Molecular composition of short-period comets from millimetre-wave spectroscopy: 21P/Giacobini-Zinner, and 64P/Swift-Gehrels

We present the results of millimetre-wave spectroscopic observations and spectral surveys of the following short-period comets: 21P/Giacobini-Zinner in September 2018, 41P/Tuttle-Giacobini-Kresák in April 2017, and 64P/Swift-Gehrels and 38P/Stephan-Oterma in December 2018, carried out with the Institut de RadioAstronomie Millimétrique (IRAM) 30-m radio telescope at wavelengths between 1 and 3 mm. Comet 21P was also observed in November 1998 with the IRAM 30-m, James Clerk Maxwell Telescope, and the Caltech submillimeter Observatory radio telescopes at wavelengths from 0.8 to 3 mm. The abundances of the following molecules have been determined in those comets: HCN, CH 3 OH, CS, H 2 CO, CH 3 CN, and H 2 S in comet 21P; HCN and CH 3 OH in 41P; HCN, CH 3 OH, and CS in 64P; and CH 3 OH in 38P. The last three comets, classiﬁed as carbon-chain typical from visible spectro-photometry, are relatively rich in methanol (3.5–5% relative to water). On the other hand, comet 21P, classiﬁed as carbon-chain depleted, shows abundances relative to water which are low for methanol (1.7%), very low for H 2 S (0.1%), and also relatively low for H 2 CO (0.16%) and CO (<2.5%). Observations of comet 21P do not show any change in activity and composition between the 1998 and 2018 per-ihelions. Sensitive upper limits on the abundances of other molecules such as CO, HNCO, HNC, or SO are also reported for these comets.


Introduction
Comets are the most pristine remnants of the formation of the Solar System 4.6 billion years ago. They sample some of the oldest and most primitive material in the Solar System, including ices, and they are thus our best window into the volatile composition of the proto-solar disk. Comets may have also played a role in the delivery of water and organic material to the early Earth (see Hartogh et al. 2011, and references therein). The latest The radio spectra are only available at the CDS via anonymous ftp to cdsarc.u-strasbg.fr (130.79.128.5) or via http://cdsarc. u-strasbg.fr/viz-bin/cat/J/A+A/651/A25 Based on observations carried out with the IRAM 30-m telescope. IRAM is supported by INSU/CNRS (France), MPG (Germany) and IGN (Spain). simulations of early Solar System evolution (Brasser & Morbidelli 2013;O'Brien et al. 2014) suggest a more complex scenario. On the one hand, ice-rich bodies that formed beyond Jupiter may have been implanted in the outer asteroid belt and participated in the supply of water to Earth or, on the other hand, current comets coming from either the Oort Cloud or the scattered disk of the Kuiper belt may have formed in the same trans-Neptunian region, sampling the same diversity of formation conditions. Understanding the diversity in the chemical and isotopic composition of cometary material is thus essential in order to assess such scenarios (Altwegg & Bockelée-Morvan 2003;Bockelée-Morvan et al. 2015).
Comet 21P/Giacobini-Zinner is a Jupiter-family comet (JFC) orbiting the Sun every 6.5 yr on a moderately low-inclination (32.0 • ) orbit. It was first discovered by M. Giacobini in 1900 and A25, page 1 of 15 30-m after perihelion (observations before perihelion suffered from bad weather). We observed comets 64P and 38P with the IRAM 30-m telescope at a time when the main target (comet 46P/Wirtanen) was unavailable on 12, 13, and 18 December 2018 UT. We searched for several molecules in those comets. We report the detections of HCN, CH 3 OH, CS, CH 3 CN, and H 2 CO and tentatively H 2 S in 21P; HCN and CH 3 OH in 41P; HCN, CH 3 OH, and CS in 64P; and a tentative detection of CH 3 OH in 38P. Sections 2 and 3 present the observations and frequency coverage. Section 4 shows the spectra of the detected molecules. Section 5 presents the information extracted from the observations for analysis and so as to compute production rates. Section 6 provides the retrieved production rates and abundances or upper limits, which are discussed and compared to other comets in Sect. 7.

Observations with the IRAM 30-m telescope
Comet 21P was observed under good to fair weather conditions (2 to 5 mm of precipitable water vapour -pwv) daily during the 11-15 November 1998 period. In 2018, during the 14-18 September allocated period, the weather was very poor, with high opacity (9-40 mm of pwv, most of the time between 10 and 16 mm). The observations were thus more concentrated on the low frequency bands of the EMIR receiver (2 and 3 mm bands, Carter et al. 2012). The comet was tracked using orbital elements JPL#K182/7 1 in 2018, where ∼4 ephemeris error was estimated afterwards.
Comet 41P observations were originally planned during two sessions, mid-March and at the end of April 2017, to observe its evolution from before to after perihelion. However, adverse weather conditions limited observations to only two nights (out of the eight scheduled) post perihelion on 26.0 April and 1.0 May 2017 UT, with average weather (3-6 mm pwv). 41P was tracked using the JPL#171/7 orbit solution. The difference in ephemeris computed with the JPL#171/18 solution of 2018 and those used by the IRAM new control system (NCS software) was negligible (<0.2 ). A map of HCN(3-2) taken on 26 April yielded a final pointing error of less than 2 .
Comets 64P and 38P were observed under good weather conditions (∼2 mm of precipitable water vapour) on 11.8 and 17.9 December 2018, to fair weather (pwv ∼5 mm) for 38P on 12.1 and 13.1 December 2018 UT. 64P was tracked using the JPL#183/4 orbit solution (JPL#35 for 38P) to compute the position in real time. Pointing was regularly checked on bright sources (rms <1 ) but ephemeris errors around 2 both in RA and Dec were found afterwards. The two comets were observed for a short time on three different days with different frequency coverage (Table 1).
In 1998 we used 3 mm, 2 mm, and 1 mm 'B100', 'C150', and 'B230' newly installed single-side band (SSB) receivers in parallel, each with a maximum 512 MHz bandwidth, connected to filter banks and correlators. In 2017-2018, we used the EMIR (Carter et al. 2012) 3 mm, 2 mm, and 1 mm band receivers in 2SB mode which offer two instantaneous bandwidths of 8 GHz per polarisation. They were connected to the Fourier Transform Spectrometer (FTS) which covers the full bandwidths with 0.195 MHz sampling (corresponding to 0.22-0.24 km s −1 at 1 mm wavelength). The central part of the band was also connected to the VESPA autocorrelator. Some lines (e.g. CH 3 OH(5 2 − 4 1 )E at 266 838.123 MHz, HCN(1-0) at 88 631.848 MHz) were also observed with VESPA using its higher spectral resolution (e.g. 20-40 kHz sampling) in a narrow bandwidth (20-80 MHz). In order to cancel the sky emission, a wobbler switching mode at a frequency of 0.5 Hz was used for most observations, with reference sky positions at ±180 in azimuth. In 1998 some observations were obtained in frequency switching mode with frequency throws of 7.2, 7.8, or 15.85 MHz.

Observations with JCMT and CSO
Comet 21P was also observed from Maunakea between 30 October and 14 November 1998. The weather was good on 30, 31 October, 1, 2, 3, and 14 November 1998 with less than 1 mm pwv, average on 4 and 15 November (2 mm pwv) and poor for Maunakea standards on 10 and 13 November (about 5 mm pwv). With the exception of some observations of the HCN(3-2) line at 265.9 GHz, most of these observations used the 0.8 mm band receivers (304-363 GHz). Some of the JCMT observations were obtained in the slow frequency switching mode, switching by ±8.1 MHz every two seconds. Other observations were obtained by beam switching with a reference at 2 in azimuth. All CSO observations and some of the JCMT observations were obtained in double-side band (DSB) mode with the two side bands separated by 3 GHz and overlapping in the resulting spectrum. Additional information is provided in Table 1. The tracking of the comet was done by interpolating ephemeris generated with the JPL-DE406#27 orbit solution.

Spectra and line intensities
Some of the spectra of 21P obtained in the frequency switching mode at JCMT or IRAM in 1998 are contaminated by stratospheric HCN, H 2 CO, or mesospheric CO. The telluric line however falls at an absolute frequency different from the comet line thanks to its Doppler shift. However the wings of the broader stratospheric lines overlap with the comet spectra and we had to remove a fit to the atmospheric line to obtain a flatter baseline in the figures presented here and to derive accurate noise and line intensities. Three lines of HCN, seven lines of CH 3 OH, and one each of CS and H 2 CO were detected with the JCMT, CSO, and IRAM telescopes in 1998. Spectra (average of several days of observations in some cases) are shown in Figs. 1, 2, 3, and 4. The daily intensities of the lines observed with JCMT and CSO are provided in Tables A.1 and A.2, respectively. The intensities of the lines observed with the IRAM 30-m telescope in 1998 and 2018 are provided in Table A.3. The mean Doppler shift of each line is computed as the first moment integral over the line width ( T b.v.dv/ T b.dv), excluding the weaker (<4%) hyperfine components for HCN, and taking the average on the two strongest hyperfine components for HCN(1-0). In 2018, lines were stronger because the comet was closer to Earth. The EMIR receivers at the IRAM 30-m telescope are also more sensitive than the receivers used in 1998, and offer a much broader bandwidth (Table 1). Spectra are shown in Figs. 5, 6, and 7.
Spectra of HCN and CH 3 OH detected in comet 41P are shown in Figs. 8 and 9 and integrated intensities are provided in Table A.4. For comet 64P, the HCN(3-2) was clearly detected on 11.8 December (Fig. 10) together with a marginal detection of methanol lines at 266.838 GHz and around 252 GHz. CH 3 OH was best detected via its 242 GHz lines (Fig. 11) on 17.9 December UT. CS(5-4) was marginally seen in this same observation. Integrated intensities or 3 − σ upper limits are provided in

Data analysis
In order to derive production rates, we assume steady-state outgassing at a constant velocity and use the Haser density profile. Excitation of the molecular rotational lines depends on radiative and collisional processes. Collisions with neutrals and electrons are taken into account but depend on the gas temperature and water production rate. Model parameters are given in Table 3.

Expansion velocity and outgassing pattern
The expansion velocity is derived from the lines observed with the best signal-to-noise ratio (S/N). Generally the Doppler velocity at half maximum intensity (VHM) in the comet frame provides a good estimate of the expansion velocity: on the observer side from negative velocities and on the opposite side (gas moving away seen on the redshifted side of the line). When possible for a double-peak line we fitted two Gaussians from which we measure the two V HMs. We took the average value of the expansion velocity deduced from the blueshifted and redshifted sides of the lines. For 21P average values are given in Table 3 for the various periods. The lines do not show a strong asymmetry (Figs. 1-7, especially looking at the very symmetric HCN(1-0) line observed with high frequency resolution in 2018 in Fig. 7). The Doppler shift is small or not significant (Table A.1-A.3, Col. 7). Therefore we assumed symmetric outgassing at a mean velocity of 0.7 to 0.73 km s −1 for 21P.
The expansion velocity we assumed for comet 41P was derived from the profile of HCN(3-2) (Fig. 8) and the CH 3 OH (2 0 − 1 −1 )E and (5 2 − 4 −1 )E lines, which were observed with the high-resolution backend VESPA. The hyperfine structure of the HCN(3-2) line has a minor effect on the broadening of the main component and was taken into account to derive the expansion velocity. From their line shape, we obtained the average values VHM = −0.8 and +0.6 km s −1 , and consequently we assumed a mean expansion velocity of 0.7 km s −1 .
For comet 64P, from the profile of the spectra of HCN(3-2) and CH 3 OH(5 0 − 4 0 )A at 241 791.367 MHz which have the highest S /N (10 and 6, respectively) we determined a gas velocity of 0.80 ± 0.05 km s −1 on the Earth-facing side, and 0.5 ± 0.1 km s −1 on the other side. Since the phase angle (Table 1) was small (less than 28 • ), in contrary to the two previous comets, the Earthfacing side was close to the day side and line asymmetry can be associated with day-to-night asymmetry in activity. We adopted the mean value v exp = 0.65 km s −1 . The mean Doppler shift of the lines (Table A.5, Col. 7) was also significantly negative, confirming the preferential outgassing on the day side at a higher velocity.
For comet 38P, which was observed at 1.6 au from the Sun, we assumed v exp = 0.6 km s −1 , using the general ∼0.8 × r −0.5 hdependence of the expansion velocity observed in comets (Biver et al. 2000(Biver et al. , 2006.

Gas temperature
Rotational temperatures were measured for methanol and CH 3 CN. In some cases a meaningful rotational temperature of HCN could also be deduced from contemporaneous observation of HCN(1-0) with IRAM 30-m and HCN(3-2) with CSO 10.4m, which probed the same region of the coma (same beam size of ∼26 ). Nevertheless the rotational temperature is not necessarily a measurement of the gas temperature T gas of the inner coma. As radio observations often sample a region larger than the collision-dominated region of the coma, the population of the rotational levels of the considered molecule evolves radiatively throughout the beam owing to spontaneous decay and infrared pumping via vibrational bands. Some series of methanol lines around 165, 157, 252, or 305 GHz sample rotational levels from the same K-ladder that evolve in a similar manner and keep memory of the thermal equilibrium in the collision-dominated region (Bockelée-Morvan et al. 1994). On the other hand, the rotational levels sampled by methanol lines around 145, 242, or 338 GHz, as well as HCN, are more sensitive to the collision rate and evolve rapidly towards a much colder fluorescence equilibrium. Consequently, their average rotational temperature in the beam is much lower than T gas . Measured values of T rot and constraints derived on the gas temperature T gas are provided in Table 2.
Several methanol lines were detected in 1998 and in 2018 in comet 21P. In 1998, however, the constraints remain loose on the gas temperature owing to low S/N (Table 2). We adopted a value of T gas = 40 K. In 2018 the rotational temperature of the methanol lines at 145 and 165 GHz (Fig. 12) could be more precisely measured and we adopted T gas = 42 K.
For comet 41P, the individual methanol lines were often only marginally detected (Table A.4), so the uncertainty in the rotational temperature is relatively large: T rot = 33 + inf −19 K, T rot = 28 ± 7 K, and T rot = 51 +53 −17 K for the 165, 252, and 242 GHz lines respectively. The rotational temperature of the 165 and 252 GHz lines are expected to provide a good estimate of T gas ,  whereas we expect a lower value for the 242 GHz lines. Hence, we used T gas = 45 K, but also tested a value of 35 K, which yields similar production rates (3-8% lower for both HCN and CH 3 OH).
We had only weak constraints on the gas temperature from the observations of methanol lines in 64P. The rotational temperature of the lines around 252 GHz is T rot = 47 +51 −19 K, which suggests a gas temperature T gas in the 33-100 K range. The rotational temperature of the 242 GHz lines is better constrained ( Fig. 13) with T rot = 18 ± 5 K. Unfortunately it does not bring a precise constraint on the gas temperature: for the assumed collision rate, this rotational temperature is compatible with any value of T gas in the 10-100 K range. We adopted T = 40 K, consistent with the values measured in comets of similar activity and heliocentric distance. In any case, a change of ±10 K does not result in a change of production rate larger than the uncertainties owing to the limited signal-to-noise of the dataset, and much larger changes in temperature have never been seen in comets of similar activity. We assumed a lower temperature T gas = 30 K for comet 38P, typical for comets of moderate activity observed at 1.6 au from the Sun (e.g. Biver et al. 2012).

Reference water production rate
OH was observed in comet 21P with the Nançay radio telescope in 1985(Gerard et al. 1988, 1998 and 2018. The 1998 data provided OH production rates in the range 3-4 × 10 28 molec. s −1 for the period of the millimetre observations (Crovisier et al. 2002), whereas Combi et al. (2011) provided slightly higher production rates, with a value around 5 × 10 28 molec. s −1 at perihelion (21 November 1998). In 2018, no reliable OH production rate could be derived from Nançay data during the IRAM observations because of the poor OH maser inversion at that time. Combi et al. (2011) also showed a trend observed at each perihelion passage of comet 21P: the activity peaks about 25 days prior to perihelion, with a steep decrease afterwards. This was also observed for the visual heliocentric magnitude, which peaked 1-2 weeks prior to perihelion in 1998 and 2018. In 2018, Roth et al. (2020) observed OH and H 2 O in the infrared yielding a water production rate in the range 3.2-3.0 × 10 28 molec. s −1 for the 7-11 September period, just before our IRAM observations. Faggi et al. (2019), also from infrared observations of H 2 O, found water production rate in the range 2.6-2.8 × 10 28 molec. s −1 for the 8 September-7 October 2018 period. We thus assumed Q H 2 O = 3.0 × 10 28 molec. s −1 at the time of IRAM observations (Table 3).
Comet 41P outgassing was monitored by several observatories during its 2017 approach, including the Solar Wind ANistropies (SWAN) instrument on the SOHO spacecraft (31 March to 20 April, Combi et al. 2019), TRAPPIST (11 March to 24 May for OH, Moulane et al. 2018) and with the Nançay radio telescope (4 March-5 June). The comet was not observed at Nançay at the time of the IRAM observations because of poor OH inversion. OH production rates of ≈1.1 × 10 28 and 0.6 × 10 28 molec. s −1 were obtained for the 28 March-3 April and 18-30 May, respectively. Most of the observations showed a plateau of activity ±20 days around perihelion with Q H 2 O ≈ 4, 3 and 8 × 10 27 molec. s −1 respectively. We adopted a median value of Q H 2 O = 4 × 10 27 molec. s −1 for the time of our observations. For comets 64P and 38P, around the time of our observations, water production rates were estimated from ground-based spectro-photometric observations of OH. D.G. Schleicher (priv. comm.) provided Q H 2 O = 7.6 × 10 27 molec. s −1 for 64P on 16 December 2018. For comet 38P his values were Q H 2 O = 5 − 4 × 10 27 molec. s −1 on 3 and 16 December 2018. We adopted water production rates of Q H 2 O = 0.8 × 10 28 molec. s −1 for 64P and Q H 2 O = 0.5 × 10 28 molec. s −1 for 38P, both to compute collision rates and determine molecular abundances. 64P and 38P were only observed in September and October at Nançay.

Production rates and abundances
Line intensities have been converted to production rates using our time-dependent non-LTE code taking into account collisions with neutrals and electrons as well as solar radiative pumping via vibrational bands for HCN, CH 3 OH, CS, H 2 CO, CO, and CH 3 CN, as used in previous papers (Biver et al. 2000(Biver et al. , 2006(Biver et al. , 2012(Biver et al. , 2014(Biver et al. , 2015(Biver et al. , 2016. Photolysis constants are also from previous papers and detailed in Biver et al. (2021). Production rates or upper limits are provided in Table 4. For species with multiple lines observed at the same time, the weighted average is provided.

Relative abundances
Abundances relative to water were computed using the average water production rates provided in Table 3. For 41P we assumed a lower water production rate (0.3 × 10 28 molec. s −1 ) for the second day (1.0 May 2017) because HCN and CH 3 OH showed a significant decrease in production rate and Moulane et al. (2018) also found a decrease in the water production rate between 26.1 April and the 28 April-3 May period. For 38P, abundances relative to water assume Q H 2 O = 0.5 × 10 28 molec. s −1 . Although no line is detected with a signal-to-noise larger than 2, the combined signal of the 13 strongest methanol lines at 165-170 GHz and 242 GHz yields a formal signal-to-noise of 3.5 and a realistic methanol abundance of 5.0 ± 1.4%. For comet 21P we took the average of 1998 and 2018 measurements as they do not show any difference.
Abundances and a comparison to values measured in other comets are provided in Table 5. Range of abundances measured in the full sample of comets and in other JFCs in the radio is also provided in the last two columns (Biver & Bockelée-Morvan 2019; Biver et al. 2021).

Evolution of production rates of comet 21P
Comet 21P was observed during a two-week period in 1998 and again in 2018 after three full orbits around the Sun. Figure 14 shows both series of production rates measurements relative to the time of perihelion. The 1998 data show a decreasing trend of the production rates of methanol and HCN over the two weeks. The 4-day span of the 2018 observation is too short to draw definitive conclusions, but shows production rates very similar to the 1998 values. The heliocentric distance in 2018 was on average 4% lower than in 1998. If the peak outgassing follows r −2 h , it should have been 8% higher in 2018. Taking this change into account, one can infer that the comet activity follows the same trend as in 1998 of continuous decrease of active surface around A25, page 7 of 15 A&A 651, A25 (2021)  perihelion. This was also observed by Combi et al. (2011) and Roth et al. (2020) for the 1998, 2005, and 2018 perihelions. Besides this small trend, production rates and relative abundances were very similar in 1998 and 2018, suggesting no significant secular trend both in activity and composition.

Discussion and conclusion
Comet 64P/Swift-Gehrels had its most favourable apparition in 2018 and was poorly investigated at previous perihelion     Notes. (a) Number of lines used for the determination of T rot . (b) For an electron density factor x ne = 0.2 (Zakharov et al. 2007). have a higher methanol-to-water abundance ratio (Table 5), which appears to be often correlated with a high C 2 /OH (and C 2 /CN for comets studied here). Comet 21P is severely depleted in H 2 S; it has the lowest H 2 S abundance relative to water observed so far, and the same is true for CS. Calmonte et al. (2016) found that the sulphur budget in comet 67P/Churyumov-Gerasimenko led to S/O ∼ 1.5% in the volatile phase, with H 2 S as a major contributor. If the ratio is similar in comet 21P, it implies that sulphur species could represent more than 1% relative to water (assuming CO 2 /H 2 O ≤25%, based on e.g. Ootsubo et al. 2012, for the O budget). In most other comets investigated at radio wavelengths (Biver & Bockelée-Morvan 2019) the total of sulphur species amounts to 1-2% relative to water (in number). Unless comet 21P is depleted in sulphur in its volatile phase, one may wonder if OCS, SO(SO 2 ), H 2 CS, CH 3 SH, or some other molecule, could be major sulphur species in its coma. This cannot be completely excluded by the upper limits we obtained (see Tables 4 and 5 for the former species and we don't have stringent upper limits for the others). However Saki et al. (2020)     end observed in comets (Table 5; Bockelée-Morvan & Biver 2017). The relatively low abundance of methanol seen in the radio around perihelion was also derived in the infrared in 1998 (Weaver et al. 1999), 2005(DiSanti et al. 2013) and 2018 (Faggi et al. 2019), with abundances around 1.3% relative to water. Comet 21P is also depleted in NH (A'Hearn et al. 1995) and NH 2 (Fink 2009) but showed unidentified emission near 8, 9 and 11 µm, some of which could be attributed to complex organics  Ootsubo et al. (2020). Unfortunately the sensitivity of the 2018 observations was limited and our upper limits on the abundance of complex organics such as NH 2 CHO or CH 3 CHO are higher than values measured in comets (Biver et al. 2021), but nevertheless show that comet 21P is not strongly enriched (by a factor >5) in these molecules. The depletion of sulphur species, carbon-chain organics and NH 2 might be present in the volatile phase of comet 21P, but compensated by an enrichment in heavier refractory species as suggested by Ootsubo et al. (2020). Its observations at two different perihelion passages, separated by 20 yr, do not show any significant change.