Ground-based Detection of Deuterated Water in Comet C/2014 Q2 (Lovejoy) at IR Wavelengths

The formation of HDO is strongly temperature dependent (for T ≤ 30 K), both in the gas phase (via ion–molecule reactions involving H₂D⁺; Willacy & Woods 2009; Ceccarelli et al. 2014; Willacy et al. 2015) and in the ice phase (via atom–addition reactions on grain surfaces). At the high temperatures of the inner protosolar nebula, the D/H ratio in water should reflect the protosolar value (Albertsson et al. 2014), while water that formed in the cold outer regions should be enhanced in deuterium relative to the protosolar value. Therefore, measurements of HDO in comets have the potential to improve our understanding of their formative conditions in the early solar system.

Over the past few decades, studies have revealed a diverse chemical composition among comets regardless of their dynamical family. There have been fewer studies of isotopic ratios in comets due largely to the lower abundances of the minor isotopologues and to technical challenges (mainly the lack of instrumental sensitivities sufficient to detect HDO), but recent advances have permitted improved statistics. Isotopic studies, particularly of HDO, are important for understanding the origin and history of cometary volatiles (for a review of results from Halley through 103P/Hartley 2, see Mumma & Charnley 2011).

New estimates of isotopic ratios in comets are beginning to comprise a statistically significant sample. Early results suggested that most comets are consistent with 2 VSMOW (where D/H is 1.558 × 10⁻⁴ and HDO/H₂O is 3.116 × 10⁻⁴), or about twice the value in Earth's oceans. Recent improvements in instrumental sensitivities and the availability of space-based instruments that can target HDO have enabled its measurement in moderately bright comets with the result that, much like the chemical diversity found for other molecules, the abundance of HDO relative to H₂O varies among both Jupiter Family and Oort Cloud (OC) comets. Whether the range of D/H ratios differs significantly among comets originating in different reservoirs is (as yet) not clear owing to inadequate statistics—especially for the ecliptic comets (Hartogh et al. 2011; Mumma & Charnley 2011; Altwegg et al. 2017).

We studied the volatile composition of the recently discovered OC comet C/2014 Q2 (Lovejoy), hereafter Lovejoy, whose favorable apparition facilitated the detection of HDO, and for which IR spectroscopy permitted the simultaneous detection of HDO and H₂O. We used the Keck Observatory on 2015 February 4–5, when comet Lovejoy was placed at 1.29 au from the Sun (heliocentric distance, R_h) and at 0.8 au from Earth (geocentric distance, Δ).

In this Letter, we present our measurements of the abundance of HDO relative to H₂O near perihelion, as well as an analysis of possible formation conditions that influenced the volatile fraction. We use our results to provide tests of chemical formation pathways of the detected molecules as related to conditions in the protoplanetary disk.

We conducted astronomical observations of comet Lovejoy on UT 2015 February 4 and 5—soon after perihelion passage (January 30), using the Near Infrared Spectrograph (NIRSPEC) at the 10-meter W. M. Keck Observatory (Keck II) atop Mauna Kea, Hawaii. Our observations of HDO occurred on February 4, yet our program also included a full chemical survey of the comet on both dates. Overall, we measured production rates for water and 14 trace species (HDO, OH, CO, CH₃OH, CH₄, C₂H₆, OCS, CN, HCN, NH₂, NH₃, H₂CO, C₂H₂, and an upper limit for CH₃D). A paper describing the overall chemistry of this comet is in preparation (L. Paganini et al. 2017, in preparation).

Lovejoy's large eccentricity (e = 0.998), original semimajor axis (a = 581 au), and high inclination to the ecliptic (i = 803) classify it as a nearly isotropic (long-period) comet originating from the OC (Nakano 2015) with Tisserand invariant T_J = 0.246. It was discovered with a faint coma on 2014 August 17 at R_h of 2.6 au, yet brightened rapidly prior to perihelion reaching 4th magnitude in 2015 January. On 2015 January 7, the comet passed relatively close to Earth (Δ = 0.47 au), increasing to 0.83 au on 2015 February 4–5. The relatively large geocentric velocity (∼34 km s⁻¹) during our observations provided optimal separation of Doppler-shifted cometary HDO lines and their terrestrial counterparts, thus granting excellent transmittance to the expected cometary emission lines. During the first half of February, the comet remained at around 5th magnitude, thus the combination of high brightness, appropriate geocentric velocities, and relatively close passage to Earth permitted the detection of volatiles with high fidelity—especially for spectroscopic techniques in the IR. Indeed, the Figure-of-Merit (FoM) for comet Lovejoy was 5.06 when observed (FoM = 10⁻²⁹ · QR · △⁻¹ · R_h^−1.5), well above values required for isotopic searches (typically, FoM > 1) and quite atypical for most cometary apparitions, that usually show lower productivity and less favorable observing conditions.

We used a slit configuration of 0432 × 24'' for the observations of comet Lovejoy, positioning the slit along the Sun–comet direction. Observations used the standard four-step nodding sequence "ABBA" to cancel thermal background and line emission from Earth's atmosphere. Initial data processing included the removal of high dark-current pixels and cosmic-ray hits, flat-fielding, spatial and spectral rectification, and spatial registration. After combining the A and B beams from the difference frames, we extracted a spectrum from the sum of nine spatial rows (178) centered on the nucleus (thus, our reduced cometary spectra sample is a 0432 × 178 rectangular extract). The extracted spectrum (Figure 1) contains emission lines of cometary gases and continuum radiation associated with cometary dust grains. We isolated the emission lines by subtracting a transmittance function synthesized for the terrestrial atmosphere (and normalized to the cometary continuum) from the combined spectrum and by fitting the spectral transmittance using the line-by-line radiative transfer model (LBLRTM; Clough et al. 2005), which accessed the HITRAN-2012 molecular database.

Zoom In Zoom Out Reset image size

Weather conditions outside the dome were characterized by temperatures slightly below freezing, relative humidity ∼14%–20%, pressure ∼610 mbar, and wind speed of ∼70 mph. Image quality (seeing) was smaller than 08 FWHM, and the airmass was 1.14–1.65. We used spectra of a flux standard star (BS1641, using the 0720 × 24'' slit configuration) to calibrate our data.

Our observing strategy allowed sensing HDO simultaneously with water, thus minimizing possible systematic effects that may result from observations on different days, with different telescopes, or both.

We obtained H₂O production rates by first determining the flux of each ro-vibrational transition detected within our sampled aperture, F (W m⁻²), together with an optimum value for rotational temperature (T_rot). The rotational temperature is obtained for an individual species by comparing modeled and observed fluxes for lines that span a sufficient range of rotational energies. The modeled spectral line intensities are derived from our custom quantum mechanical fluorescence models, which provide fluorescence efficiencies (g-factors, g) at the specified rotational temperature. This process involves modifying a test temperature iteratively until a satisfactory fit (least squares) between modeled and observed line fluxes is obtained. This optimum value for T_rot (and its uncertainty) then permits a robust measure of production rate. On February 4, we detected 25 water emission lines spanning a range of rotational energies (Table 1) sufficient to permit an accurate retrieval of T_rot and production rates. The resulting H₂O rotational temperature (78 ± 1 K) applies to our nucleus-centered aperture of 3 by 9 pixels (equivalent to 043 × 178, or 266 × 800 km at the comet). The rotational temperature obtained from measurements of CH₃OH at millimeter wavelengths is T_rot = 67 ± 0.7 K (and a gas temperature of 73 ± 1 K; Biver et al. 2015), the difference between IR and millimeter results could be related to the different beam sizes. We assumed the same T_rot for HDO (78 ± 1 K).

Table 1.  Molecular Abundances for H₂O and HDO in Comet C/2014 Q2 (Lovejoy)^a

2015 February 4 (UT 4:55–7:23)
bRh = 1.29 au; Δ = 0.83 au; Δ-dot : 33.6 km s−1; Total time on source: 60 min
Species	νc	Lines	Trot	Nucleus Qd	Global Qd
	(cm−1)		(K)	(1027 s−1)	(1027 s−1)
H2O	3368.96	25	${{78}^{+1}}_{-1}$	312.65 ± 5.95	594.04 ± 12.92
HDO	2718.00	8	(78)e	0.19 ± 0.05	0.36 ± 0.10

Notes.

^aUncertainties represent 1σ. The reported error in production rate includes the line-by-line scatter in measured column densities, along with photon noise, systematic uncertainty in the removal of the cometary continuum, and (minor) uncertainty in rotational temperature. ^bR_h: heliocentric distance; Δ: geocentric distance. Δ-dot: geocentric velocity. ^cMean wavenumber of all emission lines (used for this reduction) from a particular species. ^dGlobal production rate, after applying a measured growth factor of 1.9 ± 0.2 to the nucleus-centered production rates. Optical effects were not observed in our spectra, and thus we assume that using the same growth factor for H₂O and HDO is a correct strategy. Values for g-factors (photons molecule⁻¹ s⁻¹): H₂O = 8.39E–07, HDO = 1.37E–05. ^eWe adopted T_rot = 78 K when calculating the production rates for HDO, since T_rot determination was not possible. This is indicated as (78).

Download table as:  ASCIITypeset image

To obtain production rates, we consider the g-factors for individual molecular lines at the appropriate T_rot, as well as the specific heliocentric velocity (v_h), the monochromatic terrestrial transmittance (T_j) at the Doppler-shifted position of the cometary line, the molecular lifetime (τ), the geocentric distance Δ, and the fraction of total molecular content in the coma sampled by each pixel, f(x). We also assumed uniform decay probability with nucleocentric distance and an outflow velocity (0.85 km s⁻¹) obtained with the Odin submillimeter space observatory (Biver et al. 2016). The nucleus-centered production rate was calculated using

$$\begin{eqnarray*}&&{Q}_{{nc}}=\displaystyle \frac{4\pi {{\rm{\Delta }}}^{2}{F}_{\mathrm{lines}}}{\tau {{gT}}_{j}f(x)}.\end{eqnarray*}$$

IR observations are impacted by slit losses (i.e., loss of flux) that result from atmospheric "seeing" and slight aperture effects (e.g., imperfect centering of the cometary photo-center in the slit). To correct for these effects, we measured a growth factor (GF) in Q from the observed spatial profiles after averaging Q at diametrically opposite positions along the slit, as this averages asymmetries in outflow (following Xie & Mumma 1996). Multiplying the nucleus-centered production rate (Q_nc) by GF of 1.9 ± 0.2 resulted in total (or "global") production rates (Q_tot).

Our estimate of H₂O production rate (Q_tot or Q) is 5.9 ± 0.13 × 10²⁹ s⁻¹, while Q(HDO) is 3.6 ± 1.0 × 10²⁶ s⁻¹ (see Table 1). The resulting HDO value is well above the 3σ detection threshold after 1 hr integration on source and is obtained after combining eight lines at spectral positions where our model matches flux resulting from observed HDO emission regions within the spectral order under evaluation (Figure 2). The resulting water and HDO production rates allow the retrieval of the D/H ratio in Lovejoy's water (3.02 ± 0.87 × 10⁻⁴, or 1.94 ± 0.56 VSMOW).

Zoom In Zoom Out Reset image size

Biver et al. (2016) reported observations of water in comet Lovejoy with the Odin Space Observatory between January 30 and February 3 (after perihelion). ${\rm{Q}}({{\rm{H}}}_{2}{}^{16}{\rm{O}})$ decreased from 1.1 × 10³⁰ s⁻¹ to 5.5 × 10²⁹ s⁻¹, i.e., a maximum factor of 2 within four days (accounting for the error estimates, see Figure 2 in Biver et al. 2016). Interestingly, estimates of ${\rm{Q}}({{\rm{H}}}_{2}{}^{16}{\rm{O}})$ show also a strong variation within a given date (January 30). Considering such variations, our estimates of water production after perihelion are in agreement with Odin values. The strong variations reported by Biver et al. (2016) demonstrate the vital importance of minimizing systematic errors by measuring HDO and H₂O simultaneously with the same telescope/instrument combination.

Could there be variations of HDO release rates pre-perihelion versus post-perihelion? Even though estimates of D/H in the radio and IR used a similar Q(H₂O), the HDO production rates derived from IRAM and Keck observations differ significantly. Our IR estimate for Q(HDO) (3.6 ± 1.0 × 10²⁶ s⁻¹) is larger than the IRAM results pre-perihelion by a factor of two to three. Thus, the NIRSPEC ratio (D/H = 3.02 ± 0.87 × 10⁻⁴, or 1.94 ± 0.56 VSMOW) agrees only at the 2σ level with radio results of Biver et al. (D/H = 1.4 ± 0.4 × 10⁻⁴, or 0.89 ± 0.25 VSMOW). Two avenues could explain this discrepancy: (1) the D/H changed after perihelion, or (2) systematics may strongly influence the radio estimate of D/H in cometary water.

The first avenue argues whether seasonal or evolutionary effects could influence the observed deuterium enrichment in comets. Moores et al. (2012) reported that the D/H ratio in water vapor released from water ice mixed with dust decreased with time during sublimation. On similar experiments under controlled sublimation conditions (temperature and pressure), Lécuyer et al. (2017) indicated that such decrease is not observed on samples of pure water ice (i.e., with no dust particles). While the Lécuyer et al. experiments showed no change, the experiments did not account for the case of water mixed with dust, which would best replicate the case of cometary conditions and sublimation. Recent experiments have shown that gas-surface mechanisms may produce a temporary increase in deuterated water species due to energetic water ions, especially at short heliocentric distances (Yao & Giapis 2017).

Even though Moores et al. showed a decrease in the D/H with time and Yao & Giapis a possible instantaneous deuterium enrichment due to energetic water ions, in situ results from Rosetta do not reveal seasonal variations in HDO/H₂O or D₂O/HDO (Altwegg et al. 2017). Given the limitations of our 1 hr measurement of D/H in water, we cannot test this hypothesis. A proper assessment of possible evolutionary effects on the D/H ratio in a comet would require cadenced astronomical observations using the same facility/technique to avoid possible systematics.

Before perihelion, Biver et al. (2016) reported observations of the HDO (2₁₁–2₁₂) line at 241.562 GHz. They observed comet Lovejoy using the IRAM 30 m radio telescope during two periods: on 2015 January 13, 15, and 16 (Δ = 0.496–0.528 au) and on 2015 January 23–26 (Δ = 0.624–0.675 au). Biver et al. averaged observations from five nights (January 13, 15, 16, 23, and 24) to arrive at their estimate for Q(HDO) (1.53 ± 0.21 × 10²⁶ s⁻¹).

IRAM, however, cannot detect H₂O. Biver et al. estimated water production indirectly from observations of the OH radical maser lines at 18 cm carried out with the Nançay radio telescope. During the January 12–17 period, they estimated the average water production rate to be 5.0 ± 0.2 × 10²⁹ s⁻¹ (Biver et al. 2015, 2016). However, the Nançay observations did not overlap completely with the IRAM HDO observations during later dates (January 23–24), so Q(H₂O) values for later days involved interpolation of the Nançay (January 12–17) and Odin observations of the H₂¹⁶O line taken at 556.9 GHz (January 30–February 3). This approach using asynchronous measurements provided the value for Q(H₂O) of 5–7 × 10²⁹ s⁻¹, reported in Tables 1 and 2 in Biver et al. (2016). If one would select measurements from January 13–16 (Table 3 in Biver et al. 2016), the resulting Q(HDO) is 1.8 ± 0.4 × 10²⁶ s⁻¹, so using the indirect estimate of H₂O production rate from Nançay measurements (i.e., 5.0 ± 0.2 × 10²⁹ s⁻¹), the resulting D/H is 1.8 ± 0.4 × 10⁻⁴, or 1.16 ± 0.26 VSMOW. While such a higher value agrees with our measurements (at 1σ), there are some concerns regarding the different IRAM and Nançay beam sizes of ∼10'' versus 35 × 19', respectively.

Considering the variation of rates observed with Odin (a decrease of Q(${{\rm{H}}}_{2}{}^{16}{\rm{O}}$) from 1.1 × 10³⁰ s⁻¹ to 5.5 × 10²⁹ s⁻¹, i.e., a maximum factor of 2 within four days), it is feasible that actual water rates could depart from the smaller variation shown in Table 1 of Biver et al. (2016), and from the value for Q(H₂O) used to estimate the D/H ratio in water (∼0.89 ± 0.25 VSMOW). Indeed, the variation in water production rates observed with Odin raises some concern on the use of observations from different days, or an average of them, to obtain D/H ratios. A similar situation is observed with the high values obtained in comet C/2012 F6 (Lemmon), whose D/H estimate resulted in the highest value to date: 4.17 ± 1.03 VSMOW (Biver et al. 2016). In this regard, performing simultaneous observations at IR wavelengths (in a single setting of the spectrometer) minimizes systematic uncertainties, a prime advantage of NIRSPEC observations.

How common are comets that have Earth-like water? Most measurements of HDO/H₂O in comets observed prior to 2011 suggested a "typical" value of ∼2 VSMOW (see Table 2 and Figure 3; but also Balsiger et al. 1995; Eberhardt et al. 1995 for Halley; Meier et al. 1998; Crovisier et al. 2004 for Hale–Bopp; Bockelée-Morvan et al. 1998 for Hyakutake; Hutsemékers et al. 2008 for T7; Villanueva et al. 2009 for 8P). The view of "2 VSMOW" water in comets might have changed significantly with recent improvements in the sample size and the recent improvement in instrumental sensitivities that now permit upper limits of 1 VSMOW or better in moderately bright comets. There are now two comets consistent with 1–1.3 VSMOW: C/2009 P1 (Garradd) (Bockelée-Morvan et al. 2012) and 103P/Hartley 2 (Hartogh et al. 2011)—both with the Herschel Space Observatory, and two comets with 3σ upper limits that preclude 2 VSMOW HDO: 153P/Ikeya–Zhang (Biver et al. 2006) and 45P/Honda–Mrkos–Pajdusakova (Lis et al. 2013, Herschel). More recently, Rosetta detected HDO in comet 67P at a heliocentric distance of 3.5 au (before water was fully activated) and found (D/H)_H2O of 3.4 ± 0.41 VSMOW (Altwegg et al. 2015, 2017 report consistent values before, through, and after perihelion), while Biver et al. (2016) reported ∼4.17 VSMOW in comet C/2012 F6 (Lemmon). These recent results suggest that the range of D/H in cometary water has not been fully sampled, and/or that temporal variability among isotopologues (HDO, H₂O) sampled separately is affecting the spread in D/H values significantly. Comparison of the high-accuracy (but widely disparate) results for 103P/Hartley 2 and 67P/C-G also suggest that the spread among JFCs is larger than observed to date (Hartogh et al. 2011; Altwegg et al. 2017).

Zoom In Zoom Out Reset image size

What were the possible chemical formation pathways that influenced the volatile fraction observed in comets? Some theoretical models of early solar system dynamics suggest that the outer planets migrated after dissipation of the gaseous disk, triggering disruption of a disk of icy planetesimals when Jupiter and Saturn entered a 3:2 dynamical resonance (a scenario known as the Nice model; see Gomes et al. 2005; Morbidelli et al. 2005; Tsiganis et al. 2005). While the details differ among models (Levison et al. 2011; Nesvorný & Morbidelli 2012; Brasser & Morbidelli 2013), there is general agreement that the event scattered comets into their current reservoirs. Close approaches to Saturn and Jupiter were more likely to eject comets from the solar system rather than emplace them in the Oort Cloud. Hence, both the OC and scattered disk comet populations are expected to be dominated by comet nuclei with initial perihelia beyond ∼15 au (Duncan et al. 1987). This is thought to explain why both long- and short-period comets have similar abundances of volatiles and similar scatter in volatile abundances among those sampled to date. Similar results are now being found for variations of HDO/H₂O.

The next question is whether cometary volatile abundances were inherited from the pre-solar cloud, were formed in the protoplanetary disk, or some combination of the two. Visser et al. (2011) suggested that water in the comet-forming region (∼5–33 au) underwent one desorption event from ices in the disk mid-plane before re-condensing, while water-ice material beyond ∼33 au never desorbed and hence should represent primordial composition. Furuya et al. (2016) find that the HDO/H₂O and D₂O/H₂O ratios in a protostellar disk where water ice has sublimated (NGC 1333-IRAS 2A) can be explained by both inheritance and disk chemistry. In their model, cometary HDO/H₂O variations can be reproduced by formation of water ice in the pre-stellar stage and selective loss of HDO enroute to the disk or, alternatively, by chemical processing and turbulent mixing in the class II disk stage. In this scenario, water ice from the protosolar nebula is raised to the disk surface where it is destroyed by UV and X-ray radiation, resetting the isotopic composition (see also Furuya et al. 2013; Albertsson et al. 2014). In this case, the HDO/H₂O ice ratio increases with heliocentric distance, reflecting the temperature-dependent fractionation reactions.

Furuya et al. (2016) suggest that a better diagnostic for the origin of cometary water is the ratio of D₂O/HDO relative to HDO/H₂O, which was measured to be more than a factor of 10 in the hot inner regions of one protostellar disk (NGC 1333-IRAS 2A). Recent measurements of this ratio in comet 67P obtained a value of about 17 (Altwegg et al. 2017)—greatly exceeding the statistical value of 0.25—and in agreement with the protostellar observations and model proposed by Furuya et al.

The implications for cometary contributions to Earth's water remain unclear. As discussed above, recent observations have increased the range of known HDO/H₂O ratios in cometary water and comets with Earth-like deuterium enrichments in water have been sampled, implying that a cometary source for terrestrial water cannot be disregarded. The expected comets with D/H < 1 VSMOW have yet to be found.

The combined results for C/2014 Q2 (Lovejoy) also raise questions about whether the observed deuterium enrichment in comets could be influenced by seasonal or evolutionary effects; however, in situ results from Rosetta do not reveal seasonal variations in HDO/H₂O or D₂O/HDO for comet 67P/Churyumov–Gerasimenko (Altwegg et al. 2017). To test this, both water and HDO must be observed multiple times with the same techniques for the same comet. Such studies may become more feasible with improved high-resolution infrared spectrometers (such as iSHELL at IRTF, the upgraded NIRSPEC at Keck II expected in 2018, and the upgraded CRIRES+ at VLT expected in 2018).

The authors would like to thank the anonymous referee and N. Biver and D. Bockelée-Morvan for interesting insights about this work. We also acknowledge support by the Keck PI Data Award (L.P.), administered by the NASA Exoplanet Science Institute. E.L.G. acknowledges support from NSF Planetary Astronomy Grants AST-1211362 and AST-1615441 and ISSI-Bern. Data were obtained at the W. M. Keck Observatory from telescope time allocated to the National Aeronautics and Space Administration through the agency's scientific partnership with the California Institute of Technology and the University of California.