FIRST INFRARED BAND STRENGTHS FOR AMORPHOUS CO₂, AN OVERLOOKED COMPONENT OF INTERSTELLAR ICES

The analysis of astronomical spectra of interstellar and planetary ices continues to rely on laboratory spectra for the identification of molecules and ions. Such reference spectra must be accompanied by an unambiguous knowledge of ice composition, temperature, phase, history, and, to be applicable to the determination of molecular abundances, a quantitative measure of spectral intensity. As a specific case, recently we presented (Gerakines & Hudson 2015) new infrared (IR) results on solid methane (CH₄), an ice that has been studied spectroscopically for many years. We showed that the relatively few IR spectra published exhibit a lab-to-lab consistency, yet most if not all spectra reported are for a wholly or partially crystalline ice, and not the amorphous phase that might be expected near 10 K in the interstellar medium (ISM).

We now find that a similar study of frozen CO₂ is needed, motivated in part by the much greater interstellar abundance of CO₂ ices compared to CH₄ ices. In this paper we report new IR spectroscopic results on frozen CO₂, one of the more-common interstellar and planetary icy solids. Despite decades of attention paid to this molecular ice, and a voluminous associated literature, our new work appears to be the first to quantitatively characterize the IR spectra of carbon dioxide's amorphous phase.

Relatively few papers have addressed explicitly the IR spectra of amorphous CO₂, the first apparently being that of Falk (1987), who reported IR peaks at 2342.3 and 661.0 ${\mathrm{cm}}^{-1}$. The band shapes of these features lacked the sub-structure known for crystalline ices, and warming of Falk's samples generated spectra that were said, but not shown, to be for the crystalline material. In contrast to his work stands that of Escribano et al. (2013) who reported amorphous-CO₂ spectra, but with IR peak positions at 2328 and 655 ${\mathrm{cm}}^{-1}$, substantially different from those of Falk (1987) The thickness of the smallest ices of the former was on the order of a few nanometers compared to a few micrometers for the latter, raising the possibility that some of the spectral variations were due to different ice thicknesses. Recently, Sivaraman et al. (2013) have reported experiments in which amorphous CO₂ was made at 30 K. Their peak positions for the strongest amorphous-CO₂ spectral features were listed (spectra not shown) at 2342.0, 661.6, and 655.2 ${\mathrm{cm}}^{-1}$ which differ from those of both Falk (1987) and Escribano et al. (2013). Therefore, at present three sets of peak positions are in the literature for the same ice phase. This alone is a hindrance to applying such data to astronomical problems, but in addition none of these papers reported IR band intensities.

Adding to these uncertainties are questions surrounding the published CO₂-ice optical constants and associated spectra. Figure 1 compares IR transmission spectra for the ${\nu }_{3}$ and ${\nu }_{2}$ fundamentals of a CO₂ ice near 10 K as calculated for a thickness of 0.1 μm using the optical constants reported by four different research groups. Inspection of the traces in Figure 1 shows that there is both quantitative and qualitative disagreement in the results. Especially conspicuous is the splitting of the ${\nu }_{2}$ band (∼660 ${\mathrm{cm}}^{-1}$, 15.2 μm), a CO₂ vibration long and often used to study conditions and chemical evolution in cold interstellar regions (e.g., d'Hendecourt & Jourdain de Muizon 1989; Pontoppidan et al. 2008; Cook et al. 2011). Over 60 years ago, Osberg & Hornig (1952) showed that such a splitting near 660 cm${}^{-1}$ is expected from a factor-group analysis based on the crystal structure of CO₂. The implication is that each of the ices on which Figure 1 is based was partially or wholly crystalline, and so their associated optical constants provide no spectral-intensity information about amorphous CO₂.

Zoom In Zoom Out Reset image size

Similar comments apply to IR spectra of CO₂ ices presented elsewhere in the astrochemical literature (e.g., Sandford & Allamandola 1990), making it doubtful that amorphous-CO₂ spectra were presented in them. In the recent papers of Rocha & Pilling (2014, 2015), CO₂-ice spectra, although reported to be for a sample grown near 13 K, were said to agree with those of Hudgins et al. (1993) and Ehrenfreund et al. (1997), but those authors' ices were crystalline. In still other papers (e.g., Bennett et al. 2014), the IR reflection technique employed to acquire data, along with the small scale of the spectra presented, make an independent evaluation of phase and the extraction of band intensities for amorphous CO₂ difficult.

Our conclusion is that despite the relatively high interstellar abundance of solid CO₂, and an extensive associated literature, astrochemists still lack a mid-IR spectrum of amorphous CO₂ accompanied by a set of band strengths. Without such information for amorphous CO₂, applications to solid-state astrochemistry will be highly problematic. Therefore in this letter we describe our first results on amorphous CO₂ and its IR spectrum, all based on new laboratory measurements. We present new spectra of CO₂ ices whose phase can be unambiguously assigned, accompanied by newly determined band strengths. Finally, we consider some important CO₂-ice publications in light of our new results and comment on how our results will be useful.

The procedures and equipment used were similar to those of our recent CH₄ investigation (Gerakines & Hudson 2015; Hudson et al. 2015). The main difference between the work reported here and that from earlier CO₂ studies in other laboratories appears to be the slower rate at which our ices were grown by gas-phase condensation. The rates we used gave an increase in each sample's thickness of about 0.1 μm hr⁻¹. Doubling this rate gave crystalline CO₂ ices even at a deposition temperature of 10 K. Ice thicknesses were determined by laser interferometry using a reference index of refraction of n(670 nm) = 1.30. To calculate IR band strengths, denoted A', a density ρ = 1.28 g cm⁻³ was used, giving 1.751 × 10²² molecules ${\mathrm{cm}}^{-3}$ as the CO₂ number density. Both n(670 nm) and ρ were measured three times with CO₂ at 14 K in our laboratory with equipment and methods similar to those of Satorre et al. (2008), and with details to be supplied in a future paper. See also our similar measurements of n in Moore et al. (2010).

IR spectra were recorded as 100 scan accumulations with a resolution of 0.2 ${\mathrm{cm}}^{-1}$ on a KBr substrate in a high-vacuum chamber (10⁻⁷–10⁻⁸ torr). A resolution of 1 ${\mathrm{cm}}^{-1}$ gave essentially the same band intensities for our amorphous samples, but with lower noise, and so sometimes was employed for quick checks on the quality of a sample. The CO₂ used was obtained from Matheson Tri-Gas (Research Purity, 99.999%) Cambridge Isotope Laboratories (isotopic purity 99%). The chief contaminant seen was residual H₂O from the CO₂ reagents and possibly from our vacuum system. Trace amounts of H₂O gave weak IR peaks ($\lt 0.01$ absorbance units in all cases) in the 3700–3500 and 1600 cm${}^{-1}$ regions, far from the IR features of interest.

Uncertainties in our reported results are similar to those in our work on C₂H₂, C₂H₄, C₂H₆, and CH₄ (Hudson et al. 2014a, 2014b; Gerakines & Hudson 2015; Hudson et al. 2015) and are given in the table presented later in this paper.

Experiments began with the condensation at 10 K of room-temperature CO₂ gas to produce solid CO₂. Spectrum (a) of Figure 2 shows expansions of such an ice's ${\nu }_{3}$ (left) and ${\nu }_{2}$ (right) regions. Spectrum (b) was obtained by warming this same sample to 70 K and then recooling to 10 K. Close agreement was found between (b) and the spectra of crystalline CO₂ reported by Yamada & Person (1964), and when we deposited CO₂ at 60 K there was essentially perfect agreement for the band shapes and positions. Assigning spectrum (b) to crystalline CO₂ suggests that its precursor, spectrum (a), is that of amorphous CO₂. Supporting this assignment is the method used (slow condensation) to make the ice corresponding to (a), this spectrum's irreversible conversion to that of crystalline CO₂ on warming, and spectrum (a)ʼs lack of sharp substructure for the ${\nu }_{2}$ and ${\nu }_{3}$ features.

Zoom In Zoom Out Reset image size

The left-hand side of Figure 3 shows the result of separate, independent CO₂ depositions at 10 K for four sample thicknesses. As expected, in each spectral region an increasingly intense IR band was seen for increasingly thicker ices. Expansions around 1384 and 1277 ${\mathrm{cm}}^{-1}$ (not shown) revealed the weak ${\nu }_{1}$ and $2{\nu }_{2}$ Fermi resonance features. Since ${\nu }_{1}$ (symmetric stretching vibration) is IR-forbidden in the crystalline phase, its presence supports our assignment of these spectra to amorphous CO₂. See Hudson et al. (2015) for the activation of forbidden transitions in other amorphous ices.

Zoom In Zoom Out Reset image size

A few measurements were carried out with ${}^{13}{\mathrm{CO}}_{2}$, and the results mirrored those of our ${}^{12}{\mathrm{CO}}_{2}$ experiments. The peak positions for the stronger fundamentals for amorphous ${}^{13}{\mathrm{CO}}_{2}$ at 10 K were 2263.4 and 636.3 ${\mathrm{cm}}^{-1}$ for ${\nu }_{3}$ and ${\nu }_{2}$, respectively, giving ¹³C shifts of 66.4 and 18.4 ${\mathrm{cm}}^{-1}$.

The CO₂-ice IR spectra of Figures 2 and 3 possess many characteristics of interest, but in this paper we focus our attention on spectral profiles and intensities.

4.1. IR Spectra of Amorphous CO₂

4.2. IR Band Strengths of Amorphous CO₂

The method of Hollenberg & Dows (1961) was used to derive IR band strengths. Rearranging their equation to give

$$\begin{eqnarray}&&(2.303){\int }_{\mathrm{band}}(\mathrm{Absorbance})d\tilde{\nu }=\left({\rho }_{N}A\prime \right)h\end{eqnarray} \tag{ 1 }$$

shows that for the spectra on the left-hand side of Figure 3, a graph of 2.303 × (integrated absorbance) as a function of thickness ($h)$ will be a straight line with slope ρ ${}_{N}A\prime$, where ρ_N is the number density (molecules ${\mathrm{cm}}^{-3})$ of CO₂ molecules in the sample and A' is the apparent band strength. The right-hand side of the same figure shows the resulting plots for two amorphous-CO₂ features. Slopes are given in Table 1 along with the resulting values of A'. Similar graphs of 2.303 × (peak height) as a function of thickness gave apparent absorption coefficients (α'), also listed in Table 1. See Hudson et al. (2014a, 2014b and references therein) for additional details, including the differences between apparent and absolute intensities.

Table 1.  IR Features of Amorphous CO₂ at 10 K^a

Property	${\nu }_{1}+{\nu }_{3}$	$2{\nu }_{2}+{\nu }_{3}$	${\nu }_{3}$	${\nu }_{2}$
$\tilde{\nu }/{\mathrm{cm}}^{-1}$	3704	3597	2329	654.7
λ/μm	2.700	2.780	4.294	15.27
FWHM/${\mathrm{cm}}^{-1}$	9.8	7.1	27.1	9.2
α'/${\mathrm{cm}}^{-1}$	3890 ± 40	1557 ± 18	71,710 ± 3,016	30,380 ± 131
${\rho }_{N}A\prime$/${\mathrm{cm}}^{-2}$	70,980 ± 1490	21,860 ± 372	2,058,000 ± 84,130	302,900 ± 9363
A'/cm ${\mathrm{molecule}}^{-1}$	4.05 ± 0.10 × ${10}^{-18}$	1.25 ± 0.02 × ${10}^{-18}$	1.18 ± 0.05 × ${10}^{-16}$	1.73 ± 0.53 × ${10}^{-17}$
Integration range/${\mathrm{cm}}^{-1}$	3762–3645	3643–3577	2420–2290	680–630

Note.

^aFWHM; α' and A' denote apparent absorption coefficient and apparent band strength taken directly from a set of IR spectra using Beer's law type plots; values of n(670 nm) = 1.30 and ρ = 1.28 g ${\mathrm{cm}}^{-3}$, or ρ_N = 1.751 × 10²² molecules ${\mathrm{cm}}^{-3}$, for CO₂ were used throughout.

Download table as:  ASCIITypeset image

Figure 3 also shows that the band positions and profiles of amorphous-CO₂ IR spectra do not undergo significant shifts with ice thickness, unlike the case of, for example, crystalline C₂H₂ (Hudson et al. 2014a). This means that the band strengths (A') listed for each amorphous-CO₂ IR band in Table 1 should be applicable without modification to a wide variety of interstellar and planetary-science situations.

For applications to interstellar chemistry, perhaps the most important A' in Table 1 is $A\prime ({\nu }_{3})$ = 1.18 × ${10}^{-16}$ cm ${\mathrm{molecule}}^{-1}$. Most values now in the astrochemical literature are smaller and can be traced to older work on crystalline CO₂ at higher temperatures. For example, $A\prime ({\nu }_{3})$ values near 7.7 × ${10}^{-17}$ cm ${\mathrm{molecule}}^{-1}$ were used by Lv et al. (2012), who cited Jamieson et al. (2006) who cited Gerakines et al. (1995), who relied on the original measurement of this value by Yamada & Person (1964) for crystalline CO₂ near 80 K. The $A\prime ({\nu }_{3})$ value we are reporting for amorphous CO₂ near 10 K is about 53% larger than that of Yamada & Person (1964) for crystalline CO₂ near 80 K.

4.3. Astrochemical Applications and Implications

The immediate result of our work is that the mid-IR spectrum of amorphous CO₂ is now available, accompanied by band strengths and a knowledge that the accompanying band profiles and positions do not change substantially over the nanometer-to-micrometer range. There are several areas where these results can be important. First, an A' value for crystalline CO₂ ice has been used in many laboratory studies to calculate abundances of starting materials and of reaction products (e.g., Mennella et al. 2006; Peeters et al. 2010). Our new A' values (Table 1) will necessitate reevaluations of some of those earlier results, particularly those involving amorphous ices. Similarly, previous laboratory work in which abundance ratios, such as CO/CO₂, were calculated may need reassessing (e.g., Pilling et al. 2011). Also, kinetic studies that rely on CO₂ abundances, such as for a mass balance, almost certainly need revisiting (e.g., de Barros et al. 2014). Our IR spectra can be useful when comparisons might benefit from data on amorphous CO₂, as opposed to crystalline CO₂ (Iopollo et al. 2009, 2013; Lv et al. 2014). Edridge et al. (2013) allude to the unexpected observation of an IR feature near 2381 ${\mathrm{cm}}^{-1}$ in work on amorphous CO₂ by others, this particular band being more expected for a crystalline ice. However, Figure 7 of that earlier study (Baratta & Palumbo 1998) shows a double-peaked feature near 660 ${\mathrm{cm}}^{-1}$, and so according to our Figure 2 the sample used was partially or wholly crystalline CO₂, explaining the 2381 ${\mathrm{cm}}^{-1}$ feature (Parker & Eggers 1966).

Beyond these laboratory applications, IR band profiles of solid CO₂ have been used extensively in attempts to fit IR spectra of interstellar ices. Such fits, from which CO₂-ice abundances are derived, may need reassessing in light of the significant differences we have found in the band profiles and intensities of amorphous and crystalline CO₂ (Figure 2). As an example, Figure 4 contains a fit to the 15 μm Spitzer Space Telescope spectrum observed toward the background field star Q21-1, a line of sight that probes the dense ISM in the cold dark cloud IC 5146 (data from Whittet et al. 2009). Two laboratory components are included in roughly a 10:1 ratio, H₂O + CO₂ + CO = 100:20:3 at 20 K (Ehrenfreund et al. 1996) and amorphous CO₂ at 10 K (this study), respectively. The resulting best-fitting combination shown in Figure 4 suggests that up to ∼9% of the solid CO₂ in this line of sight may be amorphous. Previous studies using laboratory fits to interstellar quiescent-cloud CO₂ spectra conclude that the non-polar component is either a CO- or CO₂-dominated ice mixture.

Zoom In Zoom Out Reset image size

Concerning IR band strengths, those now in widest use for amorphous CO₂-containing mixtures near 10 K (Gerakines et al. 1995) derive from measurements on pure crystalline CO₂ near 80 K (Yamada & Person 1964), scaled with ice composition. Our band strengths measured near 10 K for amorphous CO₂ could be more appropriate to use in such cases.

An accurate, quantitative spectroscopic characterization of amorphous CO₂ has been presented. Our new IR transmission spectra, such as in Figures 2 and 3, will help to distinguish amorphous and crystalline CO₂ ice phases through band positions, widths, and shapes. However, just as important are the band intensities in our Table 1, which can be used to derive CO₂ abundances in ices and, in turn, deduce evolutionary and environmental information, such as thermal histories. We also suggest that a primary reason for the confusion and disagreement among the spectra of Figure 1, and other such data (e.g., Poteet et al. 2013), is that the various ice samples used were not of a single CO₂ phase, but for amorphous-crystalline mixtures. A fresh opportunity now presents itself to accurately reassess the role of frozen CO₂ in planetary and interstellar ices and the information it can convey about its surroundings. Spectra are posted on our group's web site.¹

NASA funding through the Astrophysics Research and Analysis, Cassini Data Analysis, and Outer Planets Research programs is acknowledged. Both authors received partial support from the NASA Astrobiology Institute through the Goddard Center for Astrobiology. Mark Loeffler is particularly acknowledged for constructing the ultra-high vacuum system with which he and Marla Moore measured n and ρ. Robert Ferrante and Tatiana Tway are thanked for assistance with IR measurements.