EN ROUTE TO DESTRUCTION: THE EVOLUTION IN COMPOSITION OF ICES IN COMET D/2012 S1 (ISON) BETWEEN 1.2 AND 0.34 AU FROM THE SUN AS REVEALED AT INFRARED WAVELENGTHS^*

Obtaining accurate molecular production rates requires knowledge of excitation conditions in the coma. This establishes the fractional strength of each ro-vibrational band encompassed by the lines observed and is expressed through a rotational temperature (T_rot). On all dates, T_rot was measured by comparing the relative intensities of simultaneously observed H₂O lines after correcting each for telluric transmittance at its Doppler-shifted frequency. For ISON, the NIRSPEC data lacked sufficient signal-to-noise ratio to permit measuring T_rot for species other than H₂O, and CSHELL did not sample a sufficient range in rotational energy with adequate signal-to-noise ratio except in the H2O_3A setting. Consequently, on each date we assumed that the value of T_rot measured for H₂O applied to all molecules.

This assumption has been validated in comets in which we measured T_rot for multiple species (including H₂O; e.g., C/2007 N3 Lulin; see Gibb et al. 2012), independent of observational circumstances (i.e., heliocentric and geocentric distance). In general, we observe common T_rot among different species in a given comet, for comets having very different gas production rates. T_rot primarily depends on local densities and therefore collision rates in the coma—these in turn are most strongly dependent on gas production and are less dependent on geometry. Even for the relatively few cases of differing T_rot among molecules (e.g., 86 K for H₂O, and 76 K for HCN in C/2004 Q2 Machholz; Bonev et al. 2009), the values agree within their 1σ uncertainties. This also holds for species with greatly differing photodissociation lifetimes. For example, C/2002 T7 (LINEAR) showed T_rot consistent with 100 K at R_h = 0.67–0.71 AU, for both H₂O and H₂CO (DiSanti et al. 2006; Anderson 2010) even though the lifetimes of these two molecules differ by more than an order of magnitude (see Section 5.2.3, and note c of Table 2). We also find molecular mixing ratios (i.e., abundances relative to H₂O) to be relatively insensitive to the exact (common) value of T_rot.

Table 2 lists values of T_rot used in our analysis, either as measured or adopted (the latter are shown in parentheses). Our values are based on apertures centered on the nucleus (i.e., on the location of peak H₂O emission) and are thus heavily weighted toward the inner coma. They are consistent with the measured values of T_rot averaged over this "nucleus-centered" spatial region (see below) from the H2O_3A setting on November 17 and 22 (Paper 1) and on November 19 (Dello Russo et al. 2016).

We calculated "nucleus-centered" production rates (Q_nc) using our well-established formalism (e.g., see Bonev 2005; Villanueva et al. 2011, and Section 3.2 of DiSanti et al. 2014 for additional details). These are established for each measured line and are based on its flux F_line (W m⁻², corrected for telluric transmittance), the molecular photodissociation lifetime (τ₁, s), and line g-factor (g₁, W molecule⁻¹) at the measured (or adopted) value of T_rot (both τ₁ and g₁ correspond to values at R_h = 1 AU):

$$\begin{eqnarray}&&{Q}_{\mathrm{nc},\mathrm{line}}=\displaystyle \frac{4\pi {{\rm{\Delta }}}^{2}}{{\tau }_{1}f(x)}\displaystyle \frac{{F}_{\mathrm{line}}}{{g}_{1,\mathrm{line}}}.\end{eqnarray} \tag{ 1 }$$

In our calculations we adopt a gas outflow speed v_gas = 800 R_h^−0.5 m s⁻¹, based on velocity-resolved observations of several moderately bright comets at radio wavelengths (Biver et al. 2006, 2011; Agúndez et al. 2014; Cordiner et al. 2014). The parameter f(x) represents the fraction of all molecules in the coma contained in the beam, assuming release solely from the nucleus and uniform gas outflow in the coma—f(x) was originally formulated for a circular aperture (Yamamoto 1982) and later revised for square pixels (see Appendix in Hoban et al. 1991). This geometric factor includes molecular photodissociation, expressed through a scale length Λ = v_gas(R_h = 1 AU) × τ₁ × R_h^1.5. We list f(x) by species in Table 2 for our nucleus-centered beam and refer to this as fx_01, as it enters into the expression for the corresponding column density (also tabulated; see Table 2, note e).

In practice, for each molecule Q_nc represents the mean of its nucleus-centered production rates, which are proportional to F/g for each measured individual line—or group of lines within one or more adjacent resolution elements—weighting each by the inverse square of its stochastic error. Converting to total (or global) production rate (Q_tot) requires multiplying Q_nc by a "growth factor" (GF in Table 2), obtained through application of the "Q-curve" formalism to spatial profiles of molecular emission. Figure 3 shows representative emission profiles for Comet ISON (see Appendix A.3 for details of the application).

Zoom In Zoom Out Reset image size

Fitting the dependence of Q(H₂O) on heliocentric distance allows comparison with the canonical "insolation-limited" ${R}_{{\rm{h}}}^{-2}$ dependence expected for gas production governed by incident solar flux. Inspection of Figure 4 reveals that between October 22 and November 22, Q(H₂O) spanned two orders of magnitude while R_h decreased by a factor of 3.4; thus, a heliocentric dependence steeper than ${R}_{{\rm{h}}}^{-2}$ is evident.

To quantify this, we performed linear least-squares weighted fits to the water measurements. Including all 32 measurements resulted in a heliocentric dependence Q(H₂O)  ${\rm{\propto }}{R}_{{\rm{h}}}^{(-3.28\pm 0.18)}$, much steeper than the insolation-limited case. We also performed a fit excluding points associated with periods of enhanced activity, one around November 12–14 (Biver et al. 2013; Crovisier et al. 2013; Opitom et al. 2013b; Agúndez et al. 2014) and one around November 19 (Opitom et al. 2013a); we refer to these periods as "Events 1 and 2" (see Section 4.1.2). Accordingly, we omit the three points from November 15, the point from interval 2 and the two points from interval 3 on November 19, and the point from interval 1 on November 22; these seven points are enclosed in rectangular boxes in Figure 4. Our fit to the 25 remaining points, represented by the dot-dashed line, corresponds to Q(H₂O) α${R}_{{\rm{h}}}^{-(3.10\pm 0.09)}$, still much steeper than insolation limited. We take this fit to be our best measure of the long-term heliocentric dependence of water production over the range 1.2–0.34 AU. Our two fits agree within their 1σ confidence limits.

ISON was clearly in the process of shedding substantial amounts of material, particularly within R_h ∼ 0.7 AU. This was exemplified by two periods of enhanced water production (i.e., outbursts; following the terminology of Sekanina & Kracht 2014, we refer to these as "Events"). "Event 1" occurred in the range R_h = 0.7–0.6 AU, shortly prior to our November 15 observations, and the decrease in activity we observed from then to November 16 likely reflected its cessation. The report of lengthening "coma wings" in optical images from November 14 and 16 (Boehnhardt et al. 2013) provided further evidence of the dynamic evolution in activity experienced by Comet ISON around that time.

Our higher values for Q(H₂O) observed late on November 19 and early on November 22 bracket another period of increased activity ("Event 2," around heliocentric distance 0.4 AU). (In addition to the compilation of Q(H₂O) by Sekanina and Kracht, an increase in the production rate of sodium [by a factor >3] around the time of "Event 2," between UT November 19.5 and 20.5 [prior to and following our November 19 observations, respectively], was reported from optical spectra [Schmidt et al. 2015].) Substantial variations in activity were also recognized from Solar and Heliospheric Observatory (SOHO) observations of the highly extended H Lyα coma, from which H₂O production rates as high as (1.6–2) × 10³⁰ molecules s⁻¹ were reported for the period November 21–23 (Combi et al. 2014). The SOHO observations are highly complementary to our study: compared with our observations, they provide excellent temporal coverage, however, with lower temporal resolution (e.g., days) owing to their much larger field of view (i.e., effective beam size). Such a temporal lag relative to the IR measurements is to be expected, as is a higher Q(H₂O) measured by SOHO if, for example, additional H₂O was released from icy grains in regions of the coma outside those covered by the CSHELL slit.

In our ensuing discussion of the evolution of molecular abundances with Q(H₂O) (Section 4.2.2), we associate the points corresponding to anomalously high water production with "outbursts" that are likely connected with these two events. Similarly, we associate the 25 points used for the heliocentric fit shown in Figure 4 with periods we take to represent the "baseline" (i.e., longer-term) heliocentric evolution in activity.

It is worth comparing our values for Q(H₂O) with those reported for Comet ISON by Dello Russo et al. (2016). They present results from UV, optical, and radio, together with IR observations, using NIRSPEC and CSHELL. Of particular interest is that Dello Russo et al. also see an increase in water production after UT ∼ 20 hr on November 19, the date in common with the study reported here, and their mean value for Q(H₂O) on that date ((2.70 ± 0.30) × 10²⁹ s⁻¹) is consistent with the mean we find based on the six values listed in Table 2 ((2.58 ± 0.31) × 10²⁹ s⁻¹). We next present abundance ratios for trace species, noting that it is critical to compare not only production rates of trace molecules but also water production rates (see Section 4.2.3).

The first and only apparition of Comet ISON permitted serial measurements of its activity over an unprecedented range of heliocentric distances near and (especially) to well within R_h = 1 AU. The abundance ratios of several molecules were largely independent of R_h. In particular, CO, C₂H₆, and CH₄ (Figures 5(a)–(b)) showed relatively little variation with heliocentric distance. Most values were within 1σ (and all were within 2σ) of their respective mean values. The distribution of values for CH₃OH (Figure 5(c)) showed somewhat more scatter, and its mean abundance within R_h = 0.5 AU was larger than outside 0.5 AU (Table 3), by a factor of 1.6 ± 0.2. This exceeded stochastic expectations; thus, variability in the abundance ratio of CH₃OH with (presumed) depth in the nucleus cannot be ruled out.

Within heliocentric distance 0.5 AU, NH₃, H₂CO, and HCN showed substantially increased abundance ratios compared to their values at larger R_h (Figures 5(d)–(f)). NH₃ increased by a factor of at least 4.5, the lower limit to its increase being established by the ratio of its mean abundance from November 19 and 22 (3.51%) to the 3σ upper limit obtained by combining values from October 22, 24, and 25 and November 7, spanning R_h = 1.21–0.83 AU (Q(NH₃)/Q(H₂O) < 0.78%; see Figure 5(d) and Table 3). H₂CO increased by a factor of 5.6 ± 2.1 on November 19 compared to the mean of its abundances on October 25 and November 7 (0.16% ± 0.05%). HCN increased by a factor of 4.4 ± 1.0 compared to its abundance on November 7. The mean of C₂H₂ abundances on November 19 and 22 (0.24% ± 0.02%) was higher than that measured on November 7 (Figure 5(g)); however, this increase (by a factor of 2.1 ± 0.8) is less than 3σ and therefore, although suggestive, should be considered less secure than those of NH₃, H₂CO, and HCN.

The ratios of column densities of NH₂ to those of H₂O measured simultaneously (on November 7) or contemporaneously (on November 19 and 22) indicate that NH₂/H₂O increased from 0.28% at R_h = 0.83 AU to considerably higher values (approaching or exceeding 1%) inside 0.5 AU. The value on November 7 is somewhat lower than the "typical" value for NH/OH measured among comets (∼0.43%) based on the optical survey in A'Hearn et al. (1995). The much broader spatial distribution of NH₂ on November 22 compared to November 19 likely contributed to the different column measurements on these two dates. This is discussed further in Section 5.2.

In Figure 5, we show in red measurements obtained during periods of enhanced water production, on November 15 following "Event 1," and bracketing "Event 2" on November 19 and November 22. We found that overall the abundance ratios of molecules measured in Comet ISON were independent of the total H₂O production rate. Specifically, CO, C₂H₆, CH₄, and CH₃OH were measured during periods of both "baseline" and "outburst" activity, and overall their abundance ratios showed no systematic differences in this regard. For example, the abundance for CH₃OH at 0.46 AU was obtained during "baseline" activity and that at 0.35 AU was obtained during "outburst," yet these agree to well within their 1σ uncertainties (see also Table 2, and Figure 5(c)). Similarly, the abundance ratios measured for HCN (and for C₂H₂) on November 19 and 22 were in agreement even though the first measurements (at 0.46) corresponded to "baseline" water production while the second measurements (at 0.35 AU) were referenced to Q(H₂O) obtained during a period of "outburst" that was presumably associated with Event 2. The situation for NH₃ is inconclusive since it was measured only during "baseline" periods, on both November 19 and November 22. The comparison for H₂CO is also inconclusive since it was measured only once inside 0.5 AU (on November 19), at which time Q(H₂O) was apparently transitioning from "baseline" to "outburst" regimes.

Our combined 3σ upper limit for the CO abundance ratio from October 22 and 24 (2.13%) is consistent with values measured from CSHELL observations on November 15, 16, 17, and 22. It is also consistent with the value reported from Hubble Space Telescope observations on October 21 (∼1.5%; Weaver et al. 2013) and with the 3σ upper limit reported from preliminary analysis of the NIRSPEC observations on October 22 and 24 (2.0%; Mumma et al. 2013).

The reported preliminary value for CH₃OH from October 22 and 24 (1.1% ± 0.3%; Mumma et al. 2013) agrees with our result from these two dates combined (1.4% ± 0.5%). The preliminary value for C₂H₆ (0.21% ± 0.05%; Mumma et al. 2013) is lower than our corresponding combined value from October 22 and 24 (0.45% ± 0.09%), but is consistent (at the 1σ level) with our value for R_h > 0.5 AU (0.28% ± 0.02%; see Table 3). Our abundances for CH₄ and HCN, respectively, are consistent with preliminary values, both from October 22 and 25 (0.7% ± 0.3% and <0.1%, 3σ; Mumma et al. 2013) and from November 7 (0.4% ± 0.1% and 0.06% ± 0.02%; Paganini et al. 2013).

Faggi et al. (2015) obtained an upper limit (3σ) of 2.5 × 10²⁹ s⁻¹ for NH₃ production on UT November 26 (R_h ≈ 0.17 AU) based on the (1, 1) rotational line at 23.7 GHz, using the Medicina 32 m radio antenna. Extrapolating our heliocentric fit for water production (Q(H₂O) = 1.9 × 10²⁸ ${R}_{{\rm{h}}}^{-3.1}$; Figure 4) to 0.17 AU implies Q(H₂O) = 4.6 × 10³⁰ s⁻¹. This suggests an abundance ratio Q(NH₃)/Q(H₂O) < 5.5%, consistent with our values for NH₃ from November 19 and 22 (Tables 2 and 3).

It is instructive to compare the molecular abundance ratios we find on November 19 with those reported by Dello Russo et al. (2016), as this is the only date for which results are presented both here and in that paper. For the most part these agree within their 1σ uncertainties, the notable exception being that our value for CH₃OH (1.7% ± 0.2%) is substantially larger than that found by Dello Russo et al. (1.2% ± 0.3%). However, the Dello Russo et al. value is based on their overall mean H₂O production rate ((2.70 ± 0.30) × 10²⁹ s⁻¹), whereas our value uses the mean of the three water measurements spanning interval 1 on November 19 ((2.25 ± 0.08) × 10²⁹ s⁻¹; see Table 2 and discussion in Section 4.2). Comparing our methanol production rate with our overall mean Q(H₂O) from this date ((2.58 ± 0.31) × 10²⁹ s⁻¹) results in an abundance ratio for CH₃OH of 1.5% ± 0.3%, consistent with the Dello Russo et al. value at the 1σ level. Furthermore, our abundance ratio for CH₃OH on this date is consistent with those we measured on November 18 (sampling the ν₃ Q-branch at 2844 cm⁻¹) and 22 (sampling the same feature measured on November 19, peaked near 2982 cm⁻¹). Taking the ratios of our November 19 production rates for C₂H₆, HCN, and C₂H₂ (the three other molecules measured within interval 1) to our overall mean Q(H₂O) also revises our abundance ratios (Table 2) for these three species downward somewhat, to 0.29% ± 0.05%, 0.31% ± 0.05%, and 0.21% ± 0.03%, respectively, again consistent with the values reported by Dello Russo et al., and also with the values in Table 2 based on Q(H₂O) from interval 1 on November 19.

This overall agreement between our results and those of Dello Russo et al. lends credibility to both analyses, which were conducted independently, and provides an estimate of the systematic uncertainties associated with treating these data. Systematic errors are generally difficult to assess, and the independent approaches used to analyze the November 19 ISON spectra afford one such opportunity. Factors contributing to systematic uncertainties include molecular fluorescence g-factor, terrestrial atmospheric transmittance correction (especially telluric water burden), choice of continuum baseline level underlying and local to (in frequency) superimposed emissions, and absolute flux calibration (${\rm{\Gamma }}$ in Table 1).

As a working hypothesis, it is conventional to classify comets as "organics normal," "organics depleted," or "organics enriched," based on their abundances of the four organic molecules C₂H₆, CH₃OH, HCN, and C₂H₂ relative to H₂O (e.g., see reviews by Mumma & Charnley 2011; DiSanti & Mumma 2008; Bockelée-Morvan et al. 2004, and references therein). However, these taxonomic classifications depend on existing (i.e., current) mean or median abundances among measured comets and can evolve as the number of comets studied increases. For this reason, it is more meaningful to cite abundance ratios relative to H₂O rather than taxonomic status (depleted, normal, enriched) when comparing results among observers.

Table 3 provides a summary of abundance ratios for Comet ISON and for comparison includes current mean (or for some molecules, median) values that cluster around common values for a fraction of measured comets from the Oort Cloud—for purposes of discussion only, we refer to these as "current normal" values, or "current norms" (see Table 3 note c). The current norms for C₂H₆, CH₃OH, and HCN are consistent with those quoted previously (i.e., in the review papers cited above); however, that for C₂H₂ is lower by a factor of approximately two based on the larger (and continually growing) number of measured comets.

Species having abundance ratios that did not vary significantly with heliocentric distance (CO, C₂H₆, and CH₄; Figures 5(a)–(b)) were depleted in ISON, by factors of two or more compared to their current norms. This was also the case with CH₃OH for R_h > 0.5 AU; however, within 0.5 AU its abundance was closer to "normal." For R_h > 0.5 AU, H₂CO and HCN were also depleted; however, for R_h < 0.5 AU they were enriched by differing amounts (by factors of ∼3.6 and 1.8 relative to their respective current norms). The 3σ upper limit measured for NH₃ from 1.21 to 0.83 AU was consistent with its current norm, but for R_h < 0.5 AU it was enriched by a factor of 4.7. According to this taxonomic classification, at R_h = 0.83 AU C₂H₂ was consistent with its current norm, but was enriched by a factor of ∼2 within 0.5 AU. We note, however, that according to the review papers cited above the value for C₂H₂ at 0.83 AU classifies it as "depleted," while for R_h < 0.5 AU it is in the "normal" range.

This can be visualized by removing the canonical r⁻² spatial dependence. We averaged signal from diametrically opposite distances along the profile and multiplied each spatial point along this mean profile by its corresponding projected distance ρ from the nucleus. We refer to the resulting graphical representation as an "epsilon profile" or "epsilon plot," following terminology dating to the study of CN in the coma of Comet 1P/Halley (Komitov et al. 1989).

Epsilon plots so generated from the profiles in Figure 6 are shown in Figure 7 for emissions from all gaseous species and co-measured continua in three settings on November 19 (HCN_A, H2O_3C, H2CO_A; Figure 7(a)) and two settings on November 22 (HCN_A and H2O_3C; Figure 7(b)). These are color-coded by CSHELL setting (for the continuum) and by molecule, and all are scaled arbitrarily to a common maximum value. This makes it easier to compare shapes and trends with distance from the nucleus, taken to be the zero position on the x-axis.

Zoom In Zoom Out Reset image size

On November 19 (Figure 7(a)), the continuum epsilon profiles from the three included settings tracked each other closely, suggesting that the grain outflow behavior (and seeing) did not vary appreciably over the approximately 2 hr period encompassed by the observations. Their slow decline following a maximum near ρ = 10³ km shows that at larger offset distances the E-W average profile was somewhat steeper than ${\rho }^{-1}$.

On November 22 (Figure 7(b)) the observed difference between continuum curves at offsets less than 10³ km is expected given the poorer seeing conditions during the H2O_3C observations; compare the PSF traces from UT ∼ 18 hr and 23 hr UT in Figures 6(d) and (e), respectively. However, the continuum epsilon profiles decline differently for offsets beyond ∼1000 km. The shape of the HCN_A continuum on November 22 agrees closely with that seen on November 19 (with column density decreasing faster than ρ⁻¹), while the continuum in the H2O_3C setting indicates an approximately ρ⁻¹ falloff. The origin of that difference is likely connected with changing outflow during the 5 hr temporal gap between the two measurements.

Most emission profiles we measured in Comet ISON, including those of H₂O, were consistent with release directly from the nucleus, or at least were strongly dominated by direct release. However, some subtleties were evident. The relatively steep drop experienced by H₂CO with distance from the nucleus on November 19 (Figure 7(a)) was consistent with its short photodissociation lifetime (τ₁ ∼ 5 × 10³ s, compared with approximately 8 × 10⁴ s and 7 × 10⁴ s for H₂O and HCN, respectively; Huebner et al. 1992, pp. 102–113) and hence its more abrupt photodissociation. The "secondary" increase for H₂CO around 2 × 10³ km may indicate an additional contribution from H₂CO released in the coma, as was first seen in Comet 1P/Halley from Giotto spacecraft Neutral Mass Spectrometer observations (Eberhardt 1999) and subsequently inferred from ground-based radio observations of C/1996B2 (Hyakutake; Biver et al. 1999) and C/1995 O1 (Hale-Bopp; Wink et al. 1999). Distributed emission of H₂CO was subsequently inferred in additional comets observed in the radio (Biver et al. 2002). ALMA observations of ISON on November 17.5 reported a symmetric and narrow spatial distribution of H₂CO, and significant contribution from a distributed source having a parent scale length of 230–330 km was suggested (Cordiner et al. 2014). Secondary increases are also suggested for HCN and H₂O (Figure 7), but at less pronounced levels. The secondary peaks for HCN, H₂O, and H₂CO occurred at somewhat differing offset distances, which could result from varying gas production over the nearly 2 hr of clock time encompassed by the three settings that measured these molecules on November 19.

Of the profiles shown in Figures 3 and 6, those of NH₃ are the most surprising. Although the photodissociation lifetime of ammonia is relatively short (τ₁ ∼ 4 × 10³ s; Huebner et al. 1992, pp. 169–177), Figures 6(c) and (e) show that its profile was very broad, comparable to or broader than that of the much longer-lived H₂O (τ₁ ∼ 8 × 10⁴ s) co-measured in the H2O_3C setting on both November 19 and 22. It also shows a dip in its central region that is present on both dates and is unique among spatial profiles we observed for Comet ISON, including continuum profiles. This either could result from optically thick pumping near the nucleus or alternatively could indicate release of NH₃ from a halo or shell of (ice- or grain-dominated) particles relatively close to (e.g., within a few hundred kilometers of) the nucleus. That is, its broad nature could indicate that a fraction of NH₃ was released in the coma, particularly if optical depth effects were unimportant. If optically thin, this result is highly surprising, since, unlike H₂CO, distributed release of NH₃ has not been reported for any comet.

When we applied the Q-curve formalism to co-measured NH₃ and H₂O profiles using their respective photodissociation lifetimes, the terminal region—which establishes Q_tot from measured molecular profiles (see Appendix A.3)—for H₂O was relatively level on both dates, as expected for release predominantly from the nucleus. However, in the case of NH₃, using its (considerably shorter) lifetime resulted in production rate values that continued to increase with ρ throughout the terminal region of the Q-curve. Leveling the terminal region for NH₃ required using a significantly longer effective lifetime, τ₁ ∼ 2 × 10⁴ s (or larger). This implies effective scale lengths of (at least) 5.0 × 10³ and 3.3 × 10³ km for November 19 and 22, respectively, significantly larger than those expected for NH₃ assuming release solely from the nucleus (Λ ∼ 1.5 × 10³ and 1.0 × 10³ km for R_h = 0.46 and 0.35 AU, respectively).

A pronounced secondary peak for NH₃ occurred around ρ = (1.8–2.0) × 10³ km on November 19 and around 2.0 × 10³ km on November 22, but on both dates these were displaced from the (relatively minor) secondary increases seen for simultaneously measured H₂O. Clearly, in addition to potential release in the coma, optical depth effects must be included in a detailed treatment of NH₃ emission in ISON for R_h < 0.5 AU.

The situation for HCN is also interesting. Did its higher abundance ratio inside 0.5 AU result from chemical heterogeneity in the nucleus, or was it dominated by release in the coma? If the latter, the scale for such release appears to be small, within the seeing profile.

In some comets the spatial distribution of HCN suggested the presence of (at least) two phases of ice within the nucleus, one dominated by polar and the other by apolar bonds (Mumma et al. 2011, 2014; Villanueva et al. 2011, 2012b; Paganini et al. 2012; DiSanti et al. 2014). In Comet ISON, no clear distinction between profiles of polar and apolar molecules was apparent.

Inspection of Figure 6(b) indicates that on November 19 the profile of NH₂ was similar to that of co-measured HCN. On November 22 this was also the case in the sunward-facing hemisphere; however, it more closely tracked the continuum profile antisunward. The extent to which the increase in NH₂ between 0.83 AU and within 0.5 AU tracks that of NH₃, the canonical candidate progenitor molecule of NH₂, requires detailed modeling of the outflow. This is outside the scope of our present study.