Reactive Desorption of CO Hydrogenation Products under Cold Pre-stellar Core Conditions

The total consumption of elemental carbon of the original species should be equal to the total amount of elemental carbon of the formed species, both in the solid state and, after reactive desorption, in the gas phase:

$$\begin{eqnarray}{N}_{\mathrm{consumed}}(C) & = & \bigtriangleup {N}_{\mathrm{solid}}(\mathrm{CO})=\bigtriangleup {N}_{\mathrm{solid}}({{\rm{H}}}_{2}\mathrm{CO})\\ & & +\bigtriangleup {N}_{\mathrm{solid}}({\mathrm{CH}}_{3}\mathrm{OH})+{N}_{\mathrm{RD}}({\rm{C}}),\end{eqnarray} \tag{ 1 }$$

where $\bigtriangleup {N}_{\mathrm{solid}}(\mathrm{CO})$ is the column density of CO consumption, $\bigtriangleup {N}_{\mathrm{solid}}({{\rm{H}}}_{2}\mathrm{CO})(\bigtriangleup {N}_{\mathrm{solid}}({\mathrm{CH}}_{3}\mathrm{OH}))$ is the net column density of the H₂CO (CH₃OH) formation, and ${N}_{\mathrm{RD}}({\rm{C}})$ is the column density of carbon-bearing species upon reactive desorption mechanism after H-atom bombardment. This means that an upper limit for ${N}_{\mathrm{RD}}({\rm{C}})$/N_consumed(C) can be derived by comparing accurately the total difference in CO abundance and H₂CO and CH₃OH abundances. In practice, Equation (1) is only an approximation, as it does not take into account other loss channels, such as the formation of other products than H₂CO and CH₃OH; reactive intermediates can recombine and form COMs larger than methanol, as recently shown by Chuang et al. (2016) and Fedoseev et al. (2017). The impact of this restriction is discussed in Section 3.

All experiments are performed using SURFRESIDE², an ultra-high vacuum setup fully optimized to study non-energetic atom addition reactions in interstellar ice analogs at cryogenic temperatures. Details of the design and operation of SURFRESIDE² are available from Ioppolo et al. (2013) and Fedoseev et al. (2016). The base pressure in the main chamber is ∼10⁻¹⁰ mbar, and the corresponding H₂O contamination through background gas accretion is negligible (<6.3 × 10¹⁰ molecule cm⁻² s⁻¹). A gold-plated copper substrate is mounted in the center of the main chamber and cooled by a recently upgraded closed-cycle helium cryostat that allows to vary the substrate temperature between 8 and 450 K. A silicon diode is used to monitor the temperature with 0.5 K absolute accuracy. It should be noted that the temperatures used in the present manuscript are about ∼2 K lower than the corresponding numbers used in Fuchs et al. (2009) new and more precise thermal sensors were installed on the substrate. Gaseous species, i.e., H₂CO obtained by thermal decomposition of paraformaldehyde (Sigma-Aldrich, 95%) at 60 °C–80 °C under vacuum, CH₃OH purified through multiple freeze-pump-thaw cycles and CO (Linde 2.0), are admitted into the UHV chamber through high-precision leak valves mounted under an angle of 22° from the substrate surface normal. The ices are grown with sub-monolayer precision with CO, H₂CO and CH₃OH deposition rates of ∼1.7 × 10¹², ∼1.4 × 10¹² and ∼2.2 × 10¹² molecules cm⁻² s⁻¹, respectively. The corresponding ice column densities for CO, H₂CO, and CH₃OH are ∼6.0 × 10¹⁵, ∼5.0 × 10¹⁵, and ∼8.0 × 10¹⁵ molecules cm⁻², respectively. The details of the exact deposition rate calculations are provided later.

After a preset ice thickness of the precursor species is realized, H-atoms are introduced using a Hydrogen Atom Beam Source (Tschersich 2000), mounted in a second UHV chamber. A nose-shape quartz pipe is placed along the path to efficiently quench and thermalize excited H-atoms and non-dissociated molecules through multiple collisions with the walls of the pipe. The used H-atom flux in this work amounts to ∼8 × 10¹² atoms cm⁻² s⁻¹ with an absolute uncertainty of <50% (Ioppolo et al. 2013) and H-atoms reach the surface under an incident angle of 45° from the substrate surface normal.

The ice composition is monitored in situ, before and during H-atom bombardment by Fourier Transform Reflection-Absorption InfraRed Spectroscopy in the range from 700 to 4000 cm⁻¹, with 1 cm⁻¹ resolution. The ice column density N in cm⁻² is derived from a modified Beer's law:

$$\begin{eqnarray}&&N=\displaystyle \frac{\mathrm{ln}10\cdot \int \mathrm{Abs}(\nu )d\nu }{A^{\prime} },\end{eqnarray} \tag{ 2 }$$

where $\mathrm{Abs}(\nu )$ is the band absorbance, dν is the wavenumber differential in cm⁻¹, and A' is the calibrated band strength in cm molecule⁻¹ in reflection mode. This RAIR band strength cannot be taken from the literature data available for the IR transmission spectroscopy, as the signals obtained in reflection are enhanced through substrate dipole couplings. Moreover, the determination of RAIR band strengths from transmission values cannot be realized by only compensating for different effective IR pathways in the ice. Therefore, a series of extra experiments (see Section 2.3) has been performed to determine the band strength for each of the involved species for our specific experimental settings by using the interference pattern of a HeNe laser that is reflected from the growing ice sample at 10 K (Baratta & Palumbo 1998; Brunetto et al. 2008; Fulvio et al. 2009; Bouilloud et al. 2015). For pure ice the absolute column density is calculated by the equation:

$$\begin{eqnarray}&&N=\displaystyle \frac{d\cdot \rho \cdot {N}_{a}}{M},\end{eqnarray} \tag{ 3 }$$

where d is the thickness of ice in nm, ρ is the density in ${\rm{g}}\,{\mathrm{cm}}^{-3}$, N_a is the Avogadro's constant (6.022 × 10²³ mol⁻¹), and M is the molar mass of the species. The ice thickness (${D}_{\mathrm{growing}}$) is experimentally determined by laser refractive interference,

$$\begin{eqnarray}&&{D}_{\mathrm{growing}}=k\times \displaystyle \frac{\lambda }{2n\cdot \cos ({\theta }_{f})},\end{eqnarray} \tag{ 4 }$$

where λ = 632.8 nm is the HeNe laser wavelength, n is the refractive index of a specific ice, ${\theta }_{f}\cong 3^\circ$ is the angle of refraction in the ice in degrees, and k is the number of involved fringes (Hollenberg & Dows 1961; Westley et al. 1998). The exact band-strength values may vary with changing ice composition, structure, and temperature. The reported absolute uncertainty of this method is 5%, a value that is largely determined by the density and refractive index uncertainty (Baratta & Palumbo 1998; Fulvio et al. 2009). Moreover, as the same technique is used to determine CO, H₂CO and CH₃OH band strengths, the relative error between values is expected to be lower. This makes it possible, in principle, to establish the obtained N_RD(C)/N_consumed(C) with high confidence, even when chemistry changes relative ice compositions.

The left-hand panel of Figures 2(a)–(c) shows the HeNe laser interference pattern that changes with time as the ice thickness increases upon deposition of CO, H₂CO, and CH₃OH onto the 10 K substrate. One single fringe is typically observed for the low-deposition rates used in this study. Ice density (ρ) and refractive index (n) may vary for different conditions (Bossa et al. 2015). It is important to note that the precise ice density in the amorphous phase is generally unknown for the species studied in this work. Here, the commonly accepted values in literature are taken as good approximate values. For CO, the ice density reported by Roux et al. (1980) at 20 K is used. Their deposition conditions may have been different from those in our work. For H₂CO, no density value has been reported in the solid phase, thus the value measured in liquid phase is employed (Weast & Astle 1985). For CH₃OH, the ice density is derived from the value available for its crystalline structure (Mate et al. 2009). Based on the density and refractive index values summarized in Bouilloud et al. (2015) 0.80 ± 0.01 g cm⁻³ and 1.25 ± 0.03 for CO, 0.81 ± 0.03 g cm⁻³ and 1.33 ± 0.04 for H₂CO, and 1.01 ± 0.03 g cm⁻³ and 1.33 ± 0.04 for CH₃OH–CO, H₂CO, and CH₃OH deposition rates are derived with values of ∼(1.3 ± 0.1) × 10¹⁴, ∼(4.7 ± 0.4) × 10¹³, and ∼(5.6 ± 0.5) × 10¹³ molecules cm⁻² s⁻¹. Note that these values differ in these experiments from the regular hydrogenation settings. The choice of the applied density and refractive index values is not ideal, but given the lack of more precise values, this is the best we can provide. If better values become available, the final ratio of N_RD(C)/N_consumed(C) can be easily re-evaluated.

Zoom In Zoom Out Reset image size

The corresponding CO, H₂CO, and CH₃OH plots, showing the column density upon ice deposition (horizontal axis) and the increasing IR absorption area (vertical axis) over time are presented in the right-hand panels of Figures 2(d)–(e). As explained in Section 2 (Equation (2)), from these plots the A', band-strength value, can be derived from molecule specific RAIR bands, following a linear fit. The fitted slopes for (d) CO (2142 cm⁻¹), (e) H₂CO (1497 cm⁻¹), and (f) CH₃OH (1030 cm⁻¹) yield absorption strength values A' of (5.2 ± 0.3) × 10⁻¹⁷, (3.2 ± 0.3) × 10⁻¹⁷, and (7.1 ± 0.6) × 10⁻¹⁷ cm molecule⁻¹, respectively. It should be noted that these values are only safe to use for identical experimental circumstances, i.e., identical IR incident angles, surface materials, etc.

Kerkhof et al. (1999) showed that the band strength of methanol in an ice mixture of CH₃OH:CO₂(1:1) and in pure CH₃OH ice does not vary. We checked this by performing a set of control experiments in which the CO, H₂CO, and CH₃OH band-strength values are compared with those derived in CO:H₂CO (1:1), CO:CH₃OH (1:1), and CO:H₂CO:CH₃OH (1:1:1) mixed ices. This is realized in the following way. At first, a constant deposition rate of CO molecules is used through one of the dosing lines. Then, sequentially, H₂CO and CH₃OH are introduced through other dosing lines growing an ice mixture during co-deposition. Each time a new species is allowed into the main chamber, and any changes in the CO absorbance integral area are carefully monitored. Then, in a similar way, changes in the H₂CO and CH₃OH absorbance integral area are measured. These changes are linearly correlated with their increasing column densities. From this, it is found that the CO, H₂CO, and CH₃OH band strengths are rather constant, with variations that do not exceed 2% in CO, CO:H₂CO (1:1), CO:CH₃OH (1:1), and CO:H₂CO:CH₃OH (1:1:1) ices. RAIR spectra are used only as long as signals are far from saturation, to guarantee a linear correlation with the corresponding column densities (Teolis et al. 2007).

Figure 3 presents RAIR difference spectra obtained after hydrogenation of pre-deposited CO ice for 180 minutes by an H-atom flux of ∼8 × 10¹² atoms cm⁻² s⁻¹ at (a) 10 K, (b) 12 K, and (c) 14 K, respectively. The negative peak at 2142 cm⁻¹ shows the consumption of the originally pre-deposited CO ice, and the positive peaks indicate the formation of two newly formed main products, i.e., H₂CO (1727, 1497, and 1250 cm⁻¹) and CH₃OH (1030 cm⁻¹). Very similar results have been reported and discussed extensively in previous studies of Watanabe & Kouchi (2002) and Fuchs et al. (2009). No entrapment of the formed intermediate radicals such as HCO, CH₃O, and CH₂OH is found, as these easily react with free H-atoms or other radical species.

Zoom In Zoom Out Reset image size

Figure 4 shows the changing column density obtained from the integrated RAIR absorption areas of the CO (2142 cm⁻¹), H₂CO (1497 cm⁻¹), and CH₃OH (1030 cm⁻¹) signals during H-atom accretion. The negative CO consumption peak at 10 K clearly saturates due to the limited penetration depth of H-atoms impacting on the surface. This observation is in agreement with previous findings (Fuchs et al. 2009), indicating that the CO consumption rate exhibits a strong temperature dependence; a lower temperature results in a higher CO conversion rate, fully in line with the proposed CO–H₂CO–CH₃OH reaction findings presented in the literature (Watanabe et al. 2004; Fuchs et al. 2009). This temperature dependence is explained by the different life (i.e., residence) time of H-atoms on the ice surface before desorbing from the ice surface or recombining to H₂ through interactions with other H-atoms. At 10 K and after 120 minutes of H-atom exposure, the CO depletion behavior starts slowing down, reaching a maximum value at approximately 2 × 10¹⁵ molecules cm⁻² (180 minutes), reflecting that only the few upper layers of CO molecules are accessible to accreting H-atoms.

Zoom In Zoom Out Reset image size

In Figure 5, the effective RD related column densities are shown for three different ice temperatures for 180 minutes H-atom bombardment. The increasing loss reflects that the sum of H₂CO and CH₃OH abundances is less than the amount of consumed CO molecules. At 180 minutes, the N_RD (C) value reaches ∼4.5 × 10¹⁴, ∼3.9 × 10¹⁴, and ∼1.7 × 10¹⁴ molecules cm⁻² for 10 K, 12 K, and 14 K, respectively. As discussed in Equation (1), ${N}_{\mathrm{RD}}({\rm{C}})$ reflects all products that are not recorded by the RAIRS technique in the solid state, i.e., this variable includes both the effect of reactive desorption and other elemental carbon loss channels, such as the formation of COMs with two, three, or more carbon atoms. However, the expected final abundances of these COMs cannot account for more than a few percent of the consumed CO molecules, as discussed in Fedoseev et al. (2017). Moreover, their CO stretching vibration modes are very similar to the CO stretching vibration modes used here for the CH₃OH column densities calculations; the smaller fraction of carbon atoms conserved in the formed larger COMs is actually already contributing to the CH₃OH column densities. Non-overlapping (and generally weaker) IR absorption features of COMs, i.e., around 860 cm⁻¹ for glycolaldehyde and ethylene glycol are not visible, yielding upper limits of the order of ∼5 × 10¹³ molecules cm⁻² s⁻¹, i.e., <3% of the consumed CO column density. It should be noted that previous studies on solid-state COM formation by Chuang et al. (2016), Chuang et al. (2017) and Fedoseev et al. (2017) were performed in co-deposition modus that are assumed to simulate interstellar ice conditions in a more representative way.

Zoom In Zoom Out Reset image size

In Figure 6, the obtained upper limits of ${N}_{\mathrm{RD}}({\rm{C}})$/N_consumed(C) ratios are summarized for two different time periods: from the start to the end (0–180 minutes) and from 60–180 minutes. These ratios must be strictly treated as upper limits of the reactive desorption fraction for the performed hydrogenation experiments. The abundances obtained at the end of the CO ice hydrogenation experiments after 180 minutes of H-atom bombardment are used for the calculations (black column in Figure 6), as here the largest amounts with reaction products are reached. This results in the best peak-to-noise ratios and therefore the lowest uncertainties. The resulting N_RD(C)/N_consumed(C) fractions are 0.23 ± 0.02, 0.22 ± 0.02 and 0.28 ± 0.05 for 10 K, 12 K, and 14 K, respectively. The averaged RD fraction (N_RD(C)/N_consumed(C)_{10–14 K}) amounts to 0.24 ± 0.02 for overall 180 minutes.

Zoom In Zoom Out Reset image size

A detailed analysis of the data presented in Figure 5 shows that ${N}_{\mathrm{RD}}({\rm{C}})$ does not decrease linearly with time; roughly during the first 60 minutes, the decrease is faster (for all three investigated temperatures). This may hint at startup effects, linked to structural weaknesses in the top surface layers because of any loosely bound CO molecules that get off much easier without the need of reactive desorption, for example, upon a collisional impact. For the period from 60 to 180 minutes, the ${N}_{\mathrm{RD}}({\rm{C}})$ decrease is much more linear. To exclude any perturbing contributions, therefore, the elemental carbon loss fraction is also calculated for the values derived between 60 and 180 minutes of H-atom bombardment (red column in Figure 6) resulting in somewhat lower upper values: 0.17 ± 0.05, 0.15 ± 0.02, and 0.16 ± 0.08 for 10 K, 12 K, and 14 K, respectively. The resulting averaged value (0.16 ± 0.03) amounts to ∼70% of the overall value. The latter value is likely more precise, as it does not include startup effects. However, to stay on the safe side, we report here the overall averaged RD fraction.

A very similar set of experiments is used to investigate reactive desorption upon H₂CO and CH₃OH hydrogenation. Figure 7 (left panel) presents the RAIR difference spectra obtained after hydrogenation of pre-deposited (a) H₂CO and (b) CH₃OH ice for an H-atom flux of ∼8 × 10¹² atoms cm⁻² s⁻¹ for 180 minutes at 10 K. In spectrum (a), the negative peaks at 1727, 1497, and 1250 cm⁻¹ show the depletion of the initially deposited H₂CO, while the positive peaks indicate the formation of newly formed products, i.e., CO (2142 cm⁻¹) and CH₃OH (1030 cm⁻¹). The formation of CO through H-atom abstraction reactions from H₂CO and of CH₃OH through H-atom addition reactions to H₂CO have been reported previously by Hidaka et al. (2004) and Chuang et al. (2016).

Zoom In Zoom Out Reset image size

In contrast, the CH₃OH hydrogenation experiments (Figure 7(b)) do not result in any efficient consumption of CH₃OH and the formation of hydrogenation products. Only a small negative CH₃OH peak (1050 cm⁻¹) is observed, while no secure identification of other carbon-bearing species can be made. Moreover, the consumption peak accumulates at the very beginning of the H-atom bombardment and as discussed above; this may be due to other processes than reactive desorption. As in the CO hydrogenation experiments, there is no spectroscopic evidence for radical entrapments due to high reactivity.

The evolving column density of CO, H₂CO, and CH₃OH during H-atom exposure with time is shown in the right-hand panel of Figure 7. The consumed N(H₂CO) saturates at ∼1 × 10¹⁵ molecules cm⁻² and is possibly limited by the H-atom penetration depth as discussed before. However, N(CH₃OH) saturates immediately, indicating the structural ice changes described before occurring on the surface comprising of CH₃OH molecules. As discussed for the CO hydrogenation experiments, the column density of the missing carbon during the first 60 minutes of H-atom bombardment likely includes other unwanted desorption processes. Following the same approach as described in Section 3.2, an extended calculation of the H₂CO hydrogenation experiment shows that the sum of the consumption of H₂CO and the production of CO and CH₃OH is very small; ∼2.0 × 10¹³ molecules cm⁻², i.e., 4% with respect to the CO hydrogenation experiments at 10 K. In the CH₃OH hydrogenation experiments, the ${N}_{\mathrm{RD}}({\rm{C}})$ originates only from the consumption of N(CH₃OH) due to a poor detection of H₂CO and results in 4% with respect to the CO hydrogenation experiments at 10 K. These small signals with large uncertainty complicate the determination of reactive desorption fractions. Future work is needed to provide more precise RD efficiencies for H₂CO and CH₃OH.