Collisions of CO₂ with surfaces are of interest to a broad range of scientific fields. In astronomy and astrochemistry, CO₂ occurs, a. o., in the interstellar medium (ISM) (d’Hendecourt and Jourdain de Muizon, 1989), both in gaseous and ice phases (Van Dishoeck, 2004), and in comets (Despois et al., 2005). In the chemical industry, CO₂ is a major reactant in the Cu/ZnO-based catalytic production of methanol (Rostrup-Nielsen, 1984). Collisions of CO₂ with surfaces are generally not reactive in a direct sense (Burghaus, 2014). Molecular sticking to a surface likely precedes subsequent chemical processes, especially at low-temperature conditions, such as those prevalent in interstellar space.

Molecular vibrations may affect the adsorption, desorption, and reactions of molecules to surfaces. Sibener and Lee (Sibener and Lee, 1994) found that internal vibrations of SF₆ and CCl₄ had a significant impact on the adsorption and reflection of these gas-phase molecules onto their bulk solids at low collision energies. In precursor-mediated reaction of CH₄ on Ir (111) (Dombrowski et al., 2015) and SiH₄ on Si(100) (Bisson et al., 2008), internal vibrations were found to enhance dissociative adsorption. A study of NO’s trapping probability on Au (111) found, on the contrary, no effect of vibrational excitation (Wodtke et al., 2005).

For CO₂, a study by Weida et al. (Weida et al., 1996) investigated CO₂ desorption from crystalline CO₂ ice at varying temperatures, hence varying desorption rates, using state-selected detection. They found no evidence of an effect on the desorption by the molecules’ vibrations, rotations and even orientations. From detailed balance, they argued that if there is no preference for a certain quantum state for desorption, neither will there be for the process of adsorption. However, due to the limitations of their methods, only the ν ₂ bend vibration in CO₂ was studied. Its vibrational energy (667 cm⁻¹ or 8.0 kJ/mol) is well below the desorption barrier, the sublimation energy being approximately 26 kJ/mol. The ν ₃ fundamental vibrational mode (2,349 cm⁻¹ or 28.1 kJ/mol) slightly exceeds the latter. A more recent study by Ioppolo et al. (Ioppolo et al., 2022) found restructuring and minor desorption of amorphous CO₂ ice after irradiation with resonant IR light (ν ₂ and ν ₃). For crystalline ice, they found no restructuring, but possible desorption. Their results were inconclusive on vibrational effects to desorption.

Here, we study the effect of vibrational excitation (ν ₃) on the adsorption of CO₂ onto CO₂ ice grown on a Cu(111) surface at 80 K. We excite the ν ₃ asymmetric stretch of CO₂ molecules in a molecular beam with light from a tunable continuous wave (cw)-infrared (IR) laser. We look for any changes in the sticking probability of the CO₂ by modulating the excitation and using a Fourier transform based detection scheme. An effect of vibrational excitation on either the adsorption or the desorption rate will change the effective sticking probability. We also monitor the surface with Reflection Absorption IR Spectroscopy (RAIRS) and look for any structural differences in the spectra of the resulting CO₂ ice layer with and without the additional vibrational excitation. At 80 K, the ice will be crystalline. Ioppolo et al. found no restructuring after IR irradiation of crystalline CO₂ ice. However, our situation may be different, as they irradiated the CO₂ ice after deposition, while we resonantly excite vibrations in the gaseous CO₂ molecules prior to adsorption.

Our experiments are performed using a home-built ultra-high vacuum (UHV) system, that has been used for non-state-resolved reaction dynamics measurements before (Cao et al., 2018; van Lent et al., 2019; Jansen and Juurlink, 2021). It consists of an analysis chamber with a base pressure of 2 × 10⁻¹⁰ mbar and a series of differentially pumped chambers for creation of a molecular beam. Figure 1 schematically illustrates the apparatus. The molecular beam is shaped by various skimmers that separate differential pumping stages and doses CO₂ onto a Cu (111) single crystal in the UHV chamber.

The molecular beam is created by expanding high-purity CO₂ (Linde Gas, purity 5.3) at 300 K. A flow controller feeds the gas at 4 bar into a vacuum chamber ( < 1 × 1 0 − 3 mbar) through a 28 μm circular orifice in a tungsten nozzle. Two skimmers and a third orifice select only the center part of the expansion, resulting in a molecular beam with an angular spread of less than 1°. During the expansion, kinetic and rotational energies are converted to translational energy along the beam axis, resulting in an average beam velocity of approximately 587 m/s as determined by time-of-flight (TOF) spectrometry (see Supplementary Material S1). The rotational temperature within the beam is cooled to approximately 26 K [see ref (Jansen et al., 2023)].

The molecular beam travels from the source chamber through two differentially pumped vacuum chambers and finally enters the UHV analysis chamber. In the first of these two chambers, a high-speed chopper wheel allows us to create short gas pulses from the continuous gas expansion for TOF spectrometry. The wheel can be stopped in the ‘open’ position. In the second of these chambers, a beam stopper can block the beam from hitting a sliding beam valve that separates this chamber from the UHV analysis chamber. The sliding beam valve contains various openings and controls the size and shape of the beam impinging onto the sample, a room temperature bolometer (pyro-electric detector, PED)) and the ionizer of the QMS used for TOF measurements.

On the beam axis in the analysis chamber, the edge of a disc can be rotated by an UHV-compatible stepper motor. The disc’s edge is manufactured to have two identical ‘open’ and two ‘closed’ sections, each of 90°. The closed sections have a slanted edge as not to reflect the molecular beam back into the last of the differentially pumped chambers. Combined with the beam stopper in the differential stage, this disc allows us to accurately perform King and Wells-type sticking probability measurements (King and Wells, 1972), but using a modulation technique as described below. We will refer to the beam blocker and the disc as the two beam flags from here onward.

The Cu single crystal is suspended from a copper block at the end of a vertical ‘cold finger’, i.e., a hollow tube protruding through an x, y, z, θ manipulator. The crystal is cooled by pouring liquid nitrogen (LN2) into the tube. It is kept at a constant temperature of 80 ± 1 K during experiments, as measured by a K-type thermocouple laser welded to the crystal’s edge. Adsorption of CO₂ onto the sample is monitored using an FTIR (Bruker Vertex 70) with an external LN2-cooled MCT detector to perform RAIRS during the experiments. The IR beam paths outside of the UHV chamber, including internal parts of the FTIR, are purged continuously with gaseous dry N₂ from an in-house system fueled by the exhaust from the local NMR facility.

The CO₂ ν ₃ antisymmetric stretch vibration is excited in the molecular beam with laser light from an OPO (Argos model 2,400, module D), which we will from now on simply refer to as the laser. Its output can be tuned and is stabilized on a CO₂ absorption frequency using the lamb dip measured in a reference cell. The laser light crosses the molecular beam a few cm after the initial expansion in the apparatus and 37 cm before molecules hit the Cu crystal. For this experiment, we tune the laser to the R (4) transition at 4,249.99 nm, where CO₂ molecules are excited from the v = 0, J = 4 state to the v = 1, J = 5 state. This transition was chosen because J = 4 has the highest population at the rotational temperature of the molecular beam [see (Jansen et al., 2023)]. A cylindrical lens in the path of the laser is used to achieve rapid adiabatic passage (Jansen et al., 2023; Chadwick et al., 2014), resulting in nearly full (80%) population inversion. Considering the fractional population of the ground rotational state and the level of saturation, we excite approximately 15% of the CO₂ molecules in the molecular beam. Herein, we neglect the fraction of vibrationally excited states nascent to the molecular beam. A moveable room-temperature bolometer (pyro-electric detector, Eltec model 406) in the UHV chamber is used to quantify IR power absorption by CO₂ molecules on the molecular beam axis. All laser beam paths outside of the UHV chamber, including the entire OPO and optical components required for frequency detection and stabilization, are enclosed and purged continuously by dry gaseous N₂. The > 400 ppm concentration of CO₂ in air otherwise destabilizes the OPO and results in large power losses through absorption in the 1–2 m beam path between the laser and the UHV chamber.

At the start of an experiment, the Cu crystal is flashed to 300 K to remove any buildup of CO₂, CO and H₂O. As we do not regularly perform extensive sputtering and annealing cycles for these experiments, the surface should not be considered an atomically clean Cu(111) surface. It is most likely partially oxidized, as studied previously (Zhang et al., 2023). When the crystal has cooled to 80 K, a RAIRS background spectrum is taken. Subsequently, the surface is continuously monitored with RAIRS while CO₂ is dosed from the molecular beam. The CO₂ flux is estimated to be 0.15 ML/s as referenced to the Cu atom surface density for Cu(111). The experiment is repeated multiple times with the laser tuned to the CO₂ excitation frequency and several times tuned to a non-resonant frequency as a control experiment. Between each experiment, the crystal is flashed to 300 K.

As described below, attempts to obtain information on the effective state-resolved sticking probability of CO₂ by analyzing IR peak integrals was unsuccessful. The integrals increase with the CO₂ dose time but did not allow us to detect variations in ‘laser on’ and’ laser off’ experiments. Hence, we measure the sticking probability of CO₂ onto CO₂ ice at 80 K with a modulated version of the King and Wells method (King and Wells, 1972) and a quadrupole mass spectrometer (QMS, Pfeiffer QMA 200) tuned to m/z 44.

A sticking probability experiment starts with a CO₂ ice being prepared by continuously dosing CO₂ onto the Cu crystal while being monitored with RAIRS. When the RAIRS spectrum shows sufficient intensity of the CO₂ absorption line, the King & Wells experiment is started by closing both molecular beam flags. The first flag is reopened and the beam enters the UHV chamber, but is blocked by the second flag. This causes a rise in the CO₂ partial pressure in the vacuum chamber, which is continuously monitored with the QMS. When the second flag is reopened, the beam impinges again onto the CO₂ ice. The fraction of CO₂ molecules that is adsorbed on the surface is removed from the background pressure in the chamber, resulting in a drop in the total pressure. The effective sticking probability is the ratio of the pressure drop after the beam impinges onto the ice and the pressure rise when the beam enters the chamber.

Note that the rate of desorption of CO₂ at 80 K from the CO₂ ice film is not affected by the molecular beam impinging onto the sample as the ice is many layers thick. The contribution of CO₂ evaporation from the film to the background pressure is unaltered when the beam hits the sample. Hence, our measurement solely detects if the effective sticking coefficient, i.e., the net condensation of CO₂ molecules onto the ice from the impinging beam, depends on the vibrational state population of that beam. It is only the latter that we alter by turning the laser ‘on’ and ‘off’. In the laser ‘on’ case, the heavily populated vibrationally excited state may adsorb or scatter more effectively in a direct sense than the vibrational ground state does. In case a vibrationally excited molecule adsorbs, it may also desorb again on a time scale much shorter than the experiment by V → T energy transfer. As adsorption is not activated, but desorption is, it is very unlikely that vibrational effects to both processes would be identical and cancel in an experiment that probes the effective sticking coefficient.

During the experiment, the laser is continuously modulated at 3 Hz. The measured partial pressure is directly related to the fraction of the molecular beam that is adsorbed. Therefore, we expect any effect of the vibrational excitation on the sticking probability to be visible in the measured partial pressure at the same modulation frequency. Inspection of the Fourier transform of the measured partial pressure allows us to quantify any effect of the vibrational excitation.

Using a combination of molecular beam techniques, state-resolved excitation, and RAIRS detection, we have investigated the possible influences of vibrational excitation on both the sticking of CO₂ on CO₂ ice and the structure of the resulting ice on a cryogenically cooled Cu sample. We find no evidence of an influence of the ν ₃ mode on either possible effect within our detection limits at an incident velocity of 587 m/s. With regards to the ice structure are our results consistent with the results of Ioppolo et al. (Ioppolo et al., 2022) for crystalline CO₂ ice. Neither vibrational excitation of molecules prior to impact nor when already in the ice seems to affect the ice structure significantly. With regards to the sticking probability we find that our detection limit supports that, in comparison to the measured average sticking probability for all states present in the beam, it is safe to assume that a single quantum in the ν ₃ mode is of no relevant influence to the effective sticking probability. This may result from a fortuitous cancellation of effects on the adsorption vs. direct scattering and desorption rates for the excited state, but this seems unlikely. The result, therefore, expands upon the conclusion from an earlier study that used state-resolved detection of CO₂ molecules desorbing from its ice: even vibrational energies exceeding the desorption energy seem not relevant to the growth or evaporation of CO₂ ice.