We have analyzed limb daytime observations of Titan's upper atmosphere at 3.3 μm, acquired by the visual-infrared mapping spectrometer (VIMS) on Cassini. They were previously studied by García-Comas et al. (2011) to derive CH4 densities. Here, we report an unidentified emission peaking around 3.28 μm, hidden under the methane R branch. This emission is very strong, with intensity comparable to the CH4 bands located in the same spectral region. It presents a maximum at about 950 km and extends from 600 km up to 1250 km. It is definitely pumped by solar radiation since it vanishes at night. Our analysis shows that neither methane nor the major hydrocarbon compounds already discovered in Titan's upper atmosphere are responsible for it. We have discarded many other potential candidates and suggest that the unidentified emission might be caused by aromatic compounds.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The visual-infrared mapping spectrometer (VIMS) part of the Cassini payload has sounded Titan's thermosphere during many flybys. Recently, Adriani et al.  have analyzed VIMS limb daytime observations at 3.0 μm between 500 and 1100 km to derive HCN density profiles using a dedicated non-local thermal equilibrium (non-LTE) model. Evidence of methane fluorescence at 3.3 μm in Titan's upper atmosphere from VIMS measurements was pointed out by Drossart et al.2005] and Brown et al. . That emission was used to understand the non-LTE excitation of the CH4 vibrational levels and to retrieve the CH4volume mixing ratio, helping to constrain the location of the homopause. The 3.3 μm region was studied in much more detail by García-Comas et al. ; in their paper, the VIMS limb daytime observations were used to derive accurate CH4 density profiles using an updated and sophisticated CH4non-LTE model. In this paper, we further analyze the limb daytime VIMS observations, acquired during July and August 2007 Cassini flybys, and discover an emission never identified before which is not attributable to CH4. This feature is present not only in the analyzed data set but also in all daytime VIMS observations of Titan's atmosphere. A very accurate analysis of this unidentified emission (hereafter UnE) has been performed, discarding possible instrumental artifacts. We report here the major characteristics of the UnE, i.e., its spectral location and its altitude distribution, and an extended discussion about compounds that are the potential carriers.
2 The Measurements
 VIMS is an imaging spectrometer on board the Cassini spacecraft. It generates images on up to 64 × 64 pixels at 352 separate wavelengths (spectral cubes). The infrared channel (0.85–5.2 μm) has a nominal angular resolution of 0.25 × 0.5 mrad and a spectral resolution of 16 nm [Brown et al., 2004]. VIMS is therefore capable of measuring Titan's upper atmospheric limb emission with a moderate vertical resolution (22–37 km) from its surface up to 1300 km.
 In this work, we analyze the data taken during two Titan flybys: T34 (19 July 2007) and T35 (31 August 2007). Among all the acquired measurements, we have selected the spectral cubes that cover the highest atmospheric altitudes (above 400–500 km). In order to minimize possible stray-light contaminations and to have the largest atmospheric signal, we have used measurements with the largest integration time, best vertical resolution (22–37 km), and a phase angle lower than 60°.
 The spectral and radiometric calibration of the cubes was carried out using the latest VIMS pipeline (RC17) [Filacchione et al., 2007]. The instrument noise error near 3.3 μm is about 30 nWm−2 nm−1 sr−1. VIMS IR wavelengths follow the expression λ= 0.01668 × B−0.7381 μm, where B=96−351.
 From the selected cubes, we have extracted sequences of spectra sampling Titans atmosphere at the limb, i.e., along lines perpendicular to Titan's surface. A total of 20 sequences, all of them in the Southern Hemisphere, were used. More details on the observations and the description of the data can be found in Adriani et al. , García-Comas et al. , and references therein.
 The observed atmospheric limb emission near 3.3 μm at daytime is very strong (see Figure 1), and the measurements have a reasonably good signal-to-noise ratio (∼2 for the co-added integrated spectra) up to about 1250 km above Titan's surface. These large intensities can only be explained by invoking non-LTE effects, e.g., species that are pumped by solar radiation near 3.3 μm and then fluoresce [López-Puertas and Taylor, 2001; García-Comas et al., 2011]. Without that pumping, the emission at the cold typical temperatures of Titan's upper atmosphere, ∼150 K, would be several orders of magnitude weaker.
 The emission between 450 km and 1250 km clearly shows the fingerprints of the CH4molecules, characterized by the R, Q, and P branches. However, as can be seen in Figure 1, the relative intensity of these branches changes with altitude. This “asymmetry” emerges very clearly in the vertical distribution of the radiances integrated over the spectral intervals of the three CH4branches (red line for the R branch, black line for the Q, and blue line for the P in Figure 2). We see that, while the P and Q branches have a similar altitude dependence, the intensity of the R branch has a different behavior.
 Simulations of the observed spectra with a sophisticated CH4 non-LTE model and the CH4densities derived from the Q and P branches [García-Comas et al.2011] show that the CH4emission alone cannot reproduce the excess of emission of the R branch. The model (dashed line in Figure 1) predicts very well the measured radiance (black solid line) at wavelengths larger than 3.3 μm, but it clearly underestimates the emission at wavelengths shorter than 3.3 μm. We subtracted the simulated CH4emission from the measured radiance and isolated the spectral feature shown in Figure 3, the “unidentified” emission (UnE). This emission, which has not been to date reported, exhibits a clear peak around 3.28 μm.
 In Figure 2, we have also plotted the integrated radiance of the UnE (magenta line). This emission is present at altitudes above 650 km, with a maximum at about 950 km and a secondary peak at ∼1200 km. The estimated errors in the integrated radiance of the UnE (shown by the error bars of the magenta line in Figure 2) account for the instrumental errors (more important at high altitudes) and the error in the retrieved CH4abundance. The radiance has a good signal-to-noise ratio up to near 1250 km.
3 Analysis of the “Unidentified” Emission
 The first step in our analysis of the UnE was to check that the observed emission was not due to an instrumental artifact or to calibration errors. The errors introduced by the spectral and radiometric calibrations have been discussed in detail by García-Comas et al. . In this spectral region, the spectral calibration error is smaller than 0.1% [Cruikshank et al., 2010]. The instrumental noise is about 30 nW/(m2 nm sr) for a single spectrum. The number of spectra averaged to obtain the signals shown in Figures 1and 3ranges from 5 to 16, depending on the tangent altitude. Therefore, the noise level of the residual spectra is close to 10 nW/(m2 nm sr). This implies that the peak of the UnE is about 20 times larger than the noise level.
 The emission discovered cannot be caused by an instrument's spectral artifact, for instance by some imperfect pixels, because the excess radiation is observed in the same altitude range by different pixels. Similarly, it can be ruled out that the observed feature is caused by a radiometric calibration artifact because it would appear at all altitudes, also below 700 km where the anomaly is not observed (see Figures 2and 3).
 Methane abundance or non-LTE processes affecting the CH4vibrational level populations could be potential causes of the UnE. However, since CH4 emits over the anomalous emission, in the R branch and also in the Q and P branch regions, a change in the CH4abundance would affect in a similar way the radiances in the three spectral branches and will not cause an asymmetry between the R and the Q and P branches. It is also worth noting that below 700 km such an asymmetry is not observed, demonstrating that it cannot be produced by the CH4 abundance. The same argument can be applied to the non-LTE populations since they also affect the radiance of the three branches in a similar way. Hence, any inaccuracy in the non-LTE model would just scale the whole CH4spectrum but would not produce any asymmetry. We have also ruled out rotational non-LTE effects of the CH4 energy levels since they should affect the three branches in a similar way. The change of the spectral shape with altitude could be induced by different relative contributions of the fundamental, isotopic, or hot bands of methane, as happens in the CO2near-IR spectra of the daytime atmospheres of Mars, Venus, and Earth [López-Valverde et al., 2011]. This is not, however, the case here because the spectral displacements among the different CH4vibrational bands are very small, smaller than the spectral resolution of VIMS, and the contributions of the isotopic and hot bands have been found to be negligible at these altitudes [see García-Comas et al., 2011, Figure 6].
 Attempts to assign the UnE to very weak vibrational bands of the main CH4 isotopologue or to other isotopic forms of this molecule has also failed [García-Comas et al., 2011]. The spectroscopy of the CH4molecule is rather well known, and, if such a strong emission was due to CH4, its spectroscopic data should appear in most of the existing data sets. So the UnE has to be originated from a different molecule(s) (or particles) which is (are) either as abundant as CH4 in Titan's upper atmosphere or strongly excited by solar radiation or other energy sources.
Bellucci et al.  have recently detected an absorption band at 3.4 μm in Titan's VIMS spectra. Since that wavelength is the typical signature of C–H vibrations of aliphatic chains on large organic molecules, they attributed it to aerosols formed by such molecules. The same absorption band could be responsible of an attenuation of the Q and P branches of CH4 causing the observed anomalous relative intensity of the R branch (P and Q are attenuated, while R is unchanged). However, that absorption band has been observed only at altitudes below 480 km, and we see anomalous intensities up to 1200 km. Thus, a thick layer of absorbing aerosols above 1200 km would be needed, and such a layer has not been observed. On the other hand, because the absorption frequency of these aerosols is significantly different from the frequency of our UnE, the latter cannot be due to their non-LTE emission.
 Since the spectral region where the UnE occurs is typical of the C–H stretch and the altitude range where it occurs is very far from Titan's surface, we have checked as potential carriers “light” molecules like C2H4, C2H6, C3H6, , CH3CN, and C2N2, which have been already observed in significant quantities in Titan's atmosphere [Waite et al., 2007, 2010]. As shown in Figure 4, the spectral signatures of all these molecules do not fit the observed emission. Since heavy hydrocarbon molecules might be formed in Titan's thermosphere [Sittler et al., 2009], we have also plotted in Figure 4the spectral intervals of the C–H stretching bands of the aromatic, olefinic, and aliphatic hydrocarbons [Bellamy, 1958; Silverstein and Bassler, 1967; Socrates, 2001]. As shown in the figure, the typical spectral interval of C–H stretch modes of the aliphatic hydrocarbons fall at longer wavelengths than the observed feature. On the contrary, that of the olefinic compounds falls at shorter wavelengths. The extreme end of the olefinic stretch can produce a band near 3.3 μm. However, this stretch falls in a robust range, and, when the other end of the group is attached to other organics, the band is shifted to higher frequencies, e.g., to 3.23–3.25 μm [Bellamy, 1958; Silverstein and Bassler, 1967; Socrates, 2001]. Hence, the latter can be considered as the largest wavelength of the olefinic C–H stretch modes. Alkynes with triply bonded carbons show bands at 3300cm−1(3.0 μm). As a typical example of this class of molecules, we have plotted in Figure 4the spectrum of acetylene. It is clear from the plotted spectrum that neither alkynes can be responsible of the UnE.
 As shown in Figure 4, the spectral signature of the UnE is typical of aromatic molecules, i.e., molecules containing at least one aromatic ring. The simplest molecule containing an aromatic ring is benzene, which has been recently measured in Titan's atmosphere [Waite et al., 2007]. Calculations performed (i) with the most recent benzene spectral data [Rinsland et al., 2008], (ii) using the benzene distribution in the upper atmosphere of Vuitton et al.  (which explains very well the Ion and Neutral Mass Spectrometer (INMS) measurements taken down to about 950 km [Waite et al., 2007]), and (iii) assuming that benzene's near-IR energy levels are pumped by solar radiation as those of CH4show that benzene fits rather well spectroscopically the UnE, but the concentration required to explain such a strong signal is about 700 times larger than INMS's measurements. Furthermore, the peak of the emission profile is located about 150 km above the benzene peak. Hence, benzene near-IR emission cannot explain the UnE.
 “Hot” benzene [Tsai et al., 2000] could be another candidate for explaining the UnE. Benzene is excited near 3.3 μm not only by near-IR solar radiation but also by solar UV and visible radiation. We have computed the excitation of benzene produced by UV and visible solar radiation taking into account the cross-sections of benzene in the UV and visible [Etzkorn et al., 1999; Feng et al., 2002], the solar UV-VIS flux at Titan's top atmosphere [Rottman, 2005], and a mean efficiency of 10% to 60% for wavelengths from 160 to 245 nm [Tsai et al., 2000]. We have found, however, an excitation similar to that produced by the near-IR pumping. Hence, even with the inclusion of this additional excitation, the simulated spectra would require unreasonably high benzene concentrations to reproduce the observed spectra. Furthermore, the “hot” benzene hypothesis implies a large benzene emission at longer wavelengths due to its high-energy vibrational levels (i.e., the anharmonic hot bands), and VIMS spectra do not show that feature.
 As mentioned above, besides benzene, all the aromatic compounds do present a very typical spectral signature near 3.28 μm [van Diedenhoven et al., 2004; Tielens, 2006; Bauschlicher Jr et al., 2008] (see Figure 4). Waite et al.  attributed the large ions measured by INMS in Titan's upper atmosphere [Coates et al., 2007] to polycyclic aromatic hydrocarbon (PAH) compounds. All these compounds show the aromatic ring signature in their spectra, and they might therefore be responsible for the emission we have discovered. However, how the small measured concentrations can produce such a strong emission, similar in magnitude to the strong bands of the very abundant CH4, is still unknown.
4 Summary and Conclusions
 We have analyzed Cassini/VIMS limb daytime observations of Titan's upper atmosphere in the 3.3 μm region. After removing the CH4 emission by using the sophisticated CH4non-LTE model of García-Comas et al.  and the inverted CH4densities, we have found an emission feature that is superimposed to the CH4 R branch and peaks near 3.28 μm. It is present in a wide altitude range extending from around 600 km until 1250 km, with a maximum intensity around 950 km. Furthermore, it is certainly pumped by solar radiation since it vanishes at night.
 We have considered many hypotheses to explain the existence of this anomalous emission. Potential instrumental artifacts have been discarded. Our analysis shows that it cannot be explained by the emission of weak bands or isotopologues of CH4nor by wrong CH4non-LTE modeling. We have ruled out all the major hydrocarbons already observed in Titan's atmosphere as likely explanations for it. We have also discarded other potential candidates as olefinic and aliphatic hydrocarbons because they do not fit the UnE spectrally. The unassigned spectral feature is likely due to emission of molecules or particles containing C–H bonds. Aromatic light molecules, like benzene, fit quite well the observed spectral feature, but their concentrations are not large enough to explain the magnitude of the emission. Due to its spectral position, we suggest that the UnE might be caused by polycyclic aromatic hydrocarbons (PAHs) or heterocyclic aromatic compounds (HACs), as they have been suggested to be present in Titan's upper atmosphere [Coates et al., 2007; Waite et al., 2007, 2010]. However, how the small amounts that have been measured can emit such a strong radiance is still unknown.
 Special thanks are given to the Cassini/VIMS team for planning the VIMS observations and to J.-M. Flaud for assistance with spectroscopic data. The ISAC-CNR and IAPS-INAF teams have been funded by the Cassini project of the Italian Space Agency. The IAA team was supported by the Spanish MCINN under grants, AYA2011-23552, the CONSOLIDER program , and EC FEDER funds. M.G.C. is financially supported by the MINECO under its “Ramon y Cajal” subprogram.