release of the excited molecules into the gas
mass spectrometry using a quadripolar mass spectrometer phase must take place. An example of the time evolution of
equipped with an ion extractor. The Q branches of the a typical trajectory for which desorption takes place is
REMPIð2 þ 1Þ B1Σþðv0 ¼ 0 or 1Þ ← X1Σþðv00 ¼ 0 or 1Þ presented in Fig. 1. By closely examining this trajectory we
bands near 230 nm [28] were used to probe the distribution can conclude that the dynamics of the desorption process
of rovibrational states populated after desorption. The can be described in three fundamental steps: (1) First, the
search for vibrationally excited CO in the v ¼ 2 (near excited molecule vibrates inside the aggregate while
230.5 nm), 16 (near 224.9 nm), and 19 (near 246.4 nm) retaining the initially deposited vibrational energy. No
states was also carried out, but no ionization signal was significant energy transfers to other modes (translational
found. The kinetic energy distribution of photodesorbing or rotational) are observed. (2) The excited molecule and
CO molecules was estimated by monitoring the variations one (∼75%) or two (∼25%) CO molecule(s) in its vicinity
of the ion signal with the time delay between the VUV begin to be mutually attracted and accordingly gain trans-
pulse (desorption event) and the ionizing pulse (probe lational energy, thus leading to their collision via a “kick

238001-2
 PHYSICAL REVIEW LETTERS 131, 238001 (2023)

desorption
collision
(1) (2) (3)

1 2 3 1 2 3 1 2 3

FIG. 1. Upper panel: Snapshots of a typical desorbing trajectory. The excited (blue), kicked (magenta), and desorbed (green)
molecules are highlighted. Lower panel: Vibrational, translational, and rotational energy of the excited, kicked, and desorbed CO
molecules along the trajectory depicted in the upper panel. The numbers 1 to 3 refer to the three fundamental steps of the desorption
process.

event.” At this time, sudden intra- and intermolecular plane with the two C facing each other in a displaced-
energy transfers from the vibration to the translational, stacked manner at fixed distances of 0.5, 1.5, 2.0, and 2.5 Å
rotational, and vibrational modes of the colliding molecules between the two molecules’ centers of mass in the y
occur. (3) The colliding molecules begin to move and direction (see Fig. 2). These particular geometries aim to
interact with other molecules within the aggregate leading mimic the disposition of the colliding molecules in the
to a cascade energy transfer effect; i.e., the translational and aggregate. The interaction energy EAB ¼ Etot − EA − EB
rotational energy acquired in step 2 is transferred to surface between molecules A and B was then scanned while
CO molecule(s) thus giving it (them) enough kinetic energy approaching them along the x axis. The results show that
to exceed the binding energy of the aggregate (around there exists an attractive potential energy well between the
100 meV in our case). two molecules which is not observed for configurations
The analysis of a typical trajectory where no desorption
is observed (see Fig. S2 in [30]) shows that a smaller
fraction of the vibrational energy of the excited molecule is A
transferred to the other energy modes during the kick, in
y
contrast to events where desorption occurs. The amount of y B
translational energy gained by the colliding molecules is
z x
not sufficient to observe the cascade energy transfer effect
explained above. Therefore, the kick event plays a funda-
mental role to achieve sufficient energy transfers and to
observe desorption.
To further investigate the colliding molecules kick event,
we present in Fig. 2 some interaction energy curves
between two CO molecules (A and B). The curves were
obtained under the previously described computational
conditions by fixing the bond lengths of A to its equilibrium
FIG. 2. Interaction energy evolution between two CO mole-
geometry (rA ¼ 1.14 Å) and B to the calculated outer cules (A and B) as a function of the distance between their centers
turning point of the vibrational state v ¼ 40 (rB ¼ of mass for fixed Δy values. The CO molecules bond lengths are
1.98 Å). The two molecules were then placed in the xy set to rA ¼ 1.14 and rB ¼ 1.98 Å.

238001-3
 PHYSICAL REVIEW LETTERS 131, 238001 (2023)

where the two molecules are taken at the equilibrium bond be the cause of the difficulties for the mentioned PES to
length rA ¼ rB ¼ 1.14 Å (see Fig. S3 in [30]). In general, correctly describe the interactions between a highly vibra-
starting from geometries where the molecules are far apart tionally excited CO molecule with its pair at the equilib-
(Δy ¼ 2.5 Å) to those where they are closer (Δy ¼ 0.5 Å), rium bond length. This is confirmed by the interaction
the position of the well shifts towards smaller distances and energy curves (see Fig. S4 in [30]) obtained using the
its depth increases. In particular, the closest conformation aforementioned PES that do not correctly describe the
to the kick event geometry (red curve, at distance attractive potential energy well observed in our calculations
around 3 Å in Fig. 2) exhibits the strongest interaction (see Fig. 2), which is responsible for triggering the collision
(EAB ∼ −0.8 eV). The existence of such attraction is likely between the excited CO molecule and the surrounding CO
explained by the particular evolution of the CO dipole molecule(s).
moment upon bond stretching in the rCO range explored in Finally, the remarkable quantitative and qualitative
our simulations. As has been previously shown (e.g., agreement between our simulations and experiments is
[54,59–62]), the CO displays a dipole moment of confirmed by comparing the vibrational and kinetic energy
∼0.12 D [62] in the direction Cδ− Oδþ at equilibrium, distributions of the desorbed molecules. Experimentally,
but it reverses to Cδþ Oδ− and increases at larger rCO up the photodesorbed CO molecules were found to be mostly
to its maximum value of ∼1.45 D for rCO ∼ 1.95 Å [62]. in their ground vibrational level (∼96% in the v ¼ 0 level)
Hence, a highly attractive dipole-dipole interaction is while the population of the v ¼ 1 level is estimated to
created between two CO molecules when one of them ∼4%. It should be mentioned that the chosen REMPI
starts to vibrate, especially at high v. It should be transition (near 230 nm) allows only to detect vibrationally
mentioned that a similar electrostatic behavior when the cold CO (v ¼ 0, 1). However, our simulations show also
CO stretch is excited has already been highlighted in that the desorbed molecules are predominantly vibration-
several studies on the CO-NaCl system, e.g., [63–65]. ally cold (∼95% of v ¼ 0) and desorption of molecules
The agreement between past experiments, notably [14], with v > 0 is observed only in the rare cases where the
and theory is remarkable since in both cases, the large excited molecule kicked a molecule close to the surface,
majority of UV excitations (experiments) or vibrational resulting in desorption of the latter. Figure 3 shows the
excitations (simulations) of CO molecules results in a experimental (derived from TOF spectra, see Fig. S5 in
desorption event. The high probability, observed in our [30]) and theoretical translational energy distributions of
simulations, in the indirect desorption channel is in dis- the desorbed molecules in the v ¼ 0 and J < 7 states. A
agreement with the results of van Hemert et al. [20] that non-Boltzmann kinetic energy distribution has been
show a very small contribution of the indirect pathway. observed in both cases, with the maximum of the distri-
However, in order to directly compare these results with our butions located around 25 meV, and very few molecules
calculations, it is necessary to note that our study was were detected with translational energies higher than
focused on the second part of the DIET mechanism, so it 200 meV. Therefore, only a small fraction (∼2%) of the
did not take into account the configuration space explored vibrational energy initially deposited is finally transferred
by the molecule in the excited electronic state potential to the desorbed molecule(s). In fact, even if the initial
energy surface (PES) after the UV excitation. Nonetheless, vibrational energy dissipates within the aggregate in its
our simulation provides very valuable information about different modes, only a minimum translational energy is
the mechanism of vibrational deexcitation of the CO
molecule within the aggregate. Indeed, our results clearly
indicate that the transfer from vibrational to translational
energy modes inside the CO aggregates is essential in the
desorption mechanism. The significant discrepancy regard-
ing the role of vibrational energy transfer between our
theoretical results and those obtained by van Hemert et al.
[20], could be related to the different methods used to
describe the interactions in the simulations. The analytical
PESs used in [20] (ground and excited states) are pairwise
potentials based on CO dimers interaction. The site-site
PES, frequently used to study this kind of system, may,
however, present difficulties to describe the multibody
effects in complex systems. Our results indicate that the
discrepancies primarily stem from the limited sampling of
their analytical ground state PES, which only includes the FIG. 3. Experimental kinetic energy distribution derived from
first three vibrational levels of the CO ground state. The the TOF spectra [30] (black curve) and theoretical kinetic energy
lack of points describing higher vibrational levels seems to distribution of the desorbed CO molecules (red bars).

238001-4
 PHYSICAL REVIEW LETTERS 131, 238001 (2023)

required to separate a molecule from the aggregate. This *

[email protected]
minimum energy is acquired by the future desorbed †
[email protected]
molecule(s) through a final collision. In particular, our ‡
[email protected]
simulations show that this type of event transmits between §
[email protected]
∥
100 and 300 meV to the molecule(s) translational mode. [email protected]
¶
This kinetic energy is mainly used to overcome the [email protected]
**
desorption barrier (∼100 meV) and the remaining energy [email protected]
††
(between ∼5–200 meV) is used by the molecule(s) to move [email protected]
away from the aggregate. [1] K. M. Pontoppidan, H. J. Fraser, E. Dartois, W. F. Thi, E. F.
van Dishoeck, A. C. A. Boogert, L. d’Hendecourt, A.
In summary, we performed both experiments on UV
Tielens, and S. E. Bisschop, Astron. Astrophys. 408, 981
photodesorption from CO amorphous ice and AIMD (2003).
simulations on the vibrational excitation (v ¼ 40) of one [2] A. A. Boogert, P. A. Gerakines, and D. C. Whittet, Annu.
CO within an aggregate of 50 CO molecules. A detailed Rev. Astron. Astrophys. 53, 541 (2015).
analysis of the energy transfers taking place in the [3] V. Piétu, A. Dutrey, and S. Guilloteau, Astron. Astrophys.
aggregate was conducted in the simulations to explain 467, 163 (2007).
the desorption mechanism. We have shed new light on the [4] K. Willacy and W. D. Langer, Astrophys. J. 544, 903
UV-induced photodesorption mechanism of CO from pure (2000).
CO ice by highlighting the essential role of molecular [5] K. I. Öberg, R. Murray-Clay, and E. A. Bergin, Astrophys.
vibration in triggering desorption. In particular, the theo- J. Lett. 743, L16 (2011).
[6] K. I. Öberg and E. A. Bergin, Phys. Rep. 893, 1 (2021).
retical analysis of the energy redistribution after the vibra-
[7] M. Minissale, Y. Aikawa, E. Bergin, M. Bertin, W. A.
tional excitation shows that sufficient energy transfers Brown, S. Cazaux, S. B. Charnley, A. Coutens, H. M.
between the excited molecule and neighboring one(s) Cuppen, V. Guzman, H. Linnartz, M. R. S. McCoustra,
occur, leading to the indirect desorption of CO molecule A. Rimola, J. G. Schrauwen, C. Toubin, P. Ugliengo, N.
(s) in almost all trajectories after a kick event. We have Watanabe, V. Wakelam, and F. Dulieu, ACS Earth Space
shown that this kick can be explained by a strong attractive Chem. 6, 597 (2022).
interaction created between two CO molecules when one of [8] N. Watanabe and A. Kouchi, Astrophys. J. Lett. 571, L173
them starts to vibrate. Finally, the vibrational and kinetic (2002).
energy characteristics of the desorbed molecules obtained [9] G. W. Fuchs, H. M. Cuppen, S. Ioppolo, C. Romanzin, S. E.
from the experiments and the simulations have been found Bisschop, S. Andersson, E. F. van Dishoeck, and H.
Linnartz, Astron. Astrophys. 505, 629 (2009).
to be remarkably consistent, thus providing substantial
[10] K. I. Öberg, G. W. Fuchs, Z. Awad, H. J. Fraser, S.
support to the fundamental role of vibrational deexcitation Schlemmer, E. F. Van Dishoeck, and H. Linnartz, Astro-
in the desorption mechanism. The methodology imple- phys. J. 662, L23 (2007).
mented here is easily applicable to more complex ice [11] K. I. Öberg, E. F. van Dishoeck, and H. Linnartz, Astron.
mixtures, more relevant to their field of study. Indeed, the Astrophys. 496, 281 (2009).
fate of the energy transfers, potentially leading to indirect [12] G. M. Muñoz Caro, A. Jimenez-Escobar, J. A. Martin-Gago,
desorption, when an electronically excited CO molecule is C. Rogero, C. Atienza, S. Puertas, J. M. Sobrado, and J.
surrounded by other diatomic (e.g., N2 , O2 , NO) or larger Torres-Redondo, Astron. Astrophys. 522, A108 (2010).
(e.g. CH3 OH) molecules is a fundamental question of high [13] E. C. Fayolle, M. Bertin, C. Romanzin, X. Michaut, K. I.
interest for several fields (e.g., astrophysics, terrestrial Öberg, H. Linnartz, and J.-H. Fillion, Astrophys. J. Lett.
atmosphere). 739, L36 (2011).
[14] M. Bertin, E. C. Fayolle, C. Romanzin, K. I. Öberg, X.
Michaut, A. Moudens, L. Philippe, P. Jeseck, H. Linnartz,
This project was provided with computer and storage and J.-H. Fillion, Phys. Chem. Chem. Phys. 14, 9929 (2012).
resources by GENCI at TGCC thanks to the grants 2022- [15] G. M. Muñoz Caro, Y.-J. Chen, S. Aparicio, A. Jiménez-
A0110801859 and 2023-A0130801859 and by the Centre Escobar, A. Rosu-Finsen, J. Lasne, and M. R. S. McCoustra,
de Ressources Informatiques (CRI) of the Université de Astron. Astrophys. 589, A19 (2016).
Lille. This work was funded by the ANR PIXyES project, [16] D. M. Paardekooper, G. Fedoseev, A. Riedo, and H.
Grant ANR-20-CE30-0018 of the French Agence Linnartz, Astron. Astrophys. 596, A72 (2016).
Nationale de la Recherche, by the Region Ile-de-France [17] N.-E. Sie, Y.-T. Cho, C.-H. Huang, G. M. M. Caro, L.-C.
DIM-ACAV+ program, and by the French Programme Hsiao, H.-C. Lin, and Y.-J. Chen, Astrophys. J. 938, 48
National “Physique et Chimie du Milieu Interstellaire” (2022).
[18] M. Bertin, E. C. Fayolle, C. Romanzin, H. A. M. Poderoso,
(PCMI) of CNRS/INSU cofunded by CEA and CNES. We
X. Michaut, L. Philippe, P. Jeseck, K. I. Öberg, H. Linnartz,
also acknowledge funding from Labex MiChem, part of the and J.-H. Fillion, Astrophys. J. 779, 120 (2013).
French state funds managed by the ANR within the [19] R. Dupuy, M. Bertin, G. Féraud, X. Michaut, P. Jeseck, M.
investissements d’avenir program under Reference Doronin, L. Philippe, C. Romanzin, and J.-H. Fillion,
No. ANR-11-10EX-0004-02, and from the INSU-CSAA. Astron. Astrophys. 603, A61 (2017).

238001-5
 PHYSICAL REVIEW LETTERS 131, 238001 (2023)

[20] M. C. van Hemert, J. Takahashi, and E. F. van Dishoeck, J. [46] H.-J. Werner and P. J. Knowles, J. Chem. Phys. 89, 5803
Phys. Chem. A 119, 6354 (2015). (1988).
[21] G. Kresse and J. Furthmüller, Comput. Mater. Sci. 6, 15 [47] P. J. Knowles and H.-J. Werner, Chem. Phys. Lett. 145, 514
(1996). (1988).
[22] G. Kresse and J. Furthmüller, Phys. Rev. B 54, 11169 [48] S. R. Langhoff and E. R. Davidson, Int. J. Quantum Chem.
(1996). 8, 61 (1974).
[23] J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. [49] H.-J. Werner and P. J. Knowles, J. Chem. Phys. 82, 5053
77, 3865 (1996). (1985).
[24] S. Grimme, J. Antony, S. Ehrlich, and H. Krieg, J. Chem. [50] P. J. Knowles and H.-J. Werner, Chem. Phys. Lett. 115, 259
Phys. 132, 154104 (2010). (1985).
[25] S. Grimme, S. Ehrlich, and L. Goerigk, J. Comput. Chem. [51] T. V. Mourik, A. K. Wilson, and T. H. Dunning, Mol. Phys.
32, 1456 (2011). 96, 529 (1999).
[26] R. Hilbig and R. Wallenstein, Appl. Opt. 21, 913 (1982). [52] A. K. Wilson, T. van Mourik, and T. H. Dunning, J. Mol.
[27] R. Hilbig and R. Wallenstein, IEEE J. Quantum Electron. Struct. 388, 339 (1996).
19, 194 (1983). [53] H.-J. Werner, P. J. Knowles, F. R. Manby, J. A. Black, K.
[28] S. Wurm, P. Feulner, and D. Menzel, J. Chem. Phys. 105, Doll, A. Heßelmann, D. Kats, A. Köhn, T. Korona, D. A.
6673 (1996). Kreplin, Q. Ma, T. F. Miller, A. Mitrushchenkov, K. A.
[29] F. M. Zimmermann and W. Ho, Surf. Sci. Rep. 22, 127 Peterson, I. Polyak, G. Rauhut, and M. Sibaev, J. Chem.
(1995). Phys. 152, 144107 (2020).
[30] See Supplemental Material at http://link.aps.org/ [54] N. He, M. Huang, and F. A. Evangelista, J. Phys. Chem. A
supplemental/10.1103/PhysRevLett.131.238001 for com- 127, 1975 (2023).
putational and experimental details, which includes [55] A. A. Radzig and B. M. Smirnov, Reference Data on Atoms,
Refs. [20–29,31–56]. Molecules, and Ions (Springer Science & Business Media,
[31] P. E. Blöchl, Phys. Rev. B 50, 17953 (1994). New York, 2012), Vol. 31.
[32] G. Kresse and D. Joubert, Phys. Rev. B 59, 1758 (1999). [56] S. Wurm, State-resolved investigation of the energy dis-
[33] H. J. Monkhorst and J. D. Pack, Phys. Rev. B 13, 5188 tribution over all degrees of freedom of electronically
(1976). stimulated desorbed CO from monolayers, and condensates,
[34] A. Cassidy, M. R. S. McCoustra, and D. Field, Acc. Chem. Ph.D. thesis, Technische Universität München, 1995.
Res. 56, 1909 (2023). [57] R. Basalgète, Laboratory Astrophysics applied to the VUV
[35] S. Nosé, J. Chem. Phys. 81, 511 (1984). and X-ray photo-induced desorption from molecular ices,
[36] W. G. Hoover, Phys. Rev. A 31, 1695 (1985). Ph.D. thesis, The institution is Sorbonne Université, 2022.
[37] A. Groß and A. Dianat, Phys. Rev. Lett. 98, 206107 (2007). [58] G. D. Billing, Dynamics of Molecule Surface Interaction
[38] F. Nattino, C. Díaz, B. Jackson, and G.-J. Kroes, Phys. Rev. (John Wiley & Sons, New York, 2000).
Lett. 108, 236104 (2012). [59] S. R. Langhoff and C. W. Bauschlicher Jr, J. Chem. Phys.
[39] F. Nattino, H. Ueta, H. Chadwick, M. E. van Reijzen, R. D. 102, 5220 (1995).
Beck, B. Jackson, M. C. van Hemert, and G.-J. Kroes, J. [60] M. Buldakov, V. Cherepanov, E. Koryukina, and Y. N.
Phys. Chem. Lett. 5, 1294 (2014). Kalugina, J. Phys. B 42, 105102 (2009).
[40] B. Kolb and H. Guo, J. Chem. Phys. 145, 011102 (2016). [61] J. Chen, J. Li, J. M. Bowman, and H. Guo, J. Chem. Phys.
[41] D. Novko, I. Lončarić, M. Blanco-Rey, J. I. Juaristi, and M. 153, 054310 (2020).
Alducin, Phys. Rev. B 96, 085437 (2017). [62] V. V. Meshkov, A. Y. Ermilov, A. V. Stolyarov, E. S.
[42] X. Zhou, B. Kolb, X. Luo, H. Guo, and B. Jiang, J. Phys. Medvedev, V. G. Ushakov, and I. E. Gordon, J. Quant.
Chem. C 121, 5594 (2017). Spectrosc. Radiat. Transfer 280, 108090 (2022).
[43] L. Zhou, B. Jiang, M. Alducin, and H. Guo, J. Chem. Phys. [63] J. Chen, S. Hariharan, J. Meyer, and H. Guo, J. Phys. Chem.
149, 031101 (2018). C 124, 19146 (2020).
[44] G. Füchsel, X. Zhou, B. Jiang, J. I. Juaristi, M. Alducin, H. [64] A. Nandi, P. Zhang, J. Chen, H. Guo, and J. M. Bowman,
Guo, and G.-J. Kroes, J. Phys. Chem. C 123, 2287 (2019). Nat. Chem. 13, 249 (2021).
[45] A. Rivero Santamaría, M. Alducin, R. Diez Muiño, and J. I. [65] A. D. Boese and P. Saalfrank, J. Phys. Chem. C 120, 12637
Juaristi, J. Phys. Chem. C 123, 31094 (2019). (2016).

238001-6