A TENTATIVE IDENTIFICATION OF HCN ICE ON TRITON^*

Triton has been studied spectroscopically in the visible and near-infrared regions to a maximum wavelength of ≈4 μm (Quirico et al. 1999; Grundy et al. 2002; Cruikshank 2005), and a fraction of its surface has been imaged in considerable detail by the Voyager 2 spacecraft during the flyby of the Neptune system in 1989 (Lunine 1990). The spectroscopic studies have revealed five different molecular ices: N₂, CH₄, CO, CO₂ (Cruikshank & Apt 1984; Cruikshank et al. 1984; Quirico et al. 1999), and H₂O (Cruikshank et al. 2000). The absorption bands of CH₄ are all shifted to wavelengths shorter than the laboratory values, indicating that the methane is dissolved in the nitrogen ice (Cruikshank et al. 1993). This matrix effect results from the proximity of CH₄ molecules to N₂ molecules rather than other methane, thus affecting the band energy (Quirico & Schmitt 1997a). The (2–0) N₂ band seen on Triton at 2.17 μm is relatively broad and represents the hexagonal crystalline β-phase, which is stable at temperatures above the α–β transition at 35.6 K (Grundy et al. 1993). The α-phase has a cubic structure.

Triton has a tenuous atmosphere, probably arising from the vapor pressure of the most volatile ices at the prevailing temperature of about 38 K, making it the coldest known object orbiting a planet. Particulate material is injected into the atmosphere by plumes originating in the subsurface and rising vertically to about 8 km altitude, where they are blown laterally by sublimation-driven winds. A number of surface streaks of relatively low-albedo material result from the fallout from plumes that were inactive at the time of the Voyager 2 flyby.

Photochemical models of Triton's atmosphere and surface ices predict the formation and deposition of several hydrocarbons in addition to CH₄, as well as nitriles and other molecules (Krasnopolsky & Cruikshank 1995). With the possible exception of C₂H₆ (Quirico et al. 1999), whose abundance, however, amounts at most to a few percent (DeMeo et al. 2009), none of these have been reported thus far. Most of the overtone and combination bands of the other alkanes are coincident with those of CH₄ in the best-studied region 1.0–2.5 μm, thus masking their possible presence. Nitriles should be most easily detected in the region around 4.8 μm where the fundamental C≡N stretch occurs. Our observations with AKARI were carried out in the hope to find HCN (hydrogen cyanide or formonitrile) or other nitriles that should be produced either in the atmosphere or in the surface itself.

Disk-integrated spectra of Triton between 1.8 and 5.5 μm were recorded on 2007 May 13 (UT), using both the prism and grism modes of the Infrared Camera (IRC; Onaka et al. 2007) on board ISAS/JAXA's AKARI infrared astronomy satellite (Murakami et al. 2007), which was launched on 2006 February 21 (UT). We also acquired data at mid-infrared wavelengths, but the presence of Neptune made it impossible to detect the mid-infrared thermal emission of its largest moon. The spectra of the reflected sunlight alone, however, already allow a search for new species on Triton's surface, because ground-based spectra are adversely affected by the absorption by water vapor and carbon dioxide in Earth's atmosphere. In order to confirm our first findings, we repeated the measurements in NIR-grism mode during the warm phase of the mission on 2009 November 15 and 16 (UT). Figure 1 shows the best AKARI spectra of Triton in the region 4.0–5.0 μm.

Zoom In Zoom Out Reset image size

The region 4.0–4.35 μm in Triton's spectrum is strongly absorbed. It contains the overlapping fundamental bands of β-N₂ and CO₂, and has not been previously observed. An evaluation of the relative contributions of these (and possibly other) molecules to the absorption spectrum of Triton will be undertaken in subsequent studies; in the present work we focus on the contribution of HCN to Triton's spectrum. The presence of this molecule is demonstrated by the coincidence of the ν₁ stretching mode fundamental band at 4.76 μm in the spectra from both 2007 May 13 and 2009 November 16 (UT). The depth of this band on Triton is significantly greater than the random noise level in the data (±0.05 in geometric albedo). As the two spectra were obtained at different temperatures of the detectors and with different positions of the target on the detector array—the first observation was aimed at Neptune—spurious features are easily recognizable. The fact that the narrow band at 4.76 μm is observed in both sets of data suggests it is real.

A disk-averaged spectrum was also obtained in another module: the prism (NP; Table 1). NP and NG (grism) cannot observe simultaneously, therefore the two spectra were taken 15 hr apart from each other on 2007 May 13 (UT). The lower resolution near-infrared prism spectrum was found to be consistent with the grism observations (see Figure 2) if one takes the very different spectral resolving power into account. The minor differences may simply be due to an error in the flux calibration of the slitless spectrum or hint at spatial variations of Triton's surface (the apparent planetographic longitude of this moon varied by up to 71° between the observations). The sensitivity of the InSb detectors strongly decreases at wavelengths ⩾4.5 μm, i.e., a region where in addition to this the sunlight is quite weak. Hence, any error in the dark current subtraction has a bigger effect on this part of the spectrum and might have caused the slightly larger discrepancies among the different observations.

Zoom In Zoom Out Reset image size

Raw data were reduced using the appropriate version of the IRC spectral reduction pipeline (Lorente et al. 2007) for each data set. The pointings consisted of eight or nine exposures, and each sub-frame was scrutinized for severe damage. It was only rarely found and never at a pixel corresponding to 4.76 μm, our wavelength of interest. In order to achieve maximum signal-to-noise ratio, we rejected therefore none of the sub-frames from the grism.

The spectrum taken on 2009 November 15 was less useful for our purpose than the others, because Triton was only 3 pixels away from Neptune in the cross-dispersion direction. As the FWHM of the point-spread function (PSF) for the IRC-NIR bands amounts to 3.2 pixels and one pixel has a size of 146 × 146, this means that the Neptune/Triton flux ratio is too high at certain wavelengths. Nevertheless one finds narrow absorption features at 4.68 and 4.76 μm, if one extracts a spectrum at the position of Triton, but not at the same distance from Neptune on the other side. Unfortunately larger, but spurious variations appear at the off-source position so that the lines at the wavelengths of interest cannot be considered significant. Fifteen hours later Triton was 12'' away from Neptune in the cross-dispersion direction, however, and therefore its other spectrum taken in 2009 November is free of contamination from the planet.

An intermediate case is the NG spectrum from 2007, where Neptune was included in the field of view at a distance of 6'' in the cross-dispersion direction. Its contribution to Triton's spectrum was not negligible, but could be subtracted (see Figure 3). It was obtained from off-source spectra, extracted at a similar distance from the planet as Triton, but on the other side. As this distance had to be chosen with an integer pixel number, and the center of the images of both Neptune and Triton did not fall exactly on the center or the edge of a pixel, we used two different spatial positions next to each other and averaged them. Different weights were applied in order to obtain the value for the correct in-between position. Because the change in brightness of the spectrum along the cross-dispersion direction is quite steep, we suspect that a small over-correction is one of the reasons for the negative values present in the resulting spectrum from the grism. The spectrum from the off-position was smoothed by the FWHM of image PSF, i.e., 3 pixels.

Zoom In Zoom Out Reset image size

The contribution of Neptune to Triton in prism mode was found to be smaller than in the grism mode observation from the same day, because in the slitless spectrum Neptune stood "at shorter wavelengths" relative to Triton in the dispersion direction. Therefore the "blue" part of Neptune's spectrum could not affect Triton's spectrum, and any stray light from its features would affect its moon here at other wavelengths. As a consequence, NP provides an independent check of the reality of broad features in NG. Figure 3 also proves that the features of interest in our spectrum of Triton could not have been introduced by over- or under-correction of Neptune's contribution, because there are no counterparts with the same wavelength and width in the off-spectrum. We mention, too, that the measured reflectivity, calculated by assuming the Sun to be a black body of 5800 K, agrees at 2.5 μm within 10% with measurements from the ground (Cruikshank et al. 2000, Figure 1). As Neptune's flux density is at 2.5 μm about twice as high as at 4 μm, this shows that its contribution was removed for the most part also in the wavelength region presented here.

Due to the presence of Neptune, the spectrum of Triton could not be extracted at wavelengths beyond 6 μm. But spectra were obtained outside the nominal range of the SG1 dispersion element, i.e., shortward of 5.8 μm, and an absorption feature is present at 4.8 μm. As the extended range of the SG1 is not really trustworthy, however, we cannot give an accurate signal-to-noise ratio for this line.

2.1. Field of View

2.2. Wavelength Calibration

2.3. Post-helium (Phase 3) Mission

Triton's near-infrared spectrum from 1.4 to 2.5 μm has been analyzed by Quirico et al. (1999), who successfully modeled the ground-based observational data in great detail with a scattering theory incorporating the laboratory-determined optical constants of the ices of CO, CO₂, CH₄ dissolved in N₂, and H₂O. While our ultimate goal is to model the entire spectrum from 1.4–5.0 μm, we are hampered by the lack of optical constants for the critically important CH₄:N₂ component beyond about 2.8 μm (Quirico & Schmitt 1997a). Those data have recently become available and we will soon proceed to compute the complete model. In this Letter, we focus on the limited wavelength region 4.0–5.0 μm where CH₄ has no significant band absorption, the N₂ fundamental can be approximated by pure N₂, and where we find a feature coinciding with the signature of the ice of HCN. In Figure 4, we compare the average grism spectrum from Figure 1 with the spectra of CO, CO₂, and N₂ to demonstrate their positions and approximate contributions to the strength of the absorbed region in the Triton data.

Zoom In Zoom Out Reset image size

The model spectra of the various molecules in Figure 4 should be considered schematic only, presented to show the general effects of the strong N₂ and CO₂ bands across the range 4.0–4.5 μm, and the coincidence in wavelength of the CO fundamental at 4.68 μm and HCN fundamental at 4.76 μm with discrete bands in Triton's spectrum. Only hydrogen cyanide has a band at exactly this wavelength, and one other molecule comes close, viz., ¹³CO, which peaks at 4.78 μm in the α-phase, and at a slightly longer wavelength in the β-phase of N₂. Its larger distance from the absorption line of CO, our reference point, means that it can only contribute to the shoulder of the band detected with AKARI. So our experimental data suggest that CO isotopes could contribute to the feature, but do not account for most of it (see Figure 4). We can also rule out the possibility that the fundamental vibrational mode of HCN is shifted on Triton, in spite of the fact that protons in HCN can establish hydrogen bonds, either in pure HCN crystals or mixed with other molecules like H₂O. This phenomenon affects namely the C–H stretching fundamental mode, ν₃, which is known to be shifted by 200 cm⁻¹ in the pure HCN crystal with respect to the gas and the HCN molecule isolated in a noble gas matrix (Uyemura & Maeda 1972; Müller et al. 1993). However, because the normal vibrations are strongly decoupled in this molecule—the mass ratio between N and H atoms amounts to 14—the normal mode ν₁ approximates the stretching mode of C≡N and is insensitive to the hydrogen bond. Hence, the position of this vibration mode is independent from the environment, and the peak positions reported by different authors such as Moore & Hudson (2003) and Müller et al. (1993) are actually quite similar.

There are several other discrete absorption features in the Triton spectrum that could represent still unidentified molecules but do not constitute convincing detections. We note in particular that the deep absorption structure at 4.57–4.65 μm, adjacent to the strong CO band, bears close similarity to a band in several embedded protostars and is termed "XCN" (Pendleton et al. 1999). This band is now generally regarded as arising from the OCN⁻ anion, although as Pendleton et al. (1999) point out, isonitriles have absorption bands at exactly this wavelength as well. An isonitrile is a molecule with the functional group–N≡C, i.e., the CN functionality is here connected to the rest via the nitrogen and not, as in the proper nitriles, the carbon atom. Isonitriles were considered unlikely in astrophysical sources because of the lack of the corresponding nitrile bands, but on Triton at least the simplest nitrile HCN very likely exists, opening the possibility that the 4.57–4.65 μm band structure is due to an isonitrile. OCN⁻, on the other hand, is a radiolytic product of the proton irradiation of ices containing N₂ and a source of carbon, such as CH₄ or CO (Moore & Hudson 2003). In Figure 4, we show a spectrum of OCN⁻ from Moore & Hudson (2003) to demonstrate its potential for the identification of the corresponding absorption structure in Triton's spectrum.

We note that other identifying bands of HCN (e.g., 3.043 μm) fall in regions of heavy absorption by other molecules, or are too weak to be seen, consistent with the conclusions of Quirico et al. (1999) in the analysis of the shorter wavelength spectrum.

Hydrogen cyanide is produced in an ice mixture of N₂ + CH₄ + CO (100:1:1) at temperature 12 K by 0.8 MeV proton bombardment and UV irradiation (Moore & Hudson 2003), and is stable at the temperature of Triton's surface, as well as at significantly higher temperature. The production rate in terms of the number of molecules formed per 100 eV of energy deposited was found to be 0.025 for the trinary mixture of ices noted above and about 2.6× greater for the binary mixture N₂ + CH₄ (100:1). HCN is also efficiently produced in Triton's thin (≈15 μ bar surface pressure) atmosphere by a cyclic process involving CH₄ and N₂ plus a third body, with an estimated production rate of 29 g cm⁻² Gy⁻¹, some or all of which eventually precipitates to the surface (Krasnopolsky & Cruikshank 1995). HCN has not been previously detected on a solid surface, but it was found in cometary comae (Bockelée-Morvan et al. 2004).

After having produced the first spectrum of Triton between 4 and 5 μm and tentatively identified hydrogen cyanide, we looked for the signatures of other nitriles. Compounds of this class are particularly interesting, because they can form biological molecules, such as amino acids (Hudson & Moore 2003). From five different species—ethyl cyanide (CH₃CH₂CN), methylcyanoacetylene (CH₃C₃N), dicyanoacetylene (C₄N₂), cyanoacetylene (HC₃N), and methyl cyanide (CH₃CN)—only HCCCN exhibits an absorption band that coincides with a small line, at 4.83 μm (Khanna 2005), in the measured spectrum, and we do not claim to identify this material. The strongest bands of the more complex nitriles lie sufficiently far away from 4.76 μm to make a mix up with hydrogen cyanide impossible.

Our evidence for the presence of HCN also raises the intriguing question of whether the ejecta of cryovolcanism in Triton's south polar region get their color from brown hydrogen cyanide polymers, again with possible relevance for our understanding of the origin of life (Matthews 1995). We conclude that Triton offers fascinating possibilities for the study of processed surface molecules at infrared wavelengths.

We are grateful to R. Lorente, T. Onaka, and A. Salama for helpful advice on the data reduction.

Facilities: Akari - Akari (ASTRO-F)