DETECTION OF PROPENE IN TITAN'S STRATOSPHERE

Titan, largest satellite of Saturn, is unique amongst moons by virtue of its deep atmosphere, consisting mostly of nitrogen and methane (∼95% and ∼5%, respectively, in the troposphere; Niemann et al. 2005). These two molecules are disintegrated, mainly by solar UV photons and Saturn magnetospheric electrons in the upper atmosphere, with fragments recombining to give rise to a plethora of organic molecules. These include hydrocarbons (C_xH_y) and nitriles (C_xH_y[CN]_z), many of which were first identified in Titan's stratosphere by the infrared interferometer spectrometer (IRIS) on Voyager 1 during the 1980 flyby (Hanel et al. 1981; Kunde et al. 1981; Maguire et al. 1981). Prior to the arrival of Cassini in 2004, further molecules, including the oxygen-bearing species CO, CO₂, and H₂O (Lutz et al. 1983; Samuelson et al. 1983; Coustenis et al. 1998), and later CH₃CN and C₆H₆ (Marten et al. 2002; Coustenis et al. 2003) were spectroscopically detected in Titan's atmosphere using a variety of ground- and space-based observatories. More recently the Cassini Ion and Neutral Mass Spectrometer has discovered an even greater diversity (Vuitton et al. 2007; Cui et al. 2009) in the upper atmosphere (ionosphere).

Regarding the stratospheric hydrocarbons, the C₂H_x series of acetylene (or ethyne, C₂H₂), ethylene (or ethene, C₂H₄), and ethane (C₂H₆) were all firmly detected by IRIS, confirming earlier ground-based observations. Of the C₃H_x species, methyl-acetylene (or propyne, CH₃C₂H) and propane (C₃H₈) were first identified by Maguire et al. (1981) from IRIS data. Others in the sequence including allene (CH₂CCH₂), isomer of propyne; and the alkene molecule propene (or propylene, C₃H₆); and its isomer cyclopropane (see Figure 1), have not been identified spectroscopically during the 32 yr since Voyager.

Zoom In Zoom Out Reset image size

In this Letter we report the first detection of propene in Titan's stratosphere using spectra from the Cassini Composite Infrared Spectrometer (CIRS), filling an important gap in the C₃H_x family, and providing insights into the relative abundances of alkanes, alkenes, and alkynes in Titan's atmosphere. In Section 2 we discuss our observations and data analysis method; in Section 3 we present our results; in Section 4 we discuss the context and implications of this discovery; and Section 5 is our conclusions.

2.1. Data Set

2.2. Radiative Modeling and Inversion

Figure 2(a) shows the spectral averages for FP3 in the range 880–940 cm⁻¹ (in black), along with the final model fits (colored lines). This spectral region is characterized by lines of the ethylene ν₇ band at 909, 915.5, 922.5 cm⁻¹ and from the ν₂₁ band of propane with a Q-branch at 922 cm⁻¹, modeled here using the new JPL pseudo-linelist. Note the "missing" emission in the model spectrum at 912.5 cm⁻¹, more easily seen on Figure 2(b), which shows the residual after the model is subtracted from the data. This 912.5 cm⁻¹ emission occurs in the precise location of the ν₁₉ band of propene (C₃H₆) (Lafferty et al. 2006). For comparison, at the bottom of Figure 2(b) we show a laboratory (5°C) absorbance (−log₁₀(I/I₀)) spectrum of propene recorded at the Pacific Northwest National Laboratory (PNNL; Sharpe et al. 2004) confirming the feature location. The PNNL spectrum was smoothed from a native resolution of 0.112 cm⁻¹ (boxcar apodization) to the CIRS resolution of 0.5 cm⁻¹ (Hamming apodized), and has a peak absorbance at 912.5 cm⁻¹ (in arbitrary units). To rule out any possible instrumental artifacts, we also show (in red) above each residual spectrum the corresponding standard deviation of the individual spectra in the average. Some small peaks occur due to electrical interference in the CIRS instrument, but none of these falls at 912.5 cm⁻¹.

Zoom In Zoom Out Reset image size

To increase S/N and improve the detection of propene, we averaged the non-overlapping 125, 175, and 225 km averages, ensuring no duplicate information, each weighted by the N_spec of Table 1. This average, along with the weighted best-fit spectrum, is shown in Figure 3(a). We also show the model calculation with the new (JPL pseudo-linelist) lines of propane not included, i.e., a spectrum due to ethylene and the continuum only. The residuals from both models (propane and non-propane) are shown in the lower panel (b), illustrating that the propene emission at 912.5 cm⁻¹ is not part of the propane P-band emission (the broad bump from ∼905–920 cm⁻¹).

Zoom In Zoom Out Reset image size

A spectral line listing for propene is not yet available, so we cannot include C₃H₆ in our radiative transfer model to retrieve its abundance. However we can estimate its abundance from that of propane, by comparing the relative strengths of their emissions. The radiance (I) of optically thin spectral lines is proportional to both the gas abundance (q) and the spectral absorption cross-section (k). If we additionally assume that the emission in limb viewing originates mostly from the tangent altitude z, where the atmospheric pressure is p, temperature is T_z, and Planck function is $B(\tilde{\nu },T_z)$, then we can write

$$\begin{equation} I(\tilde{\nu }, z) \propto k(\tilde{\nu }) f(\tilde{\nu },T_z, T_0) q(z) p(z) B(\tilde{\nu },T_z), \end{equation} \tag{ 1 }$$

where $f(\tilde{\nu },T_z, T_0)$ is a first-order correction factor to the overall band strength for the temperature-dependence of the occupancy of the energy states using Boltzmann statistics, similar to Equation (A11) of Rothman et al. (1998):

$$\begin{equation} f(\tilde{\nu }, T_z, T_0) = \frac{\exp (-c_2 \tilde{\nu } / T_z) / V(T_z)}{\exp (-c_2\tilde{\nu }/T_0)/V(T_0)}. \end{equation} \tag{ 2 }$$

In the above formula, T₀ is the temperature of the laboratory measurements (278 K), c₂ is the second radiation constant hc/k = 1.4388 cm K, and V(T) is the total vibrational partition function defined in Equation (B.6) of Nixon et al. (2009a). By writing this radiance formula separately for the propane emission at 922 cm⁻¹ and propene at 912.5 cm⁻¹, then taking ratios, and canceling the Planck functions and ambient pressures, we find

$$\begin{eqnarray} q_{\rm C_3H_6}(z) =& q_{\rm C_3H_8}(z) \left(\frac{k(922.0)}{k(912.5)} \right) \nonumber\\ & \times \left(\frac{f(922.0, T_z, 278)}{f(912.5, T_z, 278)} \right) \left(\frac{I(912.5, z)}{I(922.0, z)} \right). \qquad \end{eqnarray} \tag{ 3 }$$

The emission (I) at 912.5 and at 922.0 cm⁻¹ is obtained from the residual of spectral fitting (i.e., from the blue and red curves of Figure 3(b)), while the absorption cross-sections (cm² molecule⁻¹) were obtained from the 278 K absorbance spectra of C₃H₈ and C₃H₆ using the conversion factor of 9.28697 × 10⁻¹⁶ recommended by PNNL. The absorption cross-sections were smoothed to the requisite resolution and extracted at 912.5 and 922.0 cm⁻¹. Some assumptions in this technique may give rise to systematic errors, in particular the simple scaling we have used for band intensities from laboratory (278 K) to Titan stratospheric temperatures. Although this may account reasonably well for the principal "cold bands" of each gas (transitions from ground state), the contribution of hot bands to the observed emission features is unknown and is therefore the major source of uncertainty.

Figure 4 shows the final results. Panel (a) shows the temperature retrieval from the methane ν₄ band, along with the averaging ranges, while panels (b) and (c) show the gas abundances for the C₂H_x and C₃H_x hydrocarbons respectively, all derived from modeling the individual gas bands except in the case of propene. Note that for the propane vertical profile we used the retrieved abundances from the ν₂₁ band, although the ν₈ and ν₂₆ bands were also fitted. A detailed intercomparison of propane retrievals from the various bands using the JPL pseudo-linelist will be the subject of a forthcoming paper.

Zoom In Zoom Out Reset image size

In the case of propene, the emission ratioing technique (Equation (3)) was used to derive its abundance at the five altitudes, and subsequently interpolated using cubic splines onto the same, closer-spaced altitude grid range (100–250 km) as the other gases. The propene error bars were estimated by adding in quadrature the radiance relative error of the propene emission, the abundance relative error of the propane, from which the propene is scaled, and systematic uncertainties in the laboratory cross-section determination of both gases:

$$\begin{equation} \Delta q_{\rm C_3H_6} = q_{\rm C_3H_6} \times \left\lbrace \left(\frac{\Delta q_{\rm C_3H_8}}{q_{\rm C_3H_8}} \right) ^2 + \left(\frac{\Delta I_{912.5}}{I_{912.5}} \right) ^2 + 2\sigma _{\rm cs}^2 \right\rbrace ^{1/2}. \end{equation} \tag{ 4 }$$

The value of σ_cs for both C₃H₈ and C₃H₆ is estimated by PNNL at 3% total from all sources, including gas purity, pressure and temperature stability in the cell, etc. Table 1 gives the abundances at each of the five altitudes. Our derived gas profiles for the five previously known species are in good qualitative and quantitative agreement with those of Vinatier et al. (2010) at 20°S (see their Figure 3), although showing somewhat less vertical structure, likely due to the long time-base of these averages. The abundance of propene in the stratosphere is estimated here for the first time, ranging from 2.0 ± 0.8 ppbv at 100 km, to a maximum of 4.6 ± 1.5 ppbv at 200 km, and possibly declining above.

Propene has previously been detected by submillimeter techniques in a dark interstellar (molecular) cloud, TMC-1 by Marcelino et al. (2007), the first and to our knowledge the only prior astronomical sighting of this molecule. The species C₃H₆ has also been inferred from Cassini mass spectrometry of Titan's upper atmosphere as the cause of the signal at 42 amu, with an abundance of 2.3–3.4 ppmv at 1050 km (Magee et al. 2009). However, mass spectra alone cannot distinguish between isomers, so an uncertainty remained as to whether the species was propene or cyclopropane. Therefore, our result constitutes the first definitive detection of propene in a non-terrestrial planetary atmosphere.

C₃H₆ is an important component of photochemical models of Titan's atmosphere and has been included in reaction schemes for C₃H_x since the earliest post-Voyager models (Yung et al. 1984). Cassini-era models have also necessarily included C₃H₆ (e.g., Wilson & Atreya 2004; Lavvas et al. 2008a, 2008b; Wilson & Atreya 2009; Krasnopolsky 2009, 2010; Vuitton et al. 2012), but few have published stratospheric abundance predictions for propene, apparently due to the lack of observational detection. A vertical profile for propene is given in Wilson & Atreya (2004), showing an abundance of a few ppbv between 100 and 200 km, and positive vertical gradient, in approximate agreement with our results. Propene is also shown in the three versions of the Krasnopolsky model (Krasnopolsky 2009, 2010, hereafter K09 and K10). However, the gas profiles in the K09 and K10 models (defined by escape, and no escape of CH₄, respectively) are much too steep, not just for propene, but for all the hydrocarbon species. The third variant of the model, in Appendix A of K09, which uses a different eddy diffusion profile (recommended by Hörst et al. 2008), predicts a propene abundance of 2–7 × 10⁻⁸ at 100–200 km, significantly in excess of our findings. In all three versions, the abundance of C₃H₆ is comparable, or greater than that of CH₃C₂H below 500 km, in contradiction with our results.

Recently, Hébrard et al. (2013) has critically re-examined and updated the photochemistry of all C₃H_x species. In their reaction scheme, at 1000 km propene is principally produced by the reactions: H+C₃H₅ → C₃H₆ + hν (62%) and CH+C₂H₆ → C₃H₆ + H (37%). However, in the stratosphere, propene may be produced by other mechanisms, including the termolecular reaction: CH₃ + C₂H₃ + M → C₃H₆ + M. Important loss mechanisms at these altitudes are (1) photodissociation, and (2) H + C₃H₆ + M → C₃H₇ + M. The predicted stratospheric abundances of Hébrard et al. (2013) range from ∼10⁻¹⁰ at 100 km to ∼7 × 10⁻¹⁰ at 200 km, somewhat less than our findings, and also with a significantly steeper gradient. However, Hébrard et al. (2013) caution that their results are not expected to be fully valid below 200 km due to chemical processes not included in the model, such as the effects of cosmic rays.

We remark on an emerging trend amongst the abundances of small hydrocarbons in Titan's lower atmosphere, summarized as alkanes (C–C single bonds) > alkynes (C≡C triple bonds) > alkenes (C=C double bonds). This is now evident for both the C₂H_x family (C₂H₆ > C₂H₂ > C₂H₄) and the C₃H_x family (C₃H₈ > CH₃C₂H > C₃H₆). Therefore, the ordering of abundances does not follow the sequence of molecular size, or bond saturation, with alkynes being more plentiful than alkenes. A possible explanation is that alkynes, once the strong triple bond has formed, are relatively more secure than alkenes, with the weaker double bond.

Indeed, the phenomenon of more abundant triple-bonded molecules than double-bonded ones extends beyond the hydrocarbons in Titan's stratosphere, and is reflected in the roster of detected species. Nitrogen has only been found in Titan's lower atmosphere thus far in triple-bonded form (either N≡N or C≡N). In all, nine molecules with triple bonds have been detected: N₂, CO, C₂H₂, CH₃C₂H, HCN, CH₃CN, HC₃N, C₄H₂, C₂N₂: the latter three with two triple bonds apiece. In contrast, only three double-bonded species are found: CO₂, C₂H₄, and now C₃H₆. Therefore, triple bonds, once formed, appear to be persistent in Titan's atmosphere.

In this Letter we have demonstrated the first definitive detection of propene (C₃H₆) in an extra-terrestrial planetary atmosphere, with an inferred abundance of ∼2–5 ppbv (Table 1). This fills a long-standing gap in the C₃H_x series of known aliphatic hydrocarbons in Titan's stratosphere and will provide important additional constraint on photochemical models. We have also demonstrated an apparent trend in hydrocarbon abundances for the lower atmosphere: alkanes > alkynes > alkenes, within each of the C₂H_x and C₃H_x chemical families, and speculate that the triple bond is more resistant to photolysis or chemical attack than the double bond. This picture is undoubtedly simplistic and detailed photochemical modeling is required to elucidate the details, including the role played by cyclic molecules.

Our thanks to Linda Brown (JPL) and Steven Sharpe (PNNL) for providing us with molecular spectroscopic data for propane and propene: absorbances and line lists. C.A.N., D.E.J., N.G., T.M.A., and F.M.F. were supported by the NASA Cassini Mission. Part of the research was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration. V.C. was supported by the NASA Post-Doctoral Program. N.A.T. and P.G.J.I. received support for their portion of this work from the UK STFC, and NAT received additional support from the Leverhulme Trust. B.B., S.V., and A.C. were supported by the Centre National d'Études Spatiales (CNES).

Facility: Cassini - Cassini-Huygens