The Peculiar Volatile Composition of CO-dominated Comet C/2016 R2 (PanSTARRS)

We obtained Director's Discretionary Time to observe R2 PanSTARRS with the powerful iSHELL IR spectrograph on the NASA IRTF on Maunakea, Hawaii, on UT 2018 January 29 and 30. While poor weather precluded obtaining useful data on January 29, we obtained high-quality spectra on January 30. The iSHELL detector is a 2048 × 2048 pixel Hawaii H2RG array with sensitivity over a wavelength range of ∼1–5 μm. As a cross-dispersed instrument, iSHELL measures signal in many (>10) consecutive echelle orders simultaneously with complete (for λ ≤ 4 μm) or nearly complete (for λ > 4 μm) spectral coverage, a significant improvement over the previous IRTF high-resolution facility spectrograph, CSHELL (Tokunaga et al. 1990). More details on iSHELL can be found in Rayner et al. (2012, 2016).

For our observations of R2 PanSTARRS, we used the 075 wide slit, which provides a spectral resolution of R ≡ $\tfrac{\lambda }{\delta \lambda }$ ∼ 38,000 for a uniform monochromatic source. We also observed an early-type IR standard star with the 4'' wide slit to serve as a flux calibrator and telluric standard (see Section 3.1). This slit provides lower spectral resolution (R ∼ 20,000) but minimizes slit losses and therefore systematic errors in flux calibration. Both the comet and standard star were observed using the classic ABBA nodding sequence, with a 75 telescope nod (half the slit length) along the slit between the A and B positions, located equidistant to either side of the slit midpoint. We employed two grating settings: M2 and Lp1. Setting M2 covers a wavelength range of ∼4.52–5.25 μm, encompassing spectral lines of CO, H₂O, and OCS, and Lp1 covers the wavelength range ∼3.28–3.65 μm and targets emission from CH₄, C₂H₆, CH₃OH, H₂CO, and OH prompt emission (a well-established proxy for H₂O production in comets; Bonev et al. 2006). Our comet observations resulted in 29.2 minutes on source in M2 and 59.8 minutes on source in Lp1. We obtained flats and darks at the end of each observing sequence for each grating setting.

Guiding was achieved through filter imaging with the slit viewer camera, performed in specific wavelength bands independent of the wavelength regime used to obtain spectra. The slit viewer allows active guiding on sufficiently bright targets while obtaining spectra. Short-timescale guiding is achieved through a boresight guiding technique, which utilizes "spillover" flux that falls outside the slit to keep the optocenter on the slit. However, while easily visible in the guider using a broadband J filter, R2 PanSTARRS was not bright enough for active guiding. We instead performed offset guiding using a reference (guide) star in the slit viewer field of view (FOV). Owing to the small nonsidereal rates of R2 PanSTARRS, this worked very well; we verified that the comet's position with respect to the slit remained very stable over the course of our observations, with minimal adjustments needed to keep it in the slit. More details relevant to cometary observations using iSHELL are presented in DiSanti et al. (2017).

Observations of the CO(2–1) line at 230.53799 GHz were performed with the ARO SMT on 2018 February 13 at 5:16 UT using the 1.3 mm dual polarization receiver with ALMA Band 6 sideband-separating mixers. The observations began 11 hr after the first Spitzer epoch. Data acquisition was done in beam-switching mode with a +2' throw in azimuth, and standard 6 minute scans were acquired. System temperatures had peak values of 390 K but, on average, remained under 340 K. The chopper wheel method was used to determine the temperature scale for the SMT receiver systems with a beam efficiency of η = 0.74. The back-end configuration that provided the best velocity resolution (0.325 km s⁻¹ channel⁻¹) consisted of a 2048 channel 250 kHz channel⁻¹ filter bank in parallel mode. The accuracy of the pointing and tracking was checked against the JPL Horizons ephemeris position and found to be better than <1'' rms. Due to high winds, only 12 scans were obtained, but the CO line was strong enough to be seen in single scans.

We obtained Director's Discretionary Time to observe R2 PanSTARRS with Spitzer IRAC on 2018 February 12 at 18:22 UT and 2018 February 21 at 01:03 UT in order to measure the CO₂ production rate. As Spitzer is well into its post-cryogenic mission, IRAC presently observes in two passbands: one centered at 3.6 μm and the other at 4.5 μm. Both filters have broad wavelength coverage, with bandwidths of 0.8 and 1.0 μm, respectively. The 4.5 μm band has been used extensively in the past for measuring CO₂ production rates in comets, as this bandpass includes the ν₃ band of CO₂ at 4.26 μm (e.g., Reach et al. 2013; McKay et al. 2016). It also contains the ν(1–0) band of CO at 4.7 μm, but in many comets in the AKARI survey (Ootsubo et al. 2012), the CO₂ feature was at least 10 times brighter than the CO feature, and so CO₂ is typically assumed to be the dominant gas emission feature in the IRAC 4.5 μm band. This is due to the fluorescence efficiency of CO₂ being approximately an order of magnitude larger than that for CO, coupled with the fact that the CO₂ abundance in comets is often equal to or greater than the CO abundance. There are examples, however, such as C/2006 W3 (Christensen) and 29P/Schwassman-Wachmann 1, where CO emission contributes significantly to the 4.5 μm band flux (Ootsubo et al. 2012; Reach et al. 2013). The high CO production rate found for R2 PanSTARRS (Biver et al. 2018; de Val-Borro et al. 2018; Wierzchos & Womack 2018) means that CO emission may not be negligible in the Spitzer imaging and may in fact dominate signal in the 4.5 μm channel. We discuss how we account for the CO contribution in the Spitzer imaging in Section 3.3.

We supply the details of our observations in Table 1. The IRAC array is a 256 × 256 pixel InSb array covering a 52 × 52 FOV with a spatial scale of 12 pixel⁻¹, which for our observations corresponds to a projected FOV of ∼500,000 km and a spatial scale of ∼1900 km pixel⁻¹. We employed a nine-position random dither pattern. Each bandpass is observed with independent arrays, such that when one array is observing the comet, the other is on an adjacent field. The sequence takes about 12 minutes to execute. For each epoch, we performed observations of the comet field several days after each cometary observation in order to image the field without the comet in it. These images are termed "shadow observations" and provide a measurement of the background to be subtracted from the cometary images. Our observations were obtained in high dynamic range (HDR) mode, which entailed obtaining exposures with both short (1.2 s) and long (30 s) exposure times in order to avoid saturation of the inner coma while still keeping a high signal-to-noise ratio (S/N) in the fainter outer coma. Observing in HDR mode also helps protect against saturation due to bright field stars. For these observations, no pixels were saturated; therefore, to optimize the S/N, we analyzed the longest exposure time images.

We observed R2 PanSTARRS with the Large Monolithic Imager (LMI) on the 4.3 m Discovery Channel Telescope (DCT) at Lowell Observatory from 2:51 to 4:13 UT on 2018 February 21. The observations were a target-of-opportunity (ToO) request that interrupted normal observations and began within 2 hr of the Spitzer observations in order to provide a near-simultaneous constraint on the water production of R2 PanSTARRS. The DCT ToO requests are limited to 2 hr, which allowed us sufficient time to focus the instrument, observe high- and low-airmass standard stars (listed in Table 1), and obtain ∼83 minutes on R2 PanSTARRS. Conditions were photometric, with seeing ∼12. The comet was ∼33° from a 26% illuminated moon, although no stray light attributable to the moon was evident in any of our images. The comet's airmass ranged from 1.06 to 1.23.

The LMI has a 123 × 123 FOV and 6.1 K × 6.1 K e2v CCD. On-chip 3 × 3 binning resulted in a pixel scale of 036 pixel^–1. We obtained images using a standard broadband SDSS r' filter and narrowband ion (CO⁺ central wavelength/bandpass width = 4266 Å/64 Å), gas (OH 3090/62, CN 3870/62), and dust continuum (UC 3448/84, BC 4450/67) filters that are all part of the comet Hale-Bopp set (Farnham et al. 2000). Single frames were acquired at the start and end of the sequence in the two filters with the highest S/N, r' (30 s) and CO⁺ (300 s). In between, sets of exposures were acquired in OH (three exposures each of 600 s), UC (3 × 300 s), and BC (2 × 120 s), with the sets proceeding from shortest to longest wavelength in order to minimize the effects of atmospheric extinction as the airmass increased. A single CN exposure (180 s) was also acquired at the end of the sequence. The telescope followed the comet's ephemeris rate for all comet images.

We obtained Director's Discretionary Time to observe R2 PanSTARRS with the ARCES instrument on the Astrophysical Research Consortium (ARC) 3.5 m telescope at Apache Point Observatory (APO) in Sunspot, New Mexico, on UT 2018 January 30, just hours before the iSHELL observations. ARCES provides a spectral resolving power of R = 31,500 and a spectral range of 3500–10000 Å with no interorder gaps. More specifics for this instrument are discussed elsewhere (Wang et al. 2003).

Observational details are described in Table 1. For all observations, we centered the 16 × 32 slit on the optocenter of the comet. We obtained six spectra of 1800 s each over the course of the night. These spectra were averaged after extraction and calibration to increase S/N. We obtained an ephemeris generated from JPL Horizons for nonsidereal tracking of the optocenter. For short-timescale guiding, the guiding software employs a boresight technique to keep the optocenter centered in the slit. We observed a G2V star in order to remove the underlying solar continuum and Fraunhofer absorption lines, a fast-rotating (vsin(i) > 150 km s⁻¹) B star to account for telluric features, and spectra of a flux standard to establish absolute intensities of cometary emission lines. The calibration stars used are given in Table 1. We obtained spectra of a quartz lamp for flat-fielding and a ThAr lamp for wavelength calibration.

Figure 1 shows raw spectral-spatial difference frames (total A-beam minus B-beam exposures) of R2 PanSTARRS for our M2 (panel (a)) and Lp1 (panel (b)) observing sequences. We applied our general methodology for processing IR spectra (e.g., Dello Russo et al. 2006; Villanueva et al. 2011; DiSanti et al. 2014). New techniques specific to iSHELL are described in detail in DiSanti et al. (2017; see also Roth et al. 2018). We provide a brief summary of our reduction procedures below.

Zoom In Zoom Out Reset image size

Processing of each iSHELL order produces a "rectified" spectral-spatial frame, meaning each column pertains to a unique wavelength and each row to a unique spatial location along the slit. An example is shown in Figure 1 for the region of Lp1 order 157 containing the cometary R0 and R1 lines of CH₄ (panel (e)). Such rectified orders consist of three parts: the bottom and top thirds show the comet signal obtained from A- and B-beam observations, respectively (in black), and the middle third shows the combined signal [(A+B)/2] (in white). Spectral extracts for the standard star and comet are obtained by summing the signal over a range of rows in the central (combined-beam) portion of their rectified frames.

To achieve absolute flux calibration and determine the column burdens of absorbing species in the terrestrial atmosphere, we fit a synthetic atmospheric transmittance model to the standard star spectrum for each processed order. We applied this optimized atmospheric transmittance model (calculated at the airmass of R2 PanSTARRS), convolved it to the spectral resolution of the cometary observations, and scaled it to the cometary continuum level (the continuum is virtually absent for our observations of R2 PanSTARRS). Subtracting the scaled model yields the net observed cometary emission spectrum, still multiplied by monochromatic atmospheric transmittance at the Doppler-shifted frequency of each cometary line. Correcting for transmittance and incorporating flux calibration factors from our standard star spectra allows establishing line fluxes incident at the top of the terrestrial atmosphere. Fully calibrated spectral extracts for R2 PanSTARRS showing emission lines of CO and CH₄ are shown in Figure 2. All other species searched for were not detected.

Zoom In Zoom Out Reset image size

We establish molecular column densities (or upper limits) by dividing these transmittance-corrected line fluxes by appropriate line-specific fluorescence g-factors, the values of which depend on rotational temperature (T_rot). For our study of R2 PanSTARRS with iSHELL, only CO presented enough lines with high S/N that spanned a sufficient range of rotational energy to obtain a measure of T_rot. Because the solar spectrum contains CO absorption lines, an accurate treatment required using "reduced" CO g-factors that incorporate the Swings effect for the heliocentric velocity of the comet at the time of our observations (−6.8 km s⁻¹; see Table 1). Our analysis of CO provided a best-fit value T_rot = 13 ± 2 K. We assume this temperature also applies to other species, as observations of brighter comets in which T_rot was measured for multiple species demonstrate that this is generally a valid assumption (e.g., Dello Russo et al. 2011; Mumma et al. 2011; Gibb et al. 2012; DiSanti et al. 2014).

We obtained molecular production rates as follows. We extracted a "nucleus-centered" spectrum by summing the signal over 15 rows (∼25) centered on the row containing the peak emission line intensity. Application of Swings-corrected g-factors and geometric parameters (R_h, Δ, beam size at the comet) to transmittance-corrected line fluxes provides the nucleus-centered production rate, Q_nc. However, owing primarily to seeing, Q_nc invariably underestimates the actual "total" (or "global") production rate, Q_tot. To obtain Q_tot, we multiplied each Q_nc by an appropriate growth factor (GF), determined through the well-documented "Q-curve" method for analyzing spatial profiles of emissions (Dello Russo et al. 1998). For each spatial step, a "symmetrized" Q-curve was produced by averaging the signal at equal but diametrically opposed distances from the nucleus. For our observations, only CO and CH₄ were detected. Each showed bright enough emissions to allow a reliable Q-curve analysis to be performed.

We present spatial profiles and symmetrized Q-curves for CO and CH₄ in Figure 3. For each Q-curve, the GF is depicted graphically as the level of the upper horizontal line (representing Q_tot) divided by that of the corresponding lower horizontal line (representing Q_nc). We refer to the region of the coma over which Q_tot is measured as the "terminal region."

Zoom In Zoom Out Reset image size

Comparison of the profiles in Figure 3 reveals a relatively broad (and, in particular, "flat-topped") spatial distribution for the observed CO emission. This is demonstrated by its much larger GF compared to CH₄, as well as that leveling its Q-curve required beginning two steps from the nucleus instead of one step, as was used for CH₄. This could indicate extended release of CO (e.g., from grains) in the inner coma and/or optical depth in the CO lines (particularly along lines of sight passing close to the nucleus). The high CO production rates reported from millimeter observations of R2 PanSTARRS with IRAM (Biver et al. 2018) and SMT (Wierzchos & Womack 2018), as well as from our iSHELL observations (see Table 3), suggest that the IR lines of CO are affected by optical depth.

Several previous CO-rich comets revealed optically thick emissions that in all cases were most pronounced for lines of sight passing through the innermost coma. For observations of C/1995 O1 (Hale-Bopp) and C/1996 B2 (Hyakutake) with CSHELL at the IRTF, observed column densities and Q-curves were corrected for opacity in the solar pump, assuming uniform gas outflow at constant speed (DiSanti et al. 2001, 2003). Observations of C/2006 W3 (Christensen) with CRIRES at the ESO/VLT (Bonev et al. 2017) at a similar observing geometry to our observations of R2 PanSTARRS also revealed optically thick CO for a production rate only slightly lower than what has been reported for R2 PanSTARRS. Bonev et al. (2017) developed a formalism for addressing optical depth effects in CO emission based on a curve-of-growth analysis, demonstrating through their Q-curve analysis that the effects of optical depth on retrieved production rates can be quantified (and thus corrected for).

Provided that the signal in the terminal region (i.e., the ends of the slit farthest from the comet optocenter) approximates optically thin conditions, the Q-curve will level out, as was observed for these three previously observed CO-rich comets, and as we also observed for R2 PanSTARRS (Figure 3). Although in this case, for CO, the central region is optically thick, the GF still allows establishing a reliable approximation of the actual Q_tot but with the GF for CO being significantly larger than the corresponding value for the optically thin CH₄ emission in R2 PanSTARRS (see Figure 3). For all of these comets (including R2 PanSTARRS), values of Q_tot(CO) as retrieved from optically thin and thick treatments are in formal agreement (i.e., they agree to within their respective 1σ uncertainties); see Figure A4 of DiSanti et al. (2001), Figure 4 of DiSanti et al. (2003), and Figure 4 of Bonev et al. (2017). We demonstrate this by applying the methodology detailed in the Appendix of DiSanti et al. (2001) to the observed CO emission in R2 PanSTARRS. The presence of optically thick CO emission in R2 PanSTARRS is supported by the much larger GF for the observed CO profile (2.93 ± 0.04) versus that for (optically thin) CH₄ (1.70 ± 0.19), a discrepancy that is resolved by correcting for optical depth in the CO lines (resulting in 1.67 ± 0.04; see Figure 3(b)). Therefore, we conclude that our Q-curve analysis for CO mitigates the effects of optical depth on our measured CO production rate. It also reinforces our decision to apply its "corrected" GF (1.7) to obtain a realistic upper limit for comeasured OCS and, similarly, to constrain Q_tot using the (identical within 1σ uncertainty) GF from CH₄ for comeasured, undetected species included in the Lp1 setting (C₂H₆, CH₃OH, and H₂CO). Derived production rates and mixing ratios are shown in Table 3.

Our spatial profiles also permit testing for asymmetric gas outflow in the coma. This is very pronounced for CH₄ (Figure 3(c)), indicating a large enhancement in the anti-sunward-facing hemisphere, particularly considering the relatively small solar phase angle of R2 PanSTARRS at the time of our observations (∼19°) and thus the potential high degree of projection onto the sky plane. The column density of CH₄ averaged between ∼2000 and 8000 km from the nucleus (projected on the sky) is a factor of 2.65 ± 0.55 larger in the anti-sunward direction compared to the solar direction, while the CO profile is much more symmetric, its corresponding ratio being only 1.03 ± 0.02.

It is possible that the asymmetry observed for CH₄ is associated with rotation of the nucleus. However, for the measured gas speed v_exp of 0.52 km s⁻¹ (from our SMT observations; see Section 3.2), the time required to exit the iSHELL terminal region (extending to 49 from the nucleus) is approximately 4.3 hr, longer than the clock times encompassed by our M2 and Lp1 sequences (obtained consecutively and together spanning 2.65 hr of elapsed clock time). This suggests that gas in the iSHELL slit was not replenished (at least significantly) between M2 and Lp1 observations, and thus the large differences in the degrees of asymmetry observed for CO and CH₄ cannot be explained by temporally variable outgassing. Nonetheless, these results indicate that the abundance ratio CH₄/CO was higher in the anti-sunward-facing hemisphere than in the sunward-facing hemisphere by a factor of ∼2.5; therefore, assuming little to no replenishment of coma gas between M2 and Lp1 sequences, this implies a significantly higher CH₄/CO abundance ratio in the anti-sunward direction at the time of our iSHELL observations.

Figure 4 shows the spectrum resulting from coadding all 12 scans on UT February 13. We calculated the column density assuming optically thin gas (appropriate for the relatively large beam diameter of 32'') with T_rot = 23 K (Biver et al. 2018; see Section 4 for discussion of differences with iSHELL), and we calculated the production rate assuming a simple symmetric outflow expansion model with V_exp = 0.52 km s⁻¹ consistent with our spectra and other detailed modeling of the spectral line profiles (Biver et al. 2018; Wierzchos & Womack 2018). We adopt this expansion velocity for all analysis in this work. The CO production rate is (5.5 ± 0.89) × 10²⁸ mol s⁻¹ (see Table 3).

Zoom In Zoom Out Reset image size

We combined all images of the same exposure time using the MOPEX software (Makovoz & Khan 2005). This process creates a mosaic in the rest frame of the comet from the individual images, averaging overlapping data together but ignoring cosmic rays and bad pixels. Two mosaics are created: one for the comet data, the other for the shadow (background) data. We subtracted the shadow mosaic from the comet mosaic to remove the background. This includes zodiacal light and celestial sources. While this removes background stars and most of the sky background, it may not completely remove the sky background (for instance, zodiacal light for a given R.A. and decl. varies with time). We used the sky value from the adjacent image of blank sky to remove any residual background that remained after shadow subtraction. The resulting images are shown in the first two columns of Figure 5.

Zoom In Zoom Out Reset image size

After the mosaic images were created and the sky background was subtracted, we removed the dust contribution from the 4.5 μm band flux, isolating the gas emission. As the level of dust contamination is minimal, as inferred from the 3.6 μm image being much fainter than the 4.5 μm image (see Figure 5), we simply applied a scaling factor of 0.9 to the 3.6 μm image and subtracted it from the 4.5 μm image to remove the dust contribution from the 4.5 μm image. The scale factor of 0.9 was derived from a model for cometary dust accounting for the expected contributions of reflected light and thermal emission at the heliocentric distance of R2 PanSTARRS, based on the empirical coma dust model of Kelley et al. (2016) and the dust color of comet 67P/Churyumov-Gerasimenko at 2.8 au (Snodgrass et al. 2017). The resulting dust-subtracted images are shown in the third column of Figure 5. The fourth column of Figure 5 shows the dust-subtracted images normalized by a ρ⁻¹ surface brightness distribution, where ρ is the projected distance to the peak brightness, which enhances coma asymmetries. The enhanced images show strong spiral structures, as well as an ion tail that we attribute to CO⁺ (see Section 4). A more detailed analysis of the coma morphology is beyond the scope of this paper.

As mentioned in Section 2.3, CO likely contributes significant flux to the Spitzer 4.5 μm channel, especially considering the large CO production rates measured for R2 PanSTARRS at millimeter wavelengths (Biver et al. 2018; Wierzchos & Womack 2018) and in the IR with iSHELL (Figure 3(b)). Also, CO₂ is observed with strong emission at this wavelength in many comets (Ootsubo et al. 2012), with few other likely contributors (see Section 4 for a discussion of other possible contaminating species). Therefore, we assume that the gas emission at 4.5 μm predominantly arises from CO and CO₂, and we separate their contributions with a three-step process: (1) we initially assume that 100% of the gas emission flux comes from CO molecules and derive a CO production rate; (2) we subtract the contemporaneous ground-based measured CO production rate, taking care to match the projected photometric aperture used for the Spitzer observations, which leaves a residual amount; and (3) we then recharacterize this residual as a CO₂ production rate.

From the dust-subtracted image of gas emission, we measured the flux for apertures ranging from 10 to 100 pixels (12''–120'') in radius. We converted the broadband photometry to CO line fluxes in photons following the IRAC data handbook (Laine 2015). The line fluxes were then used to calculate the total number of CO molecules inside the photometric aperture (N_CO,100%) using

$$\begin{eqnarray}&&{N}_{\mathrm{CO},100 \% }=4\pi {{\rm{\Delta }}}^{2}F\displaystyle \frac{{R}_{h}^{2}}{{g}_{\mathrm{CO}}},\end{eqnarray} \tag{ 1 }$$

where Δ is the Spitzer–comet distance, F is the observed photon flux, R_h is the heliocentric distance of the comet, and g_CO is the g-factor for excitation of the CO $1\to 0$ fundamental vibrational band, which is 2.5 × 10⁻⁴ photons s⁻¹ (Debout et al. 2016). Then, the production rate Q_CO,100% is given by

$$\begin{eqnarray}&&{Q}_{\mathrm{CO},100 \% }=\displaystyle \frac{2v}{\pi \rho }{N}_{\mathrm{CO},100 \% },\end{eqnarray} \tag{ 2 }$$

where N_CO,100% is the total number of CO molecules in the photometric aperture, v is the expansion velocity, and ρ is the projected radius of the photometric aperture. We assume an expansion velocity of the coma of 0.52 km s⁻¹, consistent with CO line widths from our SMT observations (see Section 3.2) and expansion velocities reported for R2 PanSTARRS by Biver et al. (2018) and Wierzchos & Womack (2018). This approach assumes a negligible effect of photodissociation on the spatial profile in the photometric aperture, which is justified, as our photometric apertures are <10% of both the CO₂ and CO photodissociation scale lengths. We calculated production rates for a variety of aperture sizes to quantify any trends in derived production rates with aperture size. Since we found the deviations between derived production rates for different aperture sizes to be minimal (<5%), for the rest of this paper, we quote production rates derived using a photometric aperture that matches the projected distance at the comet of the ARO SMT observation beam (see Section 2.2). This is a 14 pixel radius on UT February 12 and a 15 pixel radius on UT February 21. This choice minimizes systematic errors when subtracting the expected CO flux from the Spitzer images (see below).

If we assume that all of the gas flux in the Spitzer 4.5 μm image is due to CO and calculate its corresponding production rate using Equations (1) and (2), we find Q_CO,100% ∼ 1.6 × 10²⁹ mol s⁻¹, higher than the values derived from ground-based CO observations (Biver et al. 2018; de Val-Borro et al. 2018; Wierzchos & Womack 2018; this work). Therefore, we conclude that CO is probably not the sole contributor to the observed flux. However, it is a major contributor, and accounting for it does have a strong influence on the derived CO₂ production rate. While CO contributes significantly to the observed Spitzer fluxes, the excess we attribute to CO₂ is still >50% of the observed flux, so the detection of CO₂ is robust.

To calculate ${Q}_{{\mathrm{CO}}_{2}}$ and account for the CO contribution to the Spitzer 4.5 μm images, we employ our contemporaneous observations of CO at millimeter wavelengths obtained with the ARO SMT (Section 2.2). We favor using the contemporaneous SMT results over the IRTF CO measurements from several weeks earlier for subtraction of the CO contribution from the Spitzer imaging in order to minimize effects due to possible comet variability. Moreover, the production rates from SMT and Spitzer data were obtained using the same-sized photometric aperture, whereas the iSHELL observations cover a much smaller projected region of the coma, making it more sensitive to short-timescale changes in gas production such as rotational variation. This approach minimizes systematic uncertainties introduced by possible variability in the comet's activity.

We subtracted the millimeter wavelength–derived CO production rate, Q_CO,mm (see Section 3.2), from the derived Q_CO,100% value, which leaves a residual production rate: Q_residual = Q_CO,100% − Q_CO,mm. We attribute the residual gas production to CO₂ (see Section 4 for a discussion of other possible contaminating species). This residual production rate is then converted to ${Q}_{{\mathrm{CO}}_{2}}$ by taking advantage of the scaling relationship between production rates and fluorescence efficiencies,

$$\begin{eqnarray}&&{Q}_{{\mathrm{CO}}_{2}}={Q}_{\mathrm{residual}}\displaystyle \frac{{g}_{\mathrm{CO}}}{{g}_{{\mathrm{CO}}_{2}}},\end{eqnarray} \tag{ 3 }$$

where ${g}_{{\mathrm{CO}}_{2}}$ = 2.69 × 10⁻³ photons s⁻¹ (Debout et al. 2016). Using the SMT CO production rate, we derive ${Q}_{{\mathrm{CO}}_{2}}$ = (1.0 ± 0.1) × 10²⁸ mol s⁻¹ (Table 3). As a demonstration of the sensitivity of our derived CO₂ production rate to the assumed value for CO, if Q_CO = 0, then ${Q}_{{\mathrm{CO}}_{2}}$ ∼ 1.5 × 10²⁸ mol s⁻¹, while Q_CO = 1.0 × 10²⁹ mol s⁻¹ (closer to the iSHELL value) results in ${Q}_{{\mathrm{CO}}_{2}}$ ∼ 6.0 × 10²⁷ mol s⁻¹.

Lastly, we derived the Afρ value as a proxy for the dust production based on the 3.6 μm flux levels using the same photometric aperture as for the gas photometry and assuming all of the flux is solar continuum reflected off of dust particles in the coma (i.e., negligible emissions from gaseous species and no thermal emission). We follow the methodology of A'Hearn et al. (1984) to calculate Afρ. We derive Afρ values of 896 ± 27 cm on February 12 and 884 ± 27 cm on February 21, very low for a comet of this activity level and heliocentric distance. We derive log[Afρ/Q(H₂O)] = −23.54 ± 0.03, log[Afρ/Q(CO)] = −25.79 ± 0.04, and log[Afρ/Q(CO₂)] = −25.05 ± 0.04. For similar reasons discussed above, we compare Afρ to the CO production rate measured by SMT rather than IRTF.

Due to the high quality of our Spitzer data, uncertainties in the photometry are dominated by the absolute calibration uncertainty of Spitzer IRAC, which is approximately 3% (Reach et al. 2005).

The images were bias subtracted and flat-field corrected following standard practices, and all images for a given set (OH, UC, or BC) were median combined to improve S/N and mitigate background (stellar) contamination. Absolute calibrations and gas/dust decontamination for the narrowband filters were performed using extinction coefficients determined from the two standard stars and following the procedures outlined in Farnham et al. (2000), with sky values determined from regions near the corners of the CCD that appeared to be visually free of gas/dust/ions. The CN and UC images exhibited extended filamentary structures similar to those seen with the CO⁺ filter. Although the filters were designed to isolate gas (CN) or to be a nearly emission-free continuum area (UC), the CN bandpass includes ${{\rm{N}}}_{2}^{+}$ emission (Cochran & McKay 2018; Opitom et al. 2019), and both bandpasses contain emission from CO⁺ ions (Pearse & Gaydon 1976, and references therein). For most comets, these ions are much fainter than the intended gas and/or continuum and can be safely ignored, but this is not the case for R2 PanSTARRS. We thus conclude that the features we observed are ions and that UC and CN cannot be interpreted in their normal fashion. As a result, absolute calibrations of the OH images were performed using only the BC filter to define continuum, with solar color assumed. Flux-calibrated and decontaminated images are shown in Figure 6.

Zoom In Zoom Out Reset image size

Our normal procedure for determining gas production rates from fully calibrated images (e.g., Knight & Schleicher 2015) is to measure the flux in a circular aperture centered on the central condensation and convert to a production rate using a Haser model and standard assumptions about the appropriate lifetimes and scale lengths (A'Hearn et al. 1995). However, we elected to perform a more complex analysis, detailed below, due to several factors. First, the low S/N of our "pure" OH image meant that trailed stars or their negative residuals from the continuum image that was removed could significantly impact our inferred OH signal. Second, the aforementioned difficulties in decontaminating the image may have potentially led to over- or underremoval of the sky background and/or underlying dust continuum. Finally, we identified four lines of CO⁺ from the A²Π–X²Σ "comet-tail system" (Pearse & Gaydon 1976, and references therein) that fall in the OH filter bandpass. While no ion tail structure is evident by eye in our OH image, we were nonetheless concerned about possible ion contamination that was not accounted for in the standard comet filter reduction procedures.

In order to understand the extent to which our "pure" OH image might be contaminated by improperly removed continuum and/or ions, we extracted radial profiles from the uncalibrated CO⁺, BC, and OH images after removing the background. The radial profiles were binned in distance from the nucleus, ρ, and in azimuthal angle, with a resistant mean used to screen out anomalously high or low pixel values. The dust and ion tails were oriented nearly due east, at a position angle (PA) of ∼90°, slightly offset from the anti-sunward direction of PA = 79°, so we determined radial profiles for the sunward and tailward hemispheres, as well as 90° wedges centered at 0°, 90°, 180°, and 270°. We then fit slopes to the profiles from ρ = 4000–60,000 km (the inner radius was set to be about twice the seeing disk). The exact slopes retrieved are very sensitive to the background level, but the behavior of the slopes with azimuth is not. The OH slope was the flattest, falling off as roughly ρ^−0.7 in all four quadrants. The BC slope was steeper, falling as roughly ρ^−1.1 in all four quadrants. The CO⁺ exhibited the steepest slope and was the only one that varied significantly with PA, falling as ρ^−1.4 in the sunward quadrant and ρ^−1.0 in the other quadrants. The CO⁺ slopes are consistent with the expected behavior of ions whose sunward extent is minimal, while the consistency of the OH and BC slopes at all PAs suggests that each is relatively free of ion contamination. Furthermore, the more rapid falloff of BC and CO⁺ compared to OH suggests that, even if there were problems with improper decontamination of the OH image, the flux should increasingly approach the true OH signal at larger distances, with the sunward direction providing the least-contaminated OH signal.

This exercise also demonstrated that despite our previous concern about the low S/N of individual pixels in the OH image, with appropriate binning, a clear signal could be detected to at least ρ = 1.2 × 10⁵ km. Thus, we measured a radial profile in the sunward quadrant for the "pure" OH image and used this to determine the H₂O production rate on the assumption that the coma was spherically symmetric as follows. We created a synthetic OH profile using J. Parker and M. Festou's online version¹² of the vectorial model (Festou 1981) using the standard parameters given in Table 2 and scaling for the geometry of R2 PanSTARRS during our observations. We then interpolated the model to the midpoints of our radial profile ρ bins and scaled it up or down to minimize χ² from ρ = (4000–1.2) × 10⁵ km. We repeated the process but added a second parameter, a fixed background offset that was allowed to be positive or negative, to minimize χ² again. We continued to add complexity by including the BC and CO⁺ profiles, first individually and then together, ultimately testing all combinations of vectorial model, fixed background, BC profile, and CO⁺ profile. We repeated the fitting using an OH image that was flux calibrated with the continuum component set to zero as a further test of the reduction process.

Table 2.  Parameter Values for Optical Species

Molecule	Parent Lifetime (s)a	Daughter Lifetime (s)a	g-factor (erg s−1 molecule−1)a
CN	1.3 × 104	2.1 × 105	2.6 × 10−13
C2	2.2 × 104	6.6 × 104	4.5 × 10−13
NH2	4.1 × 103	6.2 × 104	6.40 × 10−15
O ib	8.3 × 104	⋯	⋯
O ic	1.3 × 105	⋯	⋯
OH	8.3 × 104	1.3 × 105	1.54 × 10−15

Notes.

^aGiven for r = 1 au. For CN and OH, given for $\dot{r}$ = 0 km s⁻¹ but varies with $\dot{r}$. ^bFor [O i] from dissociation of H₂O into H₂ and O; the branching ratio employed is 0.07 (Bhardwaj & Raghuram 2012). ^cFor [O i] from dissociation of OH; the branching ratio for H₂O to OH + H employed is 0.855 (Huebner et al. 1992), and the branching ratio for OH to O + H is 0.094 (Bhardwaj & Raghuram 2012).

Download table as:  ASCIITypeset image

For the DCT imaging, the best fit using all parameters is a water production rate, Q(H₂O), of ∼3 × 10²⁶ mol s⁻¹. The best fit for the BC scale factor was close to the solar color, implying that the dust was properly removed in the standard processing despite our earlier concerns. Similarly, the best fits for both images required a CO⁺ component, supporting our suspicion that there was some ion contamination. In all cases, the fits yielded Q(H₂O) between 1 and 5 × 10²⁶ mol s⁻¹, with the largest variations occurring for fits using only background fluxes (e.g., no continuum or ion component). We tested reasonable deviations from the standard vectorial model assumptions, such as varying lifetimes and velocities, but these only changed Q(H₂O) at the ∼20% level or less. The fits consistently show that the OH signal is much stronger than the CO⁺ and dust continuum beyond ∼20,000 km. Figure 7 shows a "pure" OH image using our best-fit parameters (top left panel), the residual flux after removing the modeled OH component from the "pure" OH image (bottom left), and the radial profiles in our best-fit model (right).

Zoom In Zoom Out Reset image size

The residual OH image has been highly stretched to emphasize subtle features but shows a hint of overremoval in the tailward direction and slight underremoval of the inner core. Despite the large number of uncertainties and assumptions needed to deal with this unusual comet, we are confident that we have constrained Q(H₂O) to less than 10²⁷ mol s⁻¹, with a most likely value of ∼(3.1 ± 0.2) × 10²⁶ mol s⁻¹ (Table 3). The uncertainty quoted is strictly for our modeled assumptions and does not include systematic uncertainties that are difficult to quantify but could be larger.

Table 3.  Production Rates and Abundances

Instrument	Species	Q (1027 mol s−1)	X/H2Oa (%)	X/COb (%)
IRTF iSHELL	CO	95.4 ± 9.1	(3.08 ± 0.35) × 104	⋯
	CH4	0.56 ± 0.07	181 ± 25	0.59 ± 0.09
	OCS	<0.23	<74	<0.24
	C2H6	<0.085	<27	<0.089
	CH3OH	<0.36	<116	<0.38
	H2CO	<0.49	<158	<0.51
	H2Oc	<78.4	⋯	<82
	H2Od	<72.9	⋯	<76
Spitzer IRAC	CO2e	10.0 ± 1.0	3230 ± 380	18.2 ± 3.5
	Afρ	890 ± 19 cm	−23.54 ± 0.03f	−25.79 ± 0.04g
ARO SMT	CO	55.0 ± 8.9	(1.77 ± 0.31) × 104	⋯
APO ARCES	N2	4.8 ± 1.1h	1550 ± 370	5.0 ± 1.0
	CN	<0.02	<6.5	<0.021
	NH2	<0.01	<3.2	<0.010
	C2	<0.021	<6.8	<0.022
	H2O (via [O i])	10.5 ± 0.5	⋯	11.0 ± 1.2
DCT LMI	H2O (via OH)	0.31 ± 0.02	⋯	0.32 ± 0.04
	Afρ	561 ± 6 cm	−23.74 ± 0.03f	−25.99 ± 0.04g

Notes.

^aCompared to the measured value from OH observations with the DCT, as this is the most robust measure of the H₂O production rate. ^bCompared to the iSHELL CO production rate except for CO₂ and Afρ, which are compared to the SMT CO value, as this value was measured contemporaneously and used to correct the Spitzer observations for the expected CO contribution. The N₂/CO is calculated from the optical observations of ${{\rm{N}}}_{2}^{+}$ and CO⁺ as described in the text. ^cDerived from H₂O lines in the M2 setting. ^dDerived from OH prompt emission in the Lp1 setting. ^eDerived accounting for the expected CO contribution based on the SMT observations using the methodology detailed in Section 3.3. ^f $\mathrm{log}(\mathrm{Af}\rho /{Q}_{{{\rm{H}}}_{2}{\rm{O}}})$. ^g $\mathrm{log}(\mathrm{Af}\rho /{Q}_{\mathrm{CO}})$. ^hCalculated using the N₂/CO ratio derived from optical observations multiplied by the iSHELL CO production rate.

Download table as:  ASCIITypeset image

Using the observed flux in the BC image, we calculated Afρ as described in Section 3.3, although for the DCT data, we used a smaller photometric aperture projected at the comet of 10,000 km, in line with the typical apertures employed by other optical dust photometry observations. We obtain Afρ = 561 ± 6 cm (Table 3).

Spectra were extracted and calibrated using IRAF scripts that perform bias subtraction, cosmic-ray removal, flat-fielding, and wavelength calibration. We removed telluric absorption features and the reflected solar continuum from the dust coma and flux calibrated the spectra, employing our standard star observations. We assumed an exponential extinction law and extinction coefficients for APO when flux calibrating the cometary spectra (Hogg et al. 2001). More details of our reduction procedures can be found in McKay et al. (2012) and Cochran & Cochran (2002). We determined slit losses for the flux standard star observations by performing aperture photometry on the slit viewer images as described in McKay et al. (2014). Slit losses introduce a systematic error in the flux calibration of ∼10%.

In our optical spectra, we report an analysis of five molecular species (N₂⁺, CO⁺, CN, NH₂, and C₂) and one atomic species (O i). Our analysis of CN, C₂, and NH₂ employs the same empirical fitting model employed by McKay et al. (2014), which utilizes a molecular line list from Cochran & Cochran (2002) to fit Gaussian profiles to observed emission features. For N₂⁺ and CO⁺, we adapt the model from McKay et al. (2014) to include CO⁺ by adding line positions from Kuo et al. (1986) and Haridass et al. (1992, 2000) and ${{\rm{N}}}_{2}^{+}$ line positions from Dick (1978). We integrate over the fits to measure the observed flux. While we detect multiple bands of CO⁺, we use the (2, 0) band for analysis, as it has the highest S/N.

For CN, C₂, and NH₂, these fluxes are converted to production rates using a Haser model in which the input scale lengths are modified to emulate the vectorial model and g-factors from the literature. The molecular lifetimes and g-factors employed are given in Table 2. For [O i], we employ the observed [O i] 6300 Å emission as a proxy for H₂O production using a similar Haser model (see McKay et al. 2012, 2014 for more details about the Haser models employed). While [O i] 6300 Å emission is often used as a proxy for H₂O production in comets (e.g., Morgenthaler et al. 2001; Fink 2009; McKay et al. 2018), it assumes that H₂O is the dominant oxygen-bearing species in the coma, which is not the case for R2 PanSTARRS (see Table 3).

Ions do not follow a Haser profile. Therefore, we do not calculate production rates from derived fluxes for ${{\rm{N}}}_{2}^{+}$ and CO⁺, only the relative ratio of ${{\rm{N}}}_{2}^{+}$/CO⁺ using

$$\begin{eqnarray}&&\displaystyle \frac{{N}_{{{\rm{N}}}_{2}^{+}}}{{N}_{{\mathrm{CO}}^{+}}}=\displaystyle \frac{{F}_{{{\rm{N}}}_{2}^{+}}}{{F}_{{\mathrm{CO}}^{+}}}\displaystyle \frac{{g}_{{\mathrm{CO}}^{+}}}{{g}_{{{\rm{N}}}_{2}^{+}}},\end{eqnarray} \tag{ 4 }$$

where N_x denotes the column density of species x, F_x is the observed flux of species x, and g_x is the g-factor for the observed transition of species x. The g-factors employed are ${g}_{{\mathrm{CO}}^{+}}$ = 3.55 × 10⁻³ photons s⁻¹ mol⁻¹ (Magnani & A'Hearn 1986) and ${g}_{{{\rm{N}}}_{2}^{+}}$ = 7.00 × 10⁻² photons s⁻¹ mol⁻¹ (Lutz et al. 1993).

For the ARCES observations, we show our detections of ${{\rm{N}}}_{2}^{+}$ and CO⁺ in Figure 8 and the detection of [O i] 6300 Å and [O i] 5577 Å emission in Figure 9. We do not detect CN or NH₂ emission and have a tentative detection of the C₂ bandhead at 5165 Å, all of which we show in Figure 10.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

To derive upper limits on production rates, we employ empirical spectral fits of comet C/2009 P1 (Garradd) at a similar heliocentric distance using the spectral fitting model of McKay et al. (2014) and scale these models so that the strongest lines would be present at the 3σ level (for C₂, we scaled the model so that it corresponded to the candidate bandhead feature). We then integrated over this model to obtain an upper limit on the observed flux and convert to the production rate using our Haser model as described above. We present the production rate upper limits for CN, NH₂, and C₂ in Table 3. Regardless of whether the inferred flux for C₂ is interpreted as an upper limit or a detection, the upper limit inferred for C₂H₂ is unchanged, since C₂ has multiple sources in cometary comae (Combi & Fink 1997). For ARCES observations, we compare to the iSHELL observations of CO because the data were taken contemporaneously and both are narrow-slit spectrometers sampling similar portions of the coma.

From the ARCES spectra, we measure an ${{\rm{N}}}_{2}^{+}$/CO⁺ ratio of 0.05 ± 0.01. We follow the arguments of Cochran & McKay (2018) and assume that ${{\rm{N}}}_{2}^{+}$/CO⁺ = N₂/CO (see Section 4 for discussion of the validity of this assumption). Therefore, we infer an N₂/CO ratio of 0.05 ± 0.01. Using the CO production rate derived from our iSHELL observations, we derive an N₂ production rate of 4.8 × 10²⁷ mol s⁻¹ (see Table 3).

We present the H₂O production rates derived from the [O i] 6300 Å emission measured by ARCES and OH emission from DCT in Table 3. There is a clear discrepancy (factor of ∼30) between the H₂O production rate derived from the [O i] 6300 Å emission and the value derived from our OH narrowband observations. However, the methodology for deriving water production rates from [O i] 6300 Å emission assumes that H₂O photodissociation is the dominant source of O i in the coma. While this is a valid assumption for most comets and has been used in the past to derive reliable H₂O production rates (e.g., Morgenthaler et al. 2001; Fink 2009; McKay et al. 2018), the large CO and CO₂ production rates found by both ourselves and other authors compared to our derived water production rate suggest that this is not a valid assumption for R2 PanSTARRS. Not accounting for contributions from the photodissociation of other oxygen-bearing species such as CO and CO₂ when they are more abundant than H₂O results in a large overestimate of the water production rate, which is what we observe when comparing to the production rate derived from our OH observations. As OH release into the coma is always dominated by H₂O, it is a more reliable proxy for H₂O production than [O i]. Therefore, we adopt the value of ${Q}_{{{\rm{H}}}_{2}{\rm{O}}}$ inferred from our OH observations as a more accurate measure of the H₂O production rate.

The O i photochemistry in comets has been of interest for some time (e.g., Festou & Feldman 1981; Cochran 2008; Decock et al. 2013, 2015; McKay et al. 2013, 2015), and the study of both the [O i] 6300 Å line and the [O i] 5577 Å line (which we also detect in our ARCES spectra; see Figure 9) in such a CO- and CO₂-rich comet could provide new insights into O i photochemistry in comets. We measure the flux ratio of the [O i] 5577 Å line to the sum of the [O i] 6300 Å and [O i] 6364 Å lines (the oxygen line ratio) to be 0.20 ± 0.03, in agreement with the value of 0.23 ± 0.03 measured by Opitom et al. (2019). This ratio is higher than typically observed but similar to other comets observed at heliocentric distances near 3 au (McKay et al. 2012, 2015; Decock et al. 2013). This is consistent with the large CO₂/H₂O and CO/H₂O ratios observed. However, a detailed study of the O i photochemistry in R2 PanSTARRS is beyond the scope of this paper and will be pursued in a future publication.

A summary of our resulting production rates and abundances is presented in Table 3. We pursued several methods for determining the water production rate: H₂O and OH prompt emission from iSHELL, OH emission from DCT, and [O i] 6300 emission from APO ARCES. The OH data from DCT provided the best measure of H₂O production; therefore, all mixing ratios compared to H₂O use the DCT-derived Q(H₂O) as the reference. While we detected [O i] 6300 emission, there is significant evidence that this emission is actually dominated by contributions from CO₂ and/or CO photodissociation rather than H₂O, so for R2 PanSTARRS, it does not serve as a reliable proxy for H₂O (see Section 3.5).

We employ our CO₂, H₂O, and CO production rates to calculate the active areas of the cometary surface using the sublimation model of Cowan & A'Hearn (1979); these are given in Table 4. Following arguments given in Bodewits et al. (2014) and McKay et al. (2017, 2018), we adopt the slow rotator model, for which every facet of the nucleus surface is in equilibrium with the solar radiation incident upon it. While the size of the nucleus of R2 PanSTARRS is not known, we convert these active areas to active fractions for a 1 and 10 km radius nucleus, which encompasses the range of most cometary nuclei for which measurements are available, including an upper limit of R < 15 km determined for R2 PanSTARRS from Wierzchos & Womack (2018).

As our CO₂ abundance is dependent on an analysis of broadband imaging, not spectroscopy, it is possible that there are other emission features in our bandpass besides CO₂ and CO, the species commonly assumed to dominate the flux. This would lead to an overestimate of the CO₂ production rate. The leading neutral candidates are OCS and N₂O. Hot bands of water in the 4.5–5.0 μm region would normally be important (Ootsubo et al. 2012), but the extremely low H₂O/CO and H₂O/CO₂ ratios for R2 PanSTARRS mean they can be neglected.

At ∼4.9 μm, OCS has a strong band and has been observed in comets at an abundance of approximately 0.1%–0.4% with respect to H₂O (Bockelée-Morvan et al. 2004; Mumma & Charnley 2011; Dello Russo et al. 2016b, and references therein). It is covered by the M2 grating setting in our iSHELL observations (Figure 1) and was not detected, placing an upper limit on its abundance ratio compared to CO of 0.24%. At this level, OCS is not a significant contributor to the 4.5 μm band flux.

The other neutral molecule with strong emission in the 4.5 μm bandpass is N₂O. To date, N₂O has never been detected in a comet, despite Infrared Space Observatory and AKARI observations that cover this wavelength range (Crovisier et al. 1997; Ootsubo et al. 2012). As of this writing, we are not aware of any reported detection of N₂O by the Rosetta instruments around 67P/Churyumov-Gerasimenko. However, the large N₂ abundance and the presence of an intrinsically strong N₂O band at ∼4.5 μm could perhaps indicate a strong contribution of this molecule to our Spitzer photometry. The N₂O emission is too far to the blue to be covered by our iSHELL observations in the M2 grating, and the region around the N₂O emission band is heavily affected by telluric CO₂ absorption. Therefore, we cannot definitively rule out the presence of N₂O emission in our Spitzer images. The g-factor for the relevant N₂O transition is 1.5 × 10⁻³ photons s⁻¹ mol⁻¹, meaning that an N₂O production rate of ∼1.8 × 10²⁸ mol s⁻¹ could account for the residual flux we attribute to CO₂. This suggests that even at ∼1.8 × 10²⁷ mol s⁻¹, N₂O could account for 10% of the flux we attribute to CO₂. However, N₂O is also very efficient at releasing O(¹S) into the coma (Huebner et al. 1992), and at these high production rates, we may expect it to dominate the O i photochemistry and drive the oxygen line ratio to be ∼1. However, we measure an oxygen line ratio of 0.20 (Section 3.5), suggesting that N₂O is not a dominant source of O i in the coma. While O i photochemistry in comets is not well understood, this, combined with a lack of previous detections in comets, likely implies that N₂O is not contributing much more than 10% to the residual flux we attribute to CO₂.

Lastly, the Δv = 1–0 band of CO⁺ is located at ∼4.6 μm. We do indeed observe a faint tail feature with structures reminiscent of an ion tail in both Spitzer epochs (Figure 5). We attribute this to the CO⁺ band based on the large CO production rate and the strong CO⁺ tails observed in the optical for R2 PanSTARRS. However, this tail is much fainter than the neutral emission and only appears obvious in Figure 5 due to the image enhancement process. Additionally, this band is covered by our iSHELL observations, yet it is not detected. Therefore, we conclude that CO⁺ cannot account for the observed excess emission that we attribute to CO₂.

While we cannot unequivocally rule out N₂O emission, we conclude that our Spitzer 4.5 μm band imaging is most likely dominated by a combination of CO₂ and CO emission. After accounting for the expected CO contribution based on independent observations with other facilities, we think that our analysis provides an accurate measure of the CO₂ abundance in R2 PanSTARRS.

Our N₂/CO ratio agrees well with the measurements made by Cochran & McKay (2018) in 2017 December, almost 2 months before our observations, as well as measurements by Biver et al. (2018) made about a week before our observations and Opitom et al. (2019) about 2 weeks after our observations. This suggests that there was little evolution in the N₂/CO ratio over the 2017 December–2018 February time frame.

Our iSHELL CO production rate agrees well with values determined by de Val-Borro et al. (2018) and Biver et al. (2018), whose observations were 1–2 weeks before ours (although these works used an asymmetric outgassing model, while we assume symmetric outflow). Our value is higher than that found by Wierzchos & Womack (2018) using the SMT in 2017 late December and 2018 mid-January, as well as the value we measure from our SMT observations. This could be due to a different FOV or the narrow nature of our slit. There is evidence from the millimeter observations of Biver et al. (2018) and Wierzchos & Womack (2018) that there is a significant sunward outgassing component, an enhancement also observed in our Spitzer imaging (see Figure 5). As the iSHELL slit was oriented along the Sun–comet line, we may have sampled this enhancement, which, coupled with our application of a symmetric outgassing model, may result in an overestimate of the CO production rate, but we do not observe a strong asymmetry in the CO line profile (see Section 3.1, Figure 3), meaning that the iSHELL observations were not sensitive to the asymmetry inferred from the millimeter and Spitzer observations. One reason for this discrepancy could be the limited spatial coverage of the iSHELL slit (slit width 075 × length ∼6'') compared to the much larger circular beams in the submillimeter (10''–30'') and the arcminute spatial scale covered by the Spitzer imaging. A narrow feature that is only slightly offset from the solar direction could be missed by the narrow iSHELL slit while still revealing an asymmetry that would be evident only on larger spatial scales. There is evidence from the coma morphology in the Spitzer images that the asymmetry manifests as a large fan-shaped feature (Figure 5, fourth column), perhaps implying that the asymmetry would not be as pronounced in the inner coma sampled by iSHELL.

In analyzing our iSHELL spectra, we adopt the gas expansion velocity (v_exp) measured by SMT (0.52 km s⁻¹), but as the iSHELL slit samples the very inner coma, it is possible that the gas may not have been fully accelerated to the speed observed at larger nucleocentric distances. An ∼30% smaller assumed v_exp (i.e., from 0.52 to 0.35–0.40 km s⁻¹) would bring the iSHELL and SMT results for Q_CO into agreement within their respective 1σ uncertainties. As the resolving power of our iSHELL spectra (λ/Δλ ∼ 4 × 10⁴) is insufficient to resolve the cometary lines, we obtain no information regarding v_exp; therefore, we cannot confirm or refute the possibility of a lower expansion velocity in the inner coma. However, the rotational temperature we measured for CO (13 ± 2 K) is significantly lower than that obtained for CH₃OH from IRAM of 23 K (Biver et al. 2018). This discrepancy may be due to photolytic heating of the coma as the gas expands. Based on this result, we might expect a smaller expansion velocity to be more relevant to the iSHELL observations, since the terminal region spanned lines of sight offset only ∼1''–5'' from the nucleus, in contrast to the much larger SMT/IRAM beam sizes (∼10''–30''). In any case, while absolute production rates are proportional to v_exp (assuming the beam size is small compared to the photodissociation scale length, which is the case for all species reported here), relative abundances (i.e., mixing ratios) are relatively insensitive to the adopted value of v_exp as long as a common v_exp is assumed for all species, which is the case in this work. Therefore, we would expect negligible changes in abundance ratios for the other species included in the M2 and Lp1 settings (e.g., Table 3).

The different rotational temperature measured from our iSHELL observations versus the IRAM observations of Biver et al. (2018) and adopted for our SMT analysis could affect our results. For the iSHELL observations, adopting T_rot = 23 K results in an ∼10% higher CO production rate, while adopting T_rot = 13 K for the SMT observations results in a CO production rate ∼30% higher than the value for T_rot = 23 K. Adopting the higher T_rot for the iSHELL data results in an ∼70% larger CH₄ production rate, 15%–20% less sensitive upper limits on CH₃OH and C₂H₆, a 15% more sensitive upper limit on H₂CO, and a 10% more sensitive upper limit on OCS. In terms of mixing ratios compared to CO, the CH₄/CO ratio is most affected, having a value 50% higher for T_rot = 23 K than for T_rot = 13 K. The upper limits on C₂H₆/CO and CH₃OH/CO change minimally (∼5%) for the different values of T_rot, while the upper limits on H₂CO/CO and OCS/CO are approximately 20% more sensitive for T_rot = 23 K. Analysis for the other instruments does not depend on T_rot, though the slightly higher CO production rates determined from both iSHELL and SMT data using the alternate rotational temperature would result in proportionately lower values of the mixing ratios compared to CO.

Our upper limit for CH₃OH is a factor of 3 lower than the detection by Biver et al. (2018; this is independent of the value of T_rot, as are all of the comparisons in this paragraph), which could be due to temporal and/or spatial variability in CH₃OH outgassing. The different projected areas observed for the IRAM and iSHELL observations could account for this discrepancy if a fraction of CH₃OH is released from icy grains in the coma. At the same time, our H₂CO upper limit is consistent with the detection of Biver et al. (2018). Our HCN upper limit (inferred from CN) is consistent with both the upper limit found by Wierzchos & Womack (2018) and the detection reported by Biver et al. (2018). Our H₂O production rate inferred from the OH DCT data is consistent with the upper limits reported by Biver et al. (2018) from observations of the OH 18 cm line at Nancay and by Opitom et al. (2019) using narrowband photometry from the TRAPPIST telescope.

At the time of this writing, our results are the first reported detections or upper limits on CO₂, CH₄, C₂H₆, C₂H₂, OCS, and NH₃ in R2 PanSTARRS. While Opitom et al. (2019) reported a lower limit on the ${{\rm{N}}}_{2}^{+}$/NH₂ ratio (a proxy for N₂/NH₃), this is a lower limit on the column density ratio in the slit, and because of the different spatial distributions of ions and neutrals, interpreting the column density ratio as an actual abundance ratio is complicated. Assuming ${{\rm{N}}}_{2}^{+}$/CO⁺ = N₂/CO, we have used our neutral CO measurements to convert the ${{\rm{N}}}_{2}^{+}$/CO⁺ to an equivalent N₂ production rate, which can then be compared to the production rates derived for other neutrals such as NH₃ and HCN, as was also done by Wierzchos & Womack (2018). With those caveats aside, the lower limit on N₂/NH₃ provided by Opitom et al. (2019) is consistent with our constraints.

Opitom et al. (2019) also detected ${\mathrm{CO}}_{2}^{+}$ emission and derived a ${\mathrm{CO}}_{2}^{+}$/CO⁺ ratio of 1.1 ± 0.3. They did not interpret this as the CO₂/CO ratio, as CO₂ can also contribute to observed CO⁺ emission. For this reason, Opitom et al. (2019) also expressed caution when interpreting the ${{\rm{N}}}_{2}^{+}$/CO⁺ ratio as a direct measurement of N₂/CO, as CO₂ photodissociation could contribute to the observed CO⁺ emission, and therefore the measured ${{\rm{N}}}_{2}^{+}$/CO⁺ ratio actually only provides a lower limit on N₂/CO. However, from our Spitzer observations of neutral CO₂, we derive CO₂/CO ∼ 18%, and at this low abundance, CO photoionization should be the dominant source of CO⁺ ions (Huebner et al. 1992). This appears to disagree with the large ${\mathrm{CO}}_{2}^{+}$/CO⁺ ratio observed by Opitom et al. (2019). The reason for the discrepancy between ion and neutral observations is unknown but may be related to our understanding of CO₂ and CO photochemistry.

The Afρ value we find for R2 PanSTARRS from our DCT BC imaging is lower than other measurements by Opitom et al. (2019), as well as from the cometary database developed by T. Noel¹³ quoted by Biver et al. (2018). There is some evidence for temporal variability in the observations of Opitom et al. (2019), so that may explain some of the discrepancy. Variable gas contamination by strong CO⁺ emission could also be present, especially for the broadband photometry quoted by Biver et al. (2018). The Afρ value derived from our Spitzer imaging is higher than all values reported in the optical, including our DCT observations, which were contemporaneous with the Spitzer epoch on UT February 21. However, the different wavelength regimes for the IR and optical data make it unclear how comparable the Afρ values actually are. For the Spitzer data, we employed a larger photometric aperture than the optical studies (28,000 km versus 10,000 km); using a smaller aperture in line with other studies yields Afρ ∼ 1000 cm, larger than what is found for the original aperture. While Afρ is independent of aperture size for ideal comae, the presence of a dust tail and acceleration of the dust can account for the decreasing trend with aperture size we observe for Afρ. Spectral reddening of light scattered by comae can vary with wavelength, especially over such a large spectral range (Jewitt & Meech 1986). Nevertheless, we calculate the effective BC (0.45 μm) to 3.6 μm spectral slope as ∼1.8% per 100 nm, similar to other comets observed in the NIR by Jewitt & Meech (1986).

In this section, we compare R2 PanSTARRS to other comets. We first compare to the cometary population as a whole, then specifically to other comets observed at similar heliocentric distances.

4.3.1. Cometary Population

Table 5 summarizes the derived production rates (or upper limits) for R2 PanSTARRS, along with mixing ratios relative to H₂O and CO, for 12 species. For comparison, the average values of these mixing ratios for the sample of Oort Cloud comets observed to date are also presented. Most values for R2 PanSTARRS are from this work, though for CH₃OH, HCN, and H₂CO, our observations only provide upper limits, while Biver et al. (2018) secured detections, so in Table 5, we present the detected production rates for these species from Biver et al. (2018).

Table 5.  Summary of Production Rates and Relative Abundances

Instrument	Species	Q (1027 mol s−1)	X/H2O (%)	Mean X/H2Oa (%)	X/CO (%)	Mean X/COb (%)
Spitzer	CO2	10.0 ± 1.0	3230 ± 380	17.0 ± 6.0	18.2 ± 3.5	425 ± 178
IRTF iSHELL	CO	95.4 ± 9.1	(3.08 ± 0.35) × 104	4.0 ± 0.9	⋯	⋯
	CH4	0.56 ± 0.07	181 ± 25	0.88 ± 0.10	0.59 ± 0.09	22.0 ± 5.5
	C2H6	<0.085	<27	0.63 ± 0.10	<0.089	15.8 ± 4.4
	OCS	<0.23	<74	∼0.25	<0.24	∼6.3
APO ARCES	N2c	4.8 ± 1.1	1550 ± 370	≪1?d	5.0 ± 1.0	<1?d
	NH3e	<0.01	<3.2	0.91 ± 0.30	<0.010	22.8 ± 9.1
	C2H2f	<0.021	<6.8	0.16 ± 0.03	<0.022	4.0 ± 1.2
DCT LMI	H2Og	0.31 ± 0.02	⋯	⋯	0.32 ± 0.04	2500 ± 560
IRAMh	CH3OH	1.12 ± 0.07	360 ± 32	2.21 ± 0.24	1.04 ± 0.08	55.3 ± 13.8
	H2CO	0.045 ± 0.007	14.5 ± 2.4	0.33 ± 0.08	0.043 ± 0.006	8.0 ± 2.6
	HCN	(4.0 ± 1.0) × 10−3	1.3 ± 0.3	0.22 ± 0.03	(3.8 ± 1.0) × 10−3	5.5 ± 1.4

Notes.

^aMean mixing ratio compared to H₂O in the sample of Oort Cloud comets observed to date. The uncertainties reflect the standard deviation in measured values. All values except for N₂ and CO₂ are from Dello Russo et al. (2016a). The CO₂ value is from Ootsubo et al. (2012). ^bMean mixing ratio compared to CO in the sample of Oort Cloud comets observed to date. The uncertainties reflect the standard deviation in measured values. References for abundances are the same as for H₂O detailed in footnote a. ^cDerived by multiplying the derived N₂/CO ratio from our ARCES observations by the CO production rate determined from our iSHELL observations. ^dDue to the lack of observations of N₂ in comets, the mean mixing ratios compared to H₂O and CO are not meaningful to calculate. Therefore, the number included is based on past observations discussed in Cochran & McKay (2018) and has a very large uncertainty (indicated by "?"). ^eDerived from our NH₂ upper limit assuming all NH₂ is released via NH₃ photodissociation (i.e., ${Q}_{{\mathrm{NH}}_{3}}$ = ${Q}_{{\mathrm{NH}}_{2}}$). ^fDerived from our C₂ upper limit assuming all C₂ is released via C₂ photodissociation (i.e., ${Q}_{{{\rm{C}}}_{2}}$ = ${Q}_{{{\rm{C}}}_{2}{{\rm{H}}}_{2}}$). ^gDerived from our OH narrowband observations. ^hProduction rates from Biver et al. (2018). Mixing ratios compared to CO are taken directly from Biver et al. (2018), and mixing ratios for H₂O are calculated using our derived H₂O production rate.

Download table as:  ASCIITypeset image

All detected species in R2 PanSTARRS are heavily enriched compared to H₂O. Here CO is enriched by 4 orders of magnitude, with N₂ enriched by a similar amount (though the small sample size of N₂ measurements in comets makes this difficult to quantify), while CH₄, CH₃OH, and CO₂ are enriched by a factor of ∼160–200. The H₂CO and HCN have less drastic enrichments of ∼40 and 6, respectively, though these are still extraordinary enhancements never before observed in a comet. We cannot draw any conclusions regarding the abundances of C₂H₆, NH₃, C₂H₂, or OCS compared to H₂O, though our results imply that while they could be heavily enriched compared to H₂O, this enrichment is not as drastic as that inferred for species such as CO and may be more in line with that observed for species such as H₂CO or HCN.

While a definitive comparison to H₂O is not possible for all species, the strong CO detection allows a comparison of all species to CO. We find that all species searched for are heavily depleted compared to CO, except for N₂, which is enhanced. Even for species that are not detected, the upper limits are sensitive enough to demonstrate heavy depletions. All depleted species are underabundant by at least 1–3 orders of magnitude compared to other comets.

A glimpse into a rarely observed (alternative) compositional taxonomy has been provided by R2 PanSTARRS, with CO replacing H₂O as the dominant gas in the coma. The measured abundance ratio CH₄/CO (∼0.6%; Table 3) approaches the mean CH₄/H₂O and C₂H₆/H₂O ratios among comets from the Oort Cloud (Bockelée-Morvan et al. 2004; Dello Russo et al. 2016b). In contrast, C₂H₆ is strongly depleted (by a factor of at least 6.6 relative to CH₄), with C₂H₆/CO rivaling or surpassing the level of depletion of C₂H₆ (relative to H₂O) measured for disrupted comet C/1999 S4 (LINEAR), a current "end member" in terms of its severe depletion in all reported volatiles with the exception of HCN (Mumma et al. 2001). Another peculiarity of R2 PanSTARRS is the large N₂ abundance, with N₂ being the dominant reservoir of volatile nitrogen, more abundant than even H₂O (only CO₂ and CO are more abundant than N₂). Typically, NH₃ and, to a lesser extent, HCN are the most abundant nitrogen-bearing volatiles in comets. However, in R2 PanSTARRS, NH₃/N₂ < 0.21% and HCN/N₂ = 0.08 ± 0.03%. So unless another, more complicated form of volatile nitrogen that is not constrained by our observations (e.g., N₂O, C₂N₂, CH₃CN) is present at significant levels, more than 99% of the volatile nitrogen in R2 PanSTARRS is contained in N₂. There are not many measurements of the NH₃/N₂ value in comets; however, this limit for R2 PanSTARRS is much lower than that measured for comet Halley (∼1000%) and closer to the derived values for several dense molecular clouds in star-forming regions (∼0.6%; Womack et al. 1992). The very low derived relative abundance of NH₃/N₂ is consistent with the suggestion by Wierzchos & Womack (2018) that R2 PanSTARRS formed in an environment with decreased photodissociation of N₂, leading to preserving or shielding of N₂ and inhibited production pathways of hydrogen-rich species, such as HCN and NH₃ (Hily-Blant et al. 2017).

We show additional mixing ratios compared to CO₂, CH₄, CH₃OH, H₂CO, and HCN in Table 6, with the mean ratio observed among comets listed in parentheses. A subset of this compilation is shown visually as histograms in Figure 11. For mixing ratios compared to CO₂, we employ the average CO₂/H₂O ratio in the AKARI sample (Ootsubo et al. 2012), while other species like CH₃OH have their average abundance compared to H₂O derived from ground-based IR studies (Dello Russo et al. 2016b). This means that CO₂ and the other species were not observed contemporaneously, and, in most cases, the sample of AKARI and ground-based IR observations sampled different comets. This means that interpretation of the average CO₂ abundance in Table 6 is limited by the degree to which the AKARI and ground-based IR studies provide a random sample of the cometary population.

Zoom In Zoom Out Reset image size

Table 6.  Additional Mixing Ratios for R2 PanSTARRS

Species	X/CO2 (%)	X/CH4 (%)	X/CH3OH (%)	X/H2CO (%)	X/HCN (%)
CH4	5.8 ± 0.9 (19)	⋯
CH3OH	11.6 ± 1.4 (13)	200 ± 28 (250)	⋯
H2CO	0.47 ± 0.09 (1.9)	8.0 ± 1.6 (38)	4.0 ± 0.7 (15)	⋯
HCN	0.04 ± 0.01 (1.3)	0.71 ± 0.20 (25)	0.36 ± 0.09 (10)	8.9 ± 2.6 (67)	⋯
C2H6	<0.89 (3.7)	<15 (72)	<7.6 (29)	<190 (190)	<2125 (290)
C2H2	<0.21 (0.94)	<3.8 (18)	<1.9 (7.2)	<47 (48)	<525 (72)
NH3	<0.1 (5.4)	<1.8 (103)	<0.9 (41)	<22 (276)	<250 (414)

Download table as:  ASCIITypeset image

Almost all mixing ratios in R2 PanSTARRS deviate from typical values by at least a factor of 3, although for some species that were not detected, we do not have sufficient sensitivity to rule out a normal abundance (e.g., C₂H₆/HCN). The HCN is universally depleted by at least an order of magnitude compared to all detected species except H₂O. Interestingly, CH₃OH/CO₂ and CH₃OH/CH₄ are the only mixing ratios that are similar (within a factor of 2) to the average values observed among comets. Perhaps this commonality between the peculiar R2 PanSTARRS and other comets can shed light on its origin. In any case, the almost universal peculiarity of the observed abundances in R2 PanSTARRS compared to other comets implies that this is not a case of one or two species being anomalous (e.g., CO being heavily enriched and H₂O being heavily depleted), but rather a complete composition fundamentally different from the ensemble of comets observed to date.

This strong deviation from a "typical" cometary composition is also illustrated in Figure 12. Over 80% of the volatile composition of R2 PanSTARRS is CO, while H₂O is relegated to the status of a trace volatile. Also, N₂ is much more abundant than in typical comets and more abundant than the other typical trace species (i.e., other than CO, H₂O, and CO₂) combined. As a fraction of the volatile inventory of R2 PanSTARRS, CO₂ is actually fairly close to typical comets.

Zoom In Zoom Out Reset image size

The Afρ of R2 PanSTARRS is much lower than that of other comets with such high gas production observed at similar heliocentric distances, though concluding whether R2 PanSTARRS is dust- or gas-rich depends on which volatile is used as the reference. The value of log[Afρ/Q(H₂O)] determined for R2 PanSTARRS is larger by a factor of 5–10 than most comets observed at R_h ∼ 3 au in the survey by A'Hearn et al. (1995), meaning that when H₂O is the comparison gas, R2 PanSTARRS is considered quite dusty. However, log[Afρ/Q(CO₂)] is at the low end of comets observed with Spitzer and NEOWISE (Bauer et al. 2015; M. S. P. Kelley et al. 2019, in preparation) and approximately a factor of 10 lower than any Oort Cloud comet in those samples. So compared to CO₂, R2 PanSTARRS is considered a gas-rich comet. While no systematic study comparing Afρ to CO has been done, the extremely high CO production rate suggests that if CO is the reference gas, R2 PanSTARRS is incredibly gas-rich, possibly the most gas-rich comet ever observed. Even in terms of the total gas production (typically dominated by H₂O but dominated by CO in the case of R2 PanSTARRS) compared to Afρ, R2 PanSTARRS is very gas-rich.

4.3.2. Comets at Large Heliocentric Distance

Our derived active areas and fractions for H₂O, CO₂, and CO are given in Table 4. While the nucleus size of R2 PanSTARRS is not known, the very large active area required for CO would require the nucleus to be quite large (>5 km) for all the CO to come from surface sublimation. There is also the possibility of an extended source of CO, which adds additional available surface area for CO to sublimate from. However, our Spitzer data (which are sensitive to both CO and CO₂, as discussed earlier) do not show evidence for increasing production rate with photometric aperture size, as would be expected for an extended source (Combi et al. 2013; Bodewits et al. 2014; McKay et al. 2015), arguing against an extended source larger than ∼10,000 km in radius (corresponding to ∼5 pixels in our Spitzer images, the smallest photometric aperture for which reliable photometry can be performed) for these molecules. We cannot definitively rule out the possibility of an extended source of smaller spatial extent than our Spitzer photometric apertures, though our modeling of the CO spatial profile in our iSHELL observations using only optical depth effects (see Section 3.1) provides some evidence against a smaller extended source.

The derived H₂O active fraction is consistent with other comets if the nucleus is fairly small (<3 km), but this would contradict the large active area needed to explain the CO production. For a large nucleus (>10 km), this would imply a very low active fraction for H₂O sublimation, much lower than other comets (Sosa & Fernández 2011; Lis et al. 2019). This is additional evidence that the low water production is not simply an artifact of the large heliocentric distance of R2 PanSTARRS at the time of observation but rather is part of the inherent composition of this comet.

Compared to other comets observed to date, R2 PanSTARRS has an extremely anomalous composition. In the previous sections, we have demonstrated that the large heliocentric distance can only explain a small portion of the observed composition. Additionally, the depletion of highly volatile species like CO₂ and CH₄ compared to CO cannot be explained by the heliocentric distance either. Therefore, the observed anomalous abundances are not solely a consequence of the heliocentric distance. It is not likely due to thermal evolution from repeated solar passages, as this would work to deplete the most volatile species, like CO, N₂, CH₄, and CO₂, not enhance them as observed. Therefore, the composition of R2 PanSTARRS likely reflects its composition when it was formed.

The most primitive molecular forms of carbon and nitrogen in the universe are CO and N₂, respectively. They are often the starting point of chemical pathways that result in the formation of more complex molecules, such as CH₃OH, NH₃, and HCN. Therefore, the large abundance of CO and N₂ in R2 PanSTARRS compared to these more complex molecules suggests that the region of the disk where R2 PanSTARRS formed was chemically inactive and shielded from photodestruction, leaving the volatile carbon and nitrogen in their simplest forms. Both CO and N₂ are also extremely volatile, as are CH₄ and CO₂. The presence of these molecules suggests that R2 PanSTARRS must have formed in the farthest reaches of the protosolar disk in order to retain these hypervolatiles. The presence of these hypervolatiles also suggests little depletion of these ices from repeated solar passages, despite R2 PanSTARRS likely not being dynamically new and therefore having likely experienced at least several passages through the planetary region. (A dynamical analysis by Opitom et al. 2019 found that 100% of their 1000 R2 clones experienced at least three perihelion passages of less than 3 au.) Wierzchos & Womack (2018) proposed that the CO, N₂, and HCN relative abundances in the coma may be explained by the comet forming in an environment of ∼50 K (though other models suggest that N₂ requires colder temperatures, around 20 K, to condense out of the gas phase; Drozdovskaya et al. 2016) with significant shielding for N₂. Our observations showing very high N₂ with very low NH₃ abundances may provide additional support for this model, and thus the comet may more closely resemble the composition of the nitrogen-bearing volatiles of dense molecular clouds (Womack et al. 1992) and young stellar objects (Gibb et al. 2004). The large abundance of CO and CO₂ in R2 PanSTARRS is typical of interstellar apolar ice mantles, and high N₂ abundances are also expected in these ice mantles (Gibb et al. 2004). However, interstellar ice grains also have an abundant polar component that has a large water ice abundance, resulting in H₂O ice being the dominant component of interstellar medium (ISM) ice grains (Gibb et al. 2004), unlike what we observe for R2 PanSTARRS.

Of the observed species, our measurements suggest that the dominant carbon-bearing molecules are CO and CO₂, while the dominant nitrogen-bearing molecule is N₂. While CO and CO₂ are the main reservoirs of volatile carbon in comets, NH₃ and, to a lesser extent, HCN are typically the dominant reservoirs of volatile nitrogen in comets. Assuming that CO and CO₂ contain the majority of the volatile carbon and N₂ the volatile nitrogen in R2 PanSTARRS implies a C/N ratio of ∼11, while most comets, for which NH₃ is the main volatile nitrogen carrier, have a higher C/N ratio of ∼20. The solar value for C/N is ∼3.4 (Lodders 2010), so while R2 PanSTARRS has a C/N ratio closer to solar than most comets observed to date, its coma is still deficient in nitrogen compared to the Sun.

Our results show that most of the volatile oxygen in R2 PanSTARRS is locked in CO and CO₂, not H₂O, as is typically the case. This may suggest that R2 PanSTARRS formed in a region of the protosolar nebula where the C/O ratio is >1, as chemical models predict that when carbon is more abundant than oxygen in the gas phase, most oxygen will be locked into CO and CO₂, leaving little to form H₂O. However, there is also evidence from both comets and protosolar disk models that most of the water in the solar system was inherited from the presolar cloud rather than formed in the protosolar disk (Cleeves et al. 2014; Altwegg et al. 2017). If accurate, then the lack of H₂O in R2 PanSTARRS could reveal details of how inherited H₂O was distributed throughout the protosolar disk, i.e., whether H₂O was distributed heterogeneously throughout the disk. Another possible reservoir for volatile oxygen is O₂, which was detected with a surprisingly large abundance by the Rosetta spacecraft at comet 67P/Churyumov-Gerasimenko (Bieler et al. 2015; Keeney et al. 2017) and also in archival data from the Giotto spacecraft at 1P/Halley (Rubin et al. 2015b). It is extremely difficult to detect O₂ remotely, and none of our observations are sensitive to O₂, so we cannot rule out the presence of a large amount of O₂ in R2 PanSTARRS. Production of O₂ at 67P was found to correlate well with H₂O production (Bieler et al. 2015; Fougere et al. 2016), and because of this, some theories for the origin of O₂ in comets invoke a strong tie to H₂O through either radiolysis or trapping in clathrates (e.g., Mousis et al. 2016; Dulieu et al. 2017; Laufer et al. 2017). Given the very low H₂O abundance in R2 PanSTARRS, this would suggest that the O₂ abundance should also be very low. However, given the very peculiar chemistry of R2 PanSTARRS and our limited understanding of O₂ incorporation into cometary nuclei, we do not consider this argument definitive proof against a substantial O₂ abundance in R2 PanSTARRS.

All of these implications only apply to the volatile component of R2 PanSTARRS. There are no constraints on the composition of the refractory component (i.e., dust), so we cannot make any conclusions about atomic abundances in the bulk (i.e., dust and ice) composition of R2 PanSTARRS.

Biver et al. (2018) suggested that R2 PanSTARRS may be a fragment of a differentiated Kuiper Belt body in order to explain the large observed hypervolatile abundances. The relative abundances of CO, CH₄, and N₂ we observe for R2 PanSTARRS do not match the surface spectra of Pluto (Protopapa et al. 2008), though the relationship between the surface composition and the interior of large Kuiper Belt objects (KBOs) like Pluto is unclear. A detailed analysis of the dynamics of creating collisional fragments in the Kuiper Belt after differentiation and then dynamically transporting these fragments to the Oort Cloud (as well as the expected volatile composition of such fragments) must be further investigated.

Such an anomalous composition brings up the possibility that R2 PanSTARRS has an interstellar origin. While the current orbit of R2 PanSTARRS does not suggest an interstellar origin, R2 PanSTARRS could be a comet captured from another Oort Cloud in the Sun's birth cluster (or a more distant planetary system, though this is less likely) during the solar system's earliest stages. Interstellar origins have also been suggested for other comets with peculiar compositions, such as 96P/Machholz (Schleicher 2008) and C/1988 Y1 (Yanaka; Fink 1992). It has been shown that there could have been an exchange of comets between Oort Clouds in the Sun's birth cluster (Levison et al. 2010). However, it is not clear whether these comets would be expected to be significantly different compositionally from "solar" comets, as both they and their host stars would have formed from the same nebular gas.