SPUTTERING AND MOLECULAR SYNTHESIS INDUCED BY 100 keV PROTONS IN CONDENSED CO₂ AND RELEVANCE TO THE OUTER SOLAR SYSTEM

Figure 1 shows the MS measurements of the flux of CO, O₂, CO₂, and O sputtered from CO₂ films and their sum labeled ∑ at 25 and 50 K versus irradiation fluence F. Background signals for each mass were obtained with the proton beam blocked, and have been subtracted for each species. Further, we have corrected the flux for species-dependent MS sensitivity and fragmentation in the ion source (Bahr & Baragiola 2012). The accuracy of the fluxes is ∼25%.

Zoom In Zoom Out Reset image size

First, it is interesting to note the small CO₂ flux, compared to CO and O₂ at both temperatures, despite the films being pure CO₂ prior to irradiation. At 50 K, we observe the flux of CO and O₂ to increase with F, peaking at ∼5 × 10¹⁴ H⁺ cm⁻² and then decreasing at higher fluence. The CO₂ signal does not change with fluence. The maximum CO signal is ∼500 times larger than that of CO₂ at 50 K. At this temperature, we did not observe sputtered C and O atoms above the level from fragmentation of molecules in the MS.

At 25 K we could measure the direct O flux coming from the ice film at 25 K. Though the sum of the sputtered flux remained fairly constant at 25 K, variation in flux of individual species can be observed. For instance, the O/O₂ decreases ∼75-fold by 5 × 10¹⁵ H⁺ cm⁻², as the sputtered O₂ signal increases as O₂ builds up in the ice. Similarly, the O/CO signal also decreases by a factor of 15. The CO signal is ∼10 times larger than the CO₂ signal at 25 K.

Figure 2 shows the SY of initially pure CO₂ films with increasing fluence at 25 and 50 K, measured using the QCM. At 25 K, the yield varies over a small range, between 10 and 20 CO₂ eq. per ion. In contrast, the yield shows a strong dependence on fluence at 50 K. The yield is similar to that at 25 K at low fluences. It sharply increases to peak at ∼2400 by 5 × 10¹⁴ H⁺ cm⁻² and then decreases at higher fluence.

Zoom In Zoom Out Reset image size

There is good agreement between the MS (Figure 1) and the QCM measurements (Figure 2) at 25 K, as indicated by the overlap of ∑ (scaled) and the SY measurements. At 50 K, a similar trend exists between the scaled ∑ and SY, though disagreements are present at higher F, as SY decreases faster than ∑ (scaled) above 5 × 10¹⁴ H⁺ cm⁻². At the highest fluence, ∑ is nearly three times larger than SY. This may be due to a variation of the angular distribution of the sputtered flux with ion fluence and, therefore, of the fraction measured by the MS.

The top panel in Figure 3 shows changes in the CO₂ absorptions assigned to various modes of vibration as a result of ion irradiation at 25 K. The weak absorptions at ∼2.75 μm correspond to the ν₁ + ν₃ (3708 cm⁻¹) and 2ν₂ + ν₃ (3600 cm⁻¹) combination modes. The stronger absorption at ∼4.25 μm is due to the asymmetric C–O stretch mode (ν₃), while that at ∼15 μm is due to the bending mode (ν₂) (Sandford & Allamandola 1990). The stronger ν₂ and ν₃ bands exhibit LO–TO splitting as previously observed for solid CO₂ films (Baratta & Palumbo 1998).

Zoom In Zoom Out Reset image size

In the bottom panel, we show the appearance of new absorption features with increasing fluence due to synthesis of CO, CO₃, and O₃ during irradiation. The forbidden absorption band of O₂ at 1550 cm⁻¹ is not seen above the noise, but we detect O₂ in the sputtered flux (Figure 2). The residual CO₂ becomes diluted as more CO and O₂ are formed with increasing fluence. Such mixing of CO₂ in other species leads to the convergence of the LO–TO bands (Raut et al. 2012) as seen in top panel of Figure 3.

Figure 4 compares the fluence dependence of the band areas of four infrared-active species at 25 and 50 K. For CO₂, we chose to integrate the ν₁ + ν₃ band, since the stronger absorptions (ν₃, ν₂) are complicated by their LO–TO splitting and the merging of these features as the film evolves from pure CO₂ to a mixed one. These effects are absent in the weaker combination bands (see Figure 3) where the real part of the refractive index, n, in the absorption region >1 (Baratta & Palumbo 1998). We integrated the ν₁ absorption band for CO and CO₃ and the ν₃ band for O₃. The CO₂ band area decreases while those of CO, O₃, and CO₃ initially increase at both temperatures and then decrease at higher fluence. At 50 K, the CO band area is smaller than at 25 K, beginning at ∼10¹⁴ H⁺ cm⁻². Similarly, the O₃ band area is smaller at 25 K. This is consistent with our observation of sputtering where CO and O₂ are the dominant species. The band areas decrease faster at higher F due to high sputtering at 50 K.

Zoom In Zoom Out Reset image size

After irradiating to a fluence of ∼1 × 10¹⁵ H⁺ cm⁻² at 25 K, we desorbed the film by heating at 1 K minute⁻¹, while continuously monitoring the mass with the QCM and the ejected molecules with the MS (Figure 6). We do not show small C and O signals at this temperature since they are, within errors, due to fragmentation of CO and O₂ in the ion source of the MS. The desorption peak at ∼50 K is mainly due to CO and O₂. The second desorption peak at 90 K is due to residual CO₂ in the film, although it is possible that CO₂ on the unirradiated regions of the substrate contributes to the MS signal at 90 K. The absence of significant O₂ at 90 K confirms that the dominant fragmentation route is CO₂ → CO + O. We observe a small desorption peak between 60 and 70 K, which we attribute to the release of minor constituents of the film, O₃ and CO₃.

The inelastic energy is deposited by the projectiles in a track of ionizations and electronic excitations. It is then mostly transferred to motion of the CO₂ molecule as a whole or to dissociation fragments. After electron–ion recombination, the outcome of the energy deposition by an ion is a distribution of neutral dissociation fragments, the main initial products being CO, C, O, and O₂. The main dissociation channel for CO₂ is

$$\begin{equation} {\rm H}^ + + {\rm CO}_2 \to {\rm CO} + {\rm O}. \end{equation} \tag{ 1 }$$

In a dense track, the radicals and molecules can react. An example is the recombination reaction:

$$\begin{equation} {\rm CO} + {\rm O} + {\rm M} \to {\rm CO}_2 + {\rm M}, \end{equation} \tag{ 2 }$$

where M represents the matrix. A fraction of the dissociated fragments can escape the matrix cage and be trapped away from the ion track. In the bulk, the O atoms can react with additional CO₂ to form CO₃ (Moll et al. 1966) or react with another O to form O₂. Ozone synthesis can proceed via additive reactions, O + O₂ → O₃. Within a few monolayers from the surface, electronic excitations leading to atomic and molecular motion can result in sputtering, i.e., the ejection of the atomic and molecular species from the ice (Johnson & Schou 1993). The fluxes of these sputtered fragments are shown in Figure 1.

As irradiation progresses, the film composition changes with new species appearing. These can be converted by an incoming ion:

$$\begin{equation} {\rm H}^ + + {\rm CO} + {\rm CO} \to {\rm CO}^* +\, {\rm CO} \to {\rm CO}_2 + {\rm C} \end{equation} \tag{ 3 }$$

$$\begin{equation} {\rm H}^ + + {\rm O}_2 \to 2{\rm O} \end{equation} \tag{ 4 }$$

$$\begin{equation} {\rm H}^ + + {\rm O}_3 \to {\rm O}_2 + {\rm O} \end{equation} \tag{ 5 }$$

$$\begin{equation} {\rm H}^ + + {\rm CO}_3 \to {\rm CO} + {\rm O}_2 \to {\rm C} + {\rm O}_3 \end{equation} \tag{ 6 }$$

These reactions account for the species observed via infrared (Figure 3) and mass spectroscopy (Figure 1).

We observed a dramatic change in behavior of the SY with fluence and temperature, as shown in Figure 5. SY changes by <50% from the average of 15 per ion with fluence at 25 K. CO₂ is converted mainly into CO and O₂, which accumulate in the film with increasing F. These new species, though more volatile than CO₂, are stable within the film. The erosion of the film at 25 K is due to loss of molecules only a few monolayers from the ice–vacuum interface.

Zoom In Zoom Out Reset image size

In contrast, at 50 K, the remarkable increase of the SY with F is due to the escape of the CO and O₂ from the penetration depth of the ion, which in our case exceeds the film thickness. As seen in Figure 6, CO and O₂ thermally desorb when the temperature exceeds 50 K. The energy deposited by the ion first converts CO₂ into CO and O₂. These molecules can diffuse along the "hot" ion track, across the thickness of the film, and escape the ice, causing substantial mass loss that depletes 85%–90% of the initial film at the highest F.

Zoom In Zoom Out Reset image size

We estimate column densities of CO₂, CO, and O₂ and their fluence dependence for the case where the film is irradiated at 25 K (Figure 5). We find that the amount of CO₂ drops by half at a fluence of ∼1 × 10¹⁵ H⁺ cm⁻². The initial slope of the fit to the CO₂ data gives a destruction yield of 600 ± 55 CO₂ per H⁺. Each 100 keV proton deposits, on average, 22.7 keV as it traverses the 0.25 μm thick CO₂ film (Ziegler & Biersack 2013). Using this energy, we obtain the radiation yield G(−CO₂) = 2.6 ± 0.3, defined as the number of CO₂ destroyed per 100 eV of energy absorbed. This value is nearly five times larger than those for destruction of CO₂ mixed in water ice (1:1) and irradiated with 0.8 MeV H+ and Lyα at 18 K (Gerakines et al. 2000), which suggests that the matrix plays an important role in radiation stability of the CO₂.

The decomposition of CO₂ leads to the formation of CO and O₂. We derive the CO and O₂ column densities at 25 K using conservation arguments. At an average SY of 15 (Figure 2), only 3% of the initial CO₂ amount is removed by ∼1 × 10¹⁵ H⁺ cm⁻². Ignoring this small mass loss and the presence of minor radiolytic products such as CO₃ and O₃, we estimate the amount of CO and O₂ based on conservation of the number of C and O atoms in the film. In a previous study of CO₂ film irradiated with 5 keV electrons, only 2% of the O atoms generated from CO₂ destruction was estimated to react with additional CO₂ to form CO₃ (Bennett et al. 2004). The destruction of CO₂ should create CO and O₂ as per 2CO₂ → 2CO + O₂. Therefore, CO (F) = CO₂ (0) − CO₂ (F), and therefore, O₂ (F) = 1/2 CO (F). For both CO and O₂, these are upper-limit estimates. The errors on the η of CO and O₃ are due to the fluence-dependent amounts of CO₃ and O₃, which is estimated from the small desorption peak between 60 and 65 K in Figure 6.

Condensed CO₂ has been irradiated previously with 1.5 MeV He⁺ ions at 10 and 70 K (Brown 1982; Johnson et al. 1983). While they reported ejection of CO, O₂, and CO₂ that increased in steps with temperature, they did not quantitatively determine the contribution of each species to the total flux. As we will show in the discussion, this is important in astronomical applications. Also, while they observed an increase of the SY with fluence at 25 K similar to ours, their 50% increase at 50 K is much lower than the tenfold increase we measured at this temperature.

A recent study where C¹⁸O₂ films were irradiated with 46 MeV Ni¹¹⁺ reports a SY of 40,000 molecules per ion at 13 K (Seperuelo Duarte et al. 2009). In addition to CO, CO₃, and O₃, this study also identified the formation of C₃ via infrared absorption at 2039 cm⁻¹. Irradiation of solid CO₂ with 5 keV D₂⁺ ions also forms the same species, with the exception of C₃ (Ennis et al. 2011).

We also observed a previously unknown quadratic dependence of the SY for solid CO₂ at low temperatures (<25 K) on ε, the energy deposited by the projectile per monolayer in electronic excitations and ionizations. To place the data in context, we compare it to a compilation of SYs from different reports (Brown 1982; Seperuelo Duarte et al. 2009) using ε from the tables in SRIM 2013 (Ziegler & Biersack 2013). Figure 7 shows the dependence of the CO₂ yield on a dimensionless parameter, ε/U, where U is the sublimation energy, along with measurements for other condensed gases (Johnson 1996). We note that the quadratic dependence for CO₂ extends to the energies of cosmic rays (Seperuelo Duarte et al. 2009).

Zoom In Zoom Out Reset image size

CO₂ and O₂ have recently been detected in the exospheres of Saturnian moons Rhea and Dione. The abundance of O₂ is roughly twice that of CO₂ at Rhea (Teolis et al. 2010), while they exist in nearly equal proportions at Dione (Teolis & Waite 2012). Previously, a CO₂ exosphere was detected at Callisto (Carlson 1999) and O₂ exospheres at Europa and Ganymede (Hall et al. 1998). Teolis et al. (2010, and online supplementary material) explained the O₂ abundance in Rhea's exosphere using laboratory measurements of radiolysis of water ice (Teolis et al. 2009) and measurements of ion and electron fluxes taken by Cassini (Paranicas et al. 2012; Teolis et al. 2010).

In contrast, the source of the observed CO₂ is uncertain. Cruikshank et al. (2010) and Stephan et al. (2012) have suggested that, in the absence of a known endogenic source, the exospheric CO₂ is generated by sputtering of surficial CO₂ by impact from magnetospheric ions. While Cassini's infrared spectrometer indicates that Rhea's surface is pure water ice (99%) with only trace amounts of CO₂ (Stephan et al. 2012), it is possible that surface concentrations are much higher due to segregation, since CO₂ is more volatile than water. However, our results show that sputtering should produce about a thousand times more CO than CO₂, while CO has not been detected by Teolis et al. (2010). We thus conclude that other processes, such as outgassing of endogenic CO₂ or seasonal sublimation of condensed CO₂, rather than sputtering, are the sources of the tenuous CO₂ exosphere.