Collision-induced desorption of CO from Ru(0001) by hyperthermal argon and nitrogen

Highlights

• Measurement of collision-induced desorption (CID) of CO from Ru(0001) by hyperthermal argon and nitrogen beams.

• Determination of CID cross sections for CO removal.

• Quantification of the percentage of CO removable by argon and nitrogen beams.

• Evidence for N-adatom-induced weakening of the CO bond energy.

Abstract

Collision-induced desorption of CO from Ru(0001) by hyperthermal (5–9 eV) effusive beams of Ar and N + N₂ has been studied at a sample temperature of 400 K. Prompt desorption occurs with cross sections on the order of 4 Å². Based on post-exposure thermal desorption measurements, ~ 1/3 of the initial CO coverage cannot be desorbed by Ar on the time scale of the current experiments. In contrast, exposure to the mixed N + N₂ beam appears to remove all CO from the irradiated region. This is attributed to a lowering of the CO binding energy by adsorbed N-atoms. While there is no evidence of a large influence of surface diffusion on the time scale of these exposure, desorption simulations suggest that local diffusion in the periphery of the exposed region influences the measured decay.

Introduction

There is a long standing interest in collisions of gas-phase atoms and molecules with solid surfaces. This was already reviewed by Arumainayagam and Madix in 1991 [1], and frequently since then (see, e.g., [2], [3], [4], [5], [6], [7]). Most processes involving gas-phase species at solid surfaces (e.g., sticking and chemical reactions) occur with thermal or thermalized particles, but in some cases the translational or internal energy of incident atoms and molecules may drive surface processes before equilibration can occur [8], [9], [10], [11]. When incident species have hyperthermal energies (on the order of several eVs), the excess energy can enable processes that are not possible for thermalized particles [5], [12]. Examples include prompt Eley–Rideal type reactions, where the energy of the incident particle is not dissipated to the surface during the reaction [13], [14], [15], [16], [17] and collision-induced dissociative chemisorption [18], [19]. Another important case is collision-induced desorption (CID), which involves adsorbed atoms or molecules being ejected by incident hyperthermal atoms. This field has been reviewed by Asscher and Zeiri in 2003 [5]. CID works most easily for weakly bound adsorbates. In several studies, a threshold that depends on the binding energy of the adsorbate has been observed, indicating a direct recoil mechanism [5], [20].

The first experimental observation of CID was for Ar-induced desorption of CH₄ and CO from Ni(111) [21], [22]. Since then, extensive studies have been undertaken, although primarily on weakly bound molecular systems. CID has been studied for N₂ from Ru(0001) [23], [24], [25] and for chemisorbed O₂ from Ag(001) and Pt(111) [26], [27], [28], [29], [30]. These systems are described in terms of direct momentum transfer from incident rare gas atom to the adsorbed molecule, which desorbs after multiple collisions. The desorption processes are first order. In the case of CO/H/Ni(100), weakly bound CO is desorbed by Xe impact, but strongly bound CO remains on the surface [31], [32], [33], [34]. CID of chemisorbed NO on Pt(111) by hyperthermal Kr and Xe (2–4 eV) has been studied by Velic and Levis [20].

In the case of the strongly bound CO-Ru(0001) system, a desorption threshold on the order of 3–6 eV, depending on the effective surface mass during the collision, can be expected. In a previous study, we [AWK and MAG] have investigated the CID of CO from Ru(0001) with Ar atoms of ~ 6 eV average energy [35]. The most salient features of that work are summarized here since the current study represents a direct follow up.

The previous experiment was carried out with a pulsed Ar beam using a 2 slit chopper giving rise to a duty cycle of the beam of 0.5%. Prompt CID, ejecting CO with kinetic energies of around 1 eV, was observed from the initially saturated (~ 0.66 ML) overlayer for a beam incidence angle of 60^o. The desorption rate for prompt CID was CO coverage dependent. Initially, it increases with increasing exposure time and subsequently decreases again. For the starting coverage, no desorption was detected. The first CO CID signal was measured after an Ar fluence of ≥ 4.8 × 10¹⁵ atoms ⋅ cm^− 2 (equivalent to ~ 3 ML) had impinged on the surface. Residual CO remained in the exposed spot even after a cumulative exposure equivalent to 20 ML of Ar. Either not all CO in the beam spot can be removed by the Ar beam, or a non-negligible amount of CO was diffusing in from unexposed areas of the surface. Due to the long exposure times of these measurements (~ 1 h), residual CO remaining in the beam spot cannot be differentiated from potential lateral diffusion, and no attempt to do so is made in the manuscript. Consequently, the estimation of the amount of CO in the beam spot after exposures was, necessarily, very broad.

The CID process was, at least in part, instantaneous because only non-thermal CO was detected in the time-of-flight (TOF) experiments. We note however that the experiment was not very sensitive to slowly desorbing CO since such events would be spread over a large time window. In addition, the signal-to-noise ratio of the TOF data is low, so the study may not have been sensitive to all channels. The physical picture proposed is that the saturated CO overlayer can efficiently reflect the incident fast Ar atoms. CID becomes more effective as defect creation and gradual CO depletion allows for easier penetration of the overlayer. From the angular distribution of desorbing CO, it was concluded that desorption is not due to primary recoils or direct kickoff. Otherwise, the data would violate the threshold law for prompt desorption. Instead, lateral interactions in the CO layer must lead to desorption. Subsequently, once sufficient CO had been removed, the prompt CID process becomes ineffective relative to lateral displacement of CO by the incident atoms. No value for the cross section was derived from the chopped beam experiments. The effective mass for Ar scattering was equivalent to that of 1 Ru atom. In this case, the threshold for CID is around 6 eV, meaning that the majority of the beam would not contribute [35]. However, lateral energy transfer in the CO overlayer can remove this threshold.

The adsorbate structures and desorption behavior of CO on Ru(0001) have been extensively studied (see, for example, [36], [37], [38], [39]). The maximum CO coverage that can be adsorbed on Ru(0001) is temperature dependent. Broadly speaking, the thermal desorption spectrum has two distinct features although there is additional complexity (i.e., multiple components) within both of these features [37], [39]. The higher temperature desorption feature centered at ~ 475 K is typically associated with (√ 3 × √ 3)R30° surface structure and is fully saturated at a coverage of ~ 0.38 ML. The lower temperature feature (~ 400 K) saturates at ~ 0.66 ML. At T_s = 400 K, a lattice gas is predicted for coverages up to ~ 0.28 ML and (√ 3 × √ 3)R30° structures in the coverage range 0.28–0.42 ML [39]. The binding energies as a function of coverage reflect the characteristics of the desorption spectrum [37]. For coverages up to ~ 0.33 ML, the binding energy increases from ~ 1.66 eV to ~ 1.81 eV. This represents the filling of the higher temperature desorption peak. Above 0.33 ML, there is an abrupt transition to a ~ 1.24 eV binding energy with a further shift down to ~ 1.14 eV with increasing coverage.

In this paper, we present an analysis of CO CID from Ru(0001) using full beam exposures to complement the earlier work. A surface temperature (T_s) of 400 K was used to ensure saturation of the higher temperature desorption states while maintaining a low population in the less stable, lower temperature ones. Lateral diffusion is not a dominating effect in the current exposures but appears to have an influence on the periphery of the beam spot. A cross section for CID that is a sizable fraction of the unit cell is derived. Nonetheless, roughly 1/3 of the initial CO coverage remains in the beam spot after exposure, which is attributed primarily to “unremovable” CO.