A NEW SOURCE OF CO₂ IN THE UNIVERSE: A PHOTOACTIVATED ELEY–RIDEAL SURFACE REACTION ON WATER ICES

The high abundance of CO₂ in ices in the interstellar medium (ISM) greatly exceeds that in the gas phase (van Dishoeck et al. 1996). This suggests that a solid phase CO₂ synthesis process is probably involved, but the CO₂ formation mechanism has not been identified. Recent observations indicate that CO₂ (ice) formation is linked to H₂O formation (Boogert et al. 2004). Theoretical and laboratory studies suggest an efficient CO + OH → CO₂ + H reaction in CO/H₂O mixed ices due to its low activation energy barrier (Garrod & Pauly 2011; Ioppolo et al. 2011; Noble et al. 2011; Oba et al. 2010; Ruffle & Herbst 2001a; Zins et al. 2011); however, the limited mobility of CO and OH species at low temperatures inhibits the reaction probability. Carbon monoxide is the second most abundant gas species after hydrogen in the ISM (Tielens 2005). It may well be the parent molecule for the synthesis of CO₂, as well as for many organic molecules in space (Ehrenfreund & Charnley 2000). Here we report an efficient surface reaction pathway for gas phase CO conversion to solid CO₂ on amorphous solid water (ASW) films or crystalline water ice films in the temperature range 71–82 K, activated by ice absorption of Lyα (10.2 eV) radiation. A direct Eley–Rideal (E-R) process (Eley 1948; Eley & Rideal 1940; Rideal 1938) that does not involve prior energetic accommodation of CO by the ice surface is observed, when the incident CO molecules interact with surface-bound OH radicals that are produced by water photodissociation. The surface reaction is kinetically separate from the previously studied solid phase CO+OH reaction. This finding of a significant new process for CO₂ production will influence kinetic models involving CO(g) consumption and CO₂(ice) formation in molecular clouds (Ruffle & Herbst 2001a; Whittet et al. 2010) or in the atmospheres of icy moons in the solar system (Teolis et al. 2010). More generally, this result suggests that similar E-R surface reactions on ice films which contain surface-bound radical species are likely to occur for other gas molecules in space.

Photochemical processes are known to drive molecular synthesis in the ISM, especially inside thin ice films condensed on dust grains (Gerakines et al. 1996; Ruffle & Herbst 2001b; van Dishoeck et al. 2006). During photodissociation of mixed CO/H₂O ices, hydroxyl radicals are produced that can interact with CO. This reaction has been studied in the laboratory by hydrogenation of CO/O₂ binary ices (Ioppolo et al. 2011; Noble et al. 2011) and by CO reaction with non-energetic OH radicals (Oba et al. 2010; Zins et al. 2011). In contrast to these studies of bulk ices containing condensed CO molecules, here we report that gas phase CO has a remarkable reactivity with photochemically generated OH radicals on the surface of H₂O ice. This finding reveals a significant new kinetic process for CO₂(ice) formation in space.

The classical E-R mechanism for a surface reaction involves, in its simplest form, the direct collision of an incoming species from the gas phase with a surface-bound species to produce a product molecule and was originally proposed as the mechanism for a metal-catalyzed surface reaction (Rideal 1938). This type of surface reaction occurs with reactive incoming species such as atomic H (Cheng et al. 1992). In space, it is proposed that molecular H₂ formation by H+H recombination in high temperature ISM regions occurs efficiently through the E-R mechanism (Lemaire et al. 2010). A modification of the classical direct E-R reaction involves an incoming species which partially accommodates on the surface as a mobile precursor and, in its subsequent diffusive motion, encounters an immobile reactive partner which then participates with the mobile species in a chemical reaction (Harris & Kasemo 1981). Such a process should exhibit temperature dependence due to the decreasing lifetime and steady-state coverage of the mobile species as temperature is raised. We present kinetic evidence that the CO+OH (on water ice) reaction occurs by a direct E-R process not involving CO surface mobility.

The apparatus has been described previously in detail (Yuan & Yates 2013). Briefly, the apparatus is a baked high vacuum system with a base pressure of ∼10⁻⁸ torr, pumped by turbomolecular and ion pumps. Figure 1 shows an experiment which demonstrates the photochemically induced reactivity toward CO(g) of a 20 nm thick ASW film. The H₂O film is deposited at 76 K on a 0.3 cm² disk of pressed KBr powder held on a tungsten grid in the high vacuum stainless steel cell of 1.9 × 10⁻³ m³ volume (Yuan & Yates 2013). In addition, crystalline ice films were made by heating deposited ASW to 170 K until the phase transition was completed and then cooling to 76 K for similar studies. Lyα radiation (2.7 × 10¹⁸ photons m⁻² s⁻¹) (Rajappan et al. 2010, 2011), passing through a LiF window and an internal aperture, is incident at 45° to the ice film. Transmission IR spectroscopy at a 45° incidence angle to the ice surface is performed during the irradiation to observe the formation of condensed CO₂ product. A quadrupole mass spectrometer senses the pressure of flowing CO(g) before, during and after irradiation. A mixture of C¹⁶O and C¹⁸O is employed with a mole fraction, X_C18O. We measured the C¹⁸O (m/z = 30) behavior and converted it to total CO pressure using measurements of X_C18O. The CO pressure is established over the ice film using a measured CO pumping speed of 2 × 10⁻⁴ m³ s⁻¹ (Yuan & Yates 2014). When the UV light is admitted, CO begins to be consumed immediately and a steady-state CO pressure is reached in about 300 s. When the UV light is extinguished, the CO pressure rises back to the original value in about 300 s. This behavior is observed at many CO pressures and the experiment closely resembles the well-known King–Wells method used in surface science to measure adsorption kinetics (King & Wells 1974). The change of CO pressure (ΔP_CO) is found to be similar (±20%) with both thick and thin ASW films, as well as on a crystalline ice film.

Zoom In Zoom Out Reset image size

Figure 2 shows the photochemical production of condensed CO₂ isotopomers during the experiment where X_C18O = X_C16O = 0.50 ± 0.05. The strong ν₃ modes for C¹⁶O₂ and C¹⁸O¹⁶O are observed. Their frequencies are close to the typical frequencies of pure CO₂ isotopomer ices (Broekhuizen et al. 2006; Escribano et al. 2013; Falk 1987), showing that the solid CO₂ product is formed. It is seen that the two CO₂ absorption bands grow monotonically. Plots of their integrated absorbance over time display the same slope, indicating that any ¹⁸O isotope effect for CO is negligible.

Zoom In Zoom Out Reset image size

3.1. Evidence for the Eley–Rideal-type Reaction

The rate of CO(g) consumption, R_CO (molecules m⁻² s⁻¹), can be found by measuring the change of CO pressure (ΔP_CO) during UV irradiation. The ΔP_CO is directly proportional to R_CO in a rapidly pumped system (Yates 1985), as shown in Equation (1):

$$\begin{equation} R_{\rm CO} = \frac{{\Delta P_{\rm CO} \cdot S_{\rm CO} }}{{A \cdot k_B T_g }}, \end{equation} \tag{ 1 }$$

where S_CO is the measured pumping speed of CO(g), A is the irradiated sample area, k_B is the Boltzmann constant, and T_g is the CO gas temperature. At any CO pressure, the CO consumption rate during irradiation is proportional to the accumulated surface OH coverage, N_OH. When the reaction reaches steady state, the surface-bound OH production rate is equal to its consumption rate, including the OH reaction rate with CO(g), and the OH recombination rate with H or OH. Molecular dynamics studies show a large fraction of photo-produced H desorbs from the ice surface because of its weak binding energy (Andersson & van Dishoeck 2008), leaving an OH-enriched surface. We therefore postulate that the OH + OH → H₂O₂ → H₂O + O reaction is the major process that consumes the excess OH (Oba et al. 2011). After the interruption of the irradiation, the OH species remaining on the surface continue to be consumed by CO(g) and by the recombination reactions, resulting in a decrease of R_CO. An empirical fit with second-order kinetics gives $N_{\rm OH}^{\rm SS} = (4 \pm 1) \times 10^{17}\,{\rm m}^{ - 2}$ from the data in six measurements at steady state, corresponding to a fractional surface OH coverage of $\theta _{\rm OH}^{\rm SS}$ = 0.05 ML (monolayer) at steady state.

The incident flux F_CO of CO(g) on the ice surface is calculated from the kinetic theory of gases using Equation (2):

$$\begin{equation} F_{\rm CO} = \frac{{P_{\rm CO}^0 }}{{(2\pi mk_B T_g)^{1/2} }}, \end{equation} \tag{ 2 }$$

where $P_{\rm CO}^0$ is the total pressure of CO(g) before irradiation, and m is the mass of the CO molecule. Therefore, the cross section, σ_CO, for CO(g) reaction can be obtained as d[CO]/dt = F_CO·σ_CO·N_OH = R_CO. The reaction cross section, σ_CO, is found to be 6 × 10⁻²⁰ m², which is of the order of a geometrical molecular cross section, consistent with a direct E-R-type process.

When CO molecules strike the ice surface, their average residence time can be calculated from Equation (3):

$$\begin{equation} t_s = \frac{1}{\nu }e^{E_{\rm des} /k_B T}, \end{equation} \tag{ 3 }$$

where ν is the desorption frequency factor ≈10¹² s⁻¹, E_des is the CO desorption energy from the H₂O ice surface, and T is the ice temperature. The CO desorption energy from H₂O ice has been investigated experimentally and theoretically (Al-Halabi et al. 2004a, 2004b; Collings et al. 2003). Because the CO coverage is low in our experimental conditions, we use E_des = 0.125 eV here (Karssemeijer et al. 2014). Therefore at T = 76 K, the residence time for CO(g) on the H₂O ice surface is ∼2 × 10⁻⁴ s. The fractional coverage of CO adsorbed on the ASW–H₂O ice surface at a given CO(g) pressure is θ_CO = (F_CO × t_s)/N_H2O, where N_H2O ≈ 9 × 10¹⁸ m⁻² is the surface density of H₂O adsorption sites based on the density (0.94 g cm⁻³; Jenniskens & Blake 1994) of ASW. Under our experimental conditions, the CO fractional coverage is only θ_CO = 1 × 10⁻⁶ ML, i.e. five orders of magnitude smaller than the steady-state fractional OH coverage. This low coverage is consistent with the absence of the adsorbed CO IR absorption band. Thus, it is improbable under the conditions of the experiment for the reaction to occur by a mechanism where CO adsorbs on the surface and translates to an OH species.

Figure 3 shows the temperature dependence of the reaction rate. As temperature decreases the CO consumption rate remains constant. This is additional evidence to support the E-R reaction mechanism. The measured activation energy of 0.006 eV, indicative of a zero activation energy within experimental error, implies that activated diffusion processes for CO is not kinetically significant at 76 K for the CO+OH reaction. The formation of the stable intermediate compound HOCO at unobservable concentrations is probably involved (Arasa et al. 2013; Ioppolo et al. 2011).

Zoom In Zoom Out Reset image size

It should be pointed out that in the ISM, competitive reactions may influence the overall results found here. The OH+OH reaction, the H+H reaction, the CO+H and CO+O reactions (Ioppolo et al. 2011; Noble et al. 2011; Oba et al. 2010; Roser et al. 2001) may interfere with the CO+OH reaction investigated here.

3.2. Advantage of the Method

A kinetic model was used to fit the data of Figure 1 as shown in Figure 4. The model considers the flux of incident light, the cross section for H₂O photodissociation, and the incoming flux of CO(g) in the system of known pumping speed to calculate the solid curve which is superimposed on the data for one measurement. It is found that the rate of the OH+OH reaction on the irradiated ice is mainly responsible for the shape of the P_CO curve, in agreement with the observation in Figure 1 that these curves are only weakly dependent on $P_{\rm CO}^0$. A schematic for the reaction of photo-produced OH with CO(g) via a direct E-R process is shown in the inset of Figure 4, along with the proposed second order recombination of the OH species.

Zoom In Zoom Out Reset image size

To study the kinetics of the CO consumption process, CO(g) + OH → CO₂(ice)+H, we divide the experimental CO pressure curves (as shown in Figures 1, 3, and 4) into two processes. Process I describes the CO pressure decay to steady state during irradiation and Process II describes the CO pressure recovery after irradiation ends.

As mentioned earlier, the CO(g) consumption rate, R_CO, is proportional to the initial CO pressure and the surface OH coverage, N_OH:

$$\begin{equation} R_{\rm CO} = \frac{{\Delta P_{\rm CO} \cdot S_{\rm CO} }}{{A \cdot k_B T_g }} = k_1 P_{\rm CO} N_{\rm OH}. \end{equation} \tag{ 4 }$$

Because of the observed weak dependence of R_CO on CO pressure, P_CO, and the zero activation barrier, we propose that the CO(g)+OH reaction only consumes a small portion of OH species on the surface. The majority of surface OH species are consumed by the competitive OH+OH second-order recombination process. This assumption is confirmed by the following analysis.

In Process II, we assume the OH consumption rate is simply determined by the CO(g)+OH reaction rate and by the OH+OH recombination rate, despite other complex processes that could occur in an irradiated water ice:

$$\begin{equation} - \frac{{dN_{\rm OH} }}{{dt}} = R_{\rm CO} + R_{\rm OH + OH} = k_1 P_{\rm CO} N_{\rm OH} + k_2 (N_{\rm OH})^2. \end{equation} \tag{ 5 }$$

Because the CO(g)+OH reaction rate is relatively slow, the OH consumption rate is approximately a second-order kinetic process in N_OH. The empirical second-order kinetic fits of 1/R_CO versus time from the experimental data are consistent for a range of pressures, as seen in Figure 5. We use this simplified second-order kinetics assumption in our further modeling.

Zoom In Zoom Out Reset image size

By integrating Equation (5),

$$\begin{equation} - \int_{N_{\rm OH}^{\rm SS} }^{N_{\rm OH} } {\frac{{{dN}_{\rm OH} }}{{k_1 P_{\rm CO} N_{\rm OH} + k_2 (N_{\rm OH})^2 }} = \int_0^t {dt},} \end{equation} \tag{ 6 }$$

where $N_{\rm OH}^{\rm SS}$ is the OH coverage during irradiation at steady state, we get

$$\begin{equation} N_{\rm OH} = \frac{{k_1 P_{\rm CO} N_{\rm OH}^{\rm SS} }}{{k_1 P_{\rm CO} e^{k_1 P_{\rm CO} t} + k_2 N_{\rm OH}^{\rm SS} (e^{k_1 P_{\rm CO} t} - 1)}}. \end{equation} \tag{ 7 }$$

In Process I, the steady-state OH consumption rate is equilibrated with its production rate by irradiation. The kinetic equation is

$$\begin{equation} \frac{{dN_{\rm OH} }}{{dt}} = F_{h\nu } \sigma _{\rm H_2 O} N_{\rm H_2 O} - k_1 P_{\rm CO} N_{\rm OH} - k_2 (N_{\rm OH})^2, \end{equation} \tag{ 8 }$$

where F_hv is the flux of photons, σ_H2O = 1 × 10⁻²¹ m² is the Lyα photodissociation cross section of H₂O ice (van Dishoeck et al. 2006), and N_H2O is the surface density of H₂O molecules. Here we assume that only the first layer of water ice contributes to the surface OH production. Integrating Equation (8)

$$\begin{equation} \int_0^{N_{\rm OH} } {\frac{{{dN}_{\rm OH} }}{{F_{h\nu } \sigma _{\rm H_2 O} N_{\rm H_2 O} - k_1 P_{\rm CO} N_{\rm OH} - k_2 (N_{\rm OH})^2 }} = } \int_0^t {dt}, \end{equation} \tag{ 9 }$$

we get

$$\begin{eqnarray} &&{{N_{\rm OH} = \displaystyle\frac{{2c\left({1 - e^{\sqrt {k_1^2 P_{\rm CO}^2 + 4k_2 c \cdot t} } } \right)}}{{k_1 P_{\rm CO} \left({1 - e^{\sqrt {k_1^2 P_{\rm CO}^2 + 4k_2 c \cdot t} } } \right) - \sqrt {k_1^2 P_{\rm CO}^2 + 4k_2 c\left({1 + e^{\sqrt {k_1^2 P_{\rm CO}^2 + 4k_2 c \cdot t} } } \right)} }},}}\nonumber\\ \end{eqnarray} \tag{ 10 }$$

where $c = F_{h\nu } \sigma _{\rm H_2 O} N_{\rm H_2 O} \approx 2 \times 10^{16}\,{\rm m}^{ - 2}\,{\rm s}^{ - 1}$.

At steady state,

$$\begin{equation} F_{hv} \sigma _{\rm H_2 O} N_{\rm H_2 O} = k_1 P_{\rm CO} N_{\rm OH}^{\rm SS} + k_2 \left(N_{\rm OH}^{\rm SS}\right)^2. \end{equation} \tag{ 11 }$$

Here we assume a steady-state surface OH concentration, $N_{\rm OH}^{\rm SS}$, so that the rate constants k₁ and k₂ can be calculated from Equations (4) and (11). These rate constants are used in Equations (7) and (10) to model the curves to best fit the experimental data.

The value of $N_{\rm OH}^{\rm SS}$ can be estimated from the empirical second-order kinetics fit in Figure 5, and then be slightly adjusted to best fit the experimental data. An optimum $N_{\rm OH}^{\rm SS} = (4 \pm 1) \times 10^{17}\,{\rm m}^{ - 2}$ is obtained from the data of six measurements. The fitting result is shown in Figure 4 for one experiment. The modeled curve for Process II fits the experimental data well, while the curve for Process I has a small deviation from the data when P_CO approaches the steady state. This may be rationalized by postulating a small contribution from slow diffusion of OH from the underlying layers of the water ice (Andersson & van Dishoeck 2008). Since the Lyα light penetrates into the ice, the diffusing OH species produced in the bulk may contribute to the surface OH concentration, causing a slightly increasing CO consumption rate as one moves to longer times in the plateau region of Figures 1 and 4.

With the calculated values of k₁ and k₂, the OH consumption rate by the CO(g)+OH reaction and by the OH+OH recombination are 3% and 97% of the total OH production rate by irradiation at steady state, respectively. Therefore our previous assumptions that the CO(g)+OH reaction rate is relatively small and that the OH consumption rate can be simplified as a second-order kinetic process are reasonable.

In summary we have observed the reaction between photochemically produced surface-bound OH species and incident CO(g) molecules at water ice surfaces between 71 and 82 K. The reaction is found to occur by a direct Eley-Rideal mechanism where a CO(g) molecule collides and reacts directly with a surface OH species to produce CO₂(ice). This observation may explain the origin of some of the CO₂(ice) observed in the ISM as well as the CO consumption in the atmospheres of icy moons. Our experiments demonstrate a new process to convert CO(g) to CO₂ by reaction with surface OH species on astronomical ices.

C.Y. acknowledges full fellowship support from the Department of Chemistry, University of Virginia. I.R.C. acknowledges funding from a Fulbright Science and Innovation Graduate Award. We also acknowledge helpful conversations with Professor Eric Herbst and very helpful comments from the reviewer.