© The Authors 2026

1 Introduction

Jupiter and the other giant planets are constantly subject to external modifications that enrich their atmosphere with new material (e.g. Lellouch et al. 1995; Hesman et al. 2007; Hartogh et al. 2011; Cavalié et al. 2010, 2014; Moses & Poppe 2017). A major example consisted of the 21 fragments of the comet Shoemaker-Levy 9 (SL9), which impacted the Jupiter atmosphere in July 1994 at 44°S during a week (Noll et al. 1996). Molecules that were previously undetected in the stratosphere of Jupiter (H₂O, CS, HCN, OCS, and many more) were observed minutes, months (see the review of Lellouch 1996), and, for the most long-lived, years after the impacts (Lellouch et al. 2002, 2006; Moreno et al. 2003; Iino et al. 2016; Cavalié et al. 2013, 2023; Benmahi et al. 2020; Rodríguez-Ovalle et al. 2025). Other molecules that were already present before SL9, such as CO (Bézard et al. 2002), saw their stratospheric abundance enhanced after the impacts (Lellouch et al. 1997). These molecules are the direct tracers of the dynamical and chemical processes triggered by the impacts (Zahnle 1996; Lellouch 1996). Therefore, their study during and after the impacts can make it possible to reconstruct their formation and/or deposition history and their further temporal and spatial evolution in the Jovian atmosphere.

The question of the origin of these molecules, i.e. from which reservoir of material (Jovian or cometary) and in which conditions (explosion, fireball, splashback, etc.) they were formed, is still open as it involves many parameters. Firstly, the elementary composition of the comet was not known before the impacts (Crovisier 1996) and some hints can only be found during or after the impacts in the new material formed/deposited in the Jovian atmosphere (Lellouch 1996). However, the cometary material completely mixed with the Jovian air in the impact fireball, making it difficult to separate the cometary from the Jovian contribution in the newly formed species (Lellouch et al. 1995; Crovisier 1996). In addition, the different phases of impact (fragment entry, final explosion, rising fireball, plumes fallback) led to different physical processes (Zahnle 1996) and thus altered the atmosphere differently depending on the altitude and horizontal extent at which they occurred. For example, CS, CO, and HCN were probably formed by shock-induced chemistry (Zahnle 1996), recombining cometary and Jovian material in the extreme conditions of the fireball and plumes splashback in which temperatures reached several thousands of kelvins (Chapman 1996; Bjoraker et al. 1996; Nicholson 1996). They were all deposited during the plumes’ re-entry into the stratosphere at pressures of 0.05–0.2 mbar (Lellouch et al. 1997). In this case, it becomes difficult to decipher the relative contributions of cometary and Jovian materials to the final products (Lellouch 1996). On the other hand, the detection of NH₃ and H₂S (normally only present in Jupiter’s troposphere) in the lower stratosphere suggested that convection driven by the heat released at a few bars by the fragment explosion overshoot to the lower stratosphere and transported here Jovian air and cloud materials (Yelle & McGrath 1996; Griffith et al. 1997). In this case, the lower stratospheric NH₃ and H₂S probably have a purely Jovian origin.

In addition, after the formation and deposition of the species at a given altitude, their initial distribution was modified as a function of time by dynamics and chemistry. The photolysis of the short-lived species (e.g. NH₃) led with time to the enrichment in other species (e.g. HCN; Moreno et al. 2003; Lellouch et al. 2006; Moses 1996), thus altering the primordial composition of the new species and making it even more difficult to assess the origin of their material. Also, at first confined to the impact sites, the species spread out with longitude within a few months after the impacts and with latitude to cover the entire disc years later (Moreno et al. 2003; Lellouch et al. 2006; Cavalié et al. 2023).

In this context, measuring the isotopic composition of the new stratospheric species can still provide insights into their formation processes and their dominant sources of elements (Jovian or cometary), but the isotopic ratios can also diagnose the temporal evolution of the exogenic molecules from interactions in the Jovian atmosphere. Isotopic ratios are often used to trace the origin of a material because they reflect the physical and chemical conditions in which their bearing molecule was formed (Nomura et al. 2023). They are usually compared to solar values (Marty et al. 2011; Lyons et al. 2018) or to the Earth references (IAEA 1995; Coplen et al. 1992) to reveal any fractionation processes as a function of the type of object, the distance to the Sun, or the molecule (Nomura et al. 2023). Only the carbon, nitrogen, and sulphur isotopic ratios were estimated four years after SL9 impacts in exogenic HCN and CS by Matthews et al. (2002). They derived an abnormal depletion in all three heavier isotopes (¹³C, ¹⁵N, ³⁴S), with ¹²C/¹³C, ¹⁴N/¹⁵N, and ³²S/³⁴S ratios nominally higher by factors of three, ten, and three, respectively, than the Earth references of 89, 272, and 23.

Isotopic ratios have been estimated in a wide range of Solar System objects and in several of their molecules over the past 50 years. For example, there is a general trend that the ¹²C/¹³C ratio is quasi-constant throughout the Solar System at a value close to solar and terrestrial, independently of the object (meteorites, comets, giant, and terrestrial planets) or the molecule (Nomura et al. 2023; Woods & Willacy 2009; Woods 2009). This absence of notable isotopic fractionation seems to indicate that the Solar System objects accreted carbon from a common reservoir, or that all carbon reservoirs present the same isotopic composition (Fletcher et al. 2009; Nomura et al. 2023). A departure from this common trend, as reported by Matthews et al. (2002) in the carbon present in the SL9-produced HCN, is thus very intriguing. On the other hand, the nitrogen isotopic composition presents a broader range of values in the Solar System and is highly dependent on the type of the object. In the giant planet atmospheres, the ¹⁴N/¹⁵N ratio is close to solar but is super-terrestrial by a factor of 1.6 (Owen et al. 2001; Fouchet et al. 2000; Fletcher et al. 2014). Comets are the most ¹⁵N-enriched objects of the Solar System. Their ¹⁴N/¹⁵N nominal values are at least a factor of two lower than the terrestrial reference and a factor of three lower than the solar and Jovian values (Hily-Blant et al. 2017). The dispersion of values in the nitrogen isotopic ratio among Solar System objects is thus helpful to decipher the origin of the objects, as opposed to the carbon ratio. However, as for carbon, the strong depletion in ¹⁵N in HCN after SL9 completely departs from any Solar System measurements, even from the Jovian bulk, which is the least enriched. Unfortunately, no complementary observations of the isotopic ratios in Jupiter’s exogenic molecules have been performed these past 30 years. The results of Matthews et al. (2002) thus remain puzzling and no definitive answer has been found regarding the origin of the elements composing HCN.

In this work, we used 2017 Atacama Large Millimeter/submillimeter Array (ALMA) observations of HCN and two of its isotopologues, H¹³CN and HC¹⁵N, to estimate the carbon and nitrogen isotopic ratios in HCN in the Jupiter stratosphere 23 years after SL9 impacts, and to try to understand the intriguing results of Matthews et al. (2002) in 1998. We detail the dataset in Section 2 and the radiative-transfer model used to simulate the observations in Section 3. We then present the results in Section 4 and discuss and compare them to other measurements in Section 5. We give our conclusions in Section 6.

2 Observations

We used the ALMA Cycle 4 dataset already presented in Cavalié et al. (2021, 2023) from which they derived, respectively, stratospheric wind measurements and vertical abundance profiles from the CO (J=3–2) line at 345.7960 GHz and the HCN (J=4–3) line at 354.5055 GHz. General information on the dataset is presented here, but more details on the observation setup and data reduction can be found in Cavalié et al. (2021, 2023).

The Jupiter observations were recorded on March 22, 2017 as part of the 2016.1.01235.S project. The ALMA configuration at the time of the observations consisted of 42 antennas with a diameter of 12 m and baselines extending from 15.1 to 160.7 m. The resulting beam size is 1.0–1.2″ (east-west and north-south) with a position angle of 87.3°. The complete dataset consists of four different spectral cubes of Jupiter, taken in four spectral bands, from which we extracted two smaller cubes at 345.3398 and 344.2003 GHz encompassing, respectively, the H¹³CN and HC¹⁵N (J=4–3) lines. These are the only isotopic lines present in the dataset. The continuum level was subtracted from these cubes. The spectral resolution is Δν = 0.488 MHz, resulting in a spectral resolving power of about 7 × 10⁵ at 344–345 GHz.

Jupiter’s angular diameter was 43.8″ at the equator, and its rotation axis was tilted by 23.7°¹. The sub-observer latitude was −3.5° and the sub-observer longitude was 65.4°W² at the beginning of the observation. During the 24 min on-source integration time, the planet rotated by 15°; we thus observed the planetary western limb in a longitude range spanning 155–170°W and 335–350°W at the eastern limb.

The small beam size compared to the planet’s disk resulted in a latitudinal resolution of 2–3° per beam at the low latitudes and up to 8° per beam at the higher latitudes. From the knowledge of Jupiter’s geometry and using the continuum map of Jupiter, we found a pointing offset of ΔRA=0.053″ and ΔDec=−0.039″ and re-centred the planetary disc accordingly. For both spectral cubes (H¹³CN and HC¹⁵N), we oversampled the beam by a factor of four and extracted 540 spectra at the planetary limb, which is defined by the projected planetary ellipse at the 1 bar level. We only extracted spectra at the limb where we have the highest line sensitivity.

The same procedure was performed by Cavalié et al. (2023) to extract HCN spectra at the limb. Because the spatial resolution is slightly better at the HCN (J=4–3) transition, they extracted a total of 557 limb spectra. The total 24 min observation time was sufficient to detect the HCN line in each of these individual limb spectra and reach signal-to-noise ratios (SNR) of the order of 25 per beam. An example of an HCN individual spectrum is presented in Fig. 1. However, because H¹³CN and HC¹⁵N are significantly less abundant than H¹²C¹⁴N, the resulting lines are too faint to be detected in an individual limb-extracted spectrum. Therefore, to increase the SNR, we needed to co-add spectra, and we adopted the following procedure: (1) for each spectrum and on both limbs, we subtracted the Doppler shift caused by the beam-convolved planet rotation to re-centre all the lines at the line rest frequency; then (2) we averaged the spectra on the selected latitudinal range that gives the best SNR; and (3) we subtracted any continuum baseline left with the averaging step. The best latitudinal range was found to be 50°S-50°N (i.e. 303 selected spectra) and corresponds to the highest and rather constant HCN column density with latitude (Cavalié et al. 2023). The final spectra obtained with this method are illustrated in Figs. 1 (right), 2, and 3 for HCN, H¹³CN, and HC¹⁵N, respectively. We detected the H¹³CN and HC¹⁵N lines with an SNR at a line peak of 24 and five, respectively, and with noise levels at 1σ of 2.1 and 2.6 mJy/beam.

Fig. 1 HCN limb spectra. (left) HCN limb spectrum extracted at a single position on the eastern equatorial limb at 3°N. (right) HCN spectrum averaged over the two limbs and on the latitudinal range 50°S-50°N. The observed spectra are presented in black and the modelled spectra in red. The differences between the observed and the modelled spectra are shown below with the residuals. The 0 Jy/beam level is depicted by the thin solid black line. The dashed-dotted lines represent the 1σ noise level in the single spectrum on the left panel, and the dashed lines the 1σ noise level from the right panel’s averaged spectrum. For the single pointing spectra, the continuum level was subtracted and the lines were centred at the rest frequency. The averaged spectra follow the same averaging method described in Section 2. Observations come from the dataset presented in Cavalié et al. (2023). Modelled spectra were computed with the procedure detailed in Section 3.

3 Modelling

We used the radiative-transfer model detailed in Cavalié et al. (2019) (and references therein) to simulate the HCN, H¹³CN, and HC¹⁵N lines in the same atmospheric, geometric, and instrumental conditions of the observations. The most important code inputs are summarised in Section 3.1. As the isotopologue lines are optically thin (the opacities at the line centre are 0.015 and 0.004, respectively), the line amplitude is sensitive to the atmospheric temperature and to the abundance of the species. We detail the 3D fields employed in the modelling in Sections 3.2 to 3.4.

3.1 Radiative-transfer model

The radiative transfer model from Cavalié et al. (2019) is a line-by-line model accounting for Jupiter’s 3D ellipsoidal geometry in the same conditions as the observations (see Section 2). The background composition of H₂, He, CH₄, NH₃, and PH₃ used for the continuum is the same as in Cavalié et al. (2023).

The pressure broadening coefficient, γ, and the exponent for the temperature dependency, denoted n, were not measured for the H¹³CN and HC¹⁵N (J=4–3) transitions. Rohart et al. (1987) only gave values for the J=1–0 transition of HC¹⁵N: γ = 0.1308 cm⁻¹.atm⁻¹ and n = 0.752. Here for consistency, in the nominal model, we adopted the pressure-broadening parameters of the H¹²C¹⁴N (J=4–3) transition from Rohart et al. (1987) (and used by Cavalié et al. 2023) for both heavier isotopologues, i.e. γ = 0.1447 cm⁻¹.atm⁻¹ and n = 0.756. In addition, we rescaled the HC¹⁵N (J=1–0) parameters using the H¹²C¹⁴N (J=1–0) and (J=4–3) values to estimate the parameters at the HC¹⁵N (J=4–3) transition: γ = 0.1305 cm⁻¹.atm⁻¹ and n = 0.770. The difference implied on the derived HC¹⁵N abundance by choosing either these rescaled values or the nominal ones is accounted for in the uncertainties. In the absence of information on the H¹³CN parameters, the same approach as for HC¹⁵N was performed for H¹³CN.

We computed spatially and spectrally convolved spectra of H¹³CN and HC¹⁵N at each of the 303 selected positions extracted from the limb, contained between 50°S and 50°N. The modelled continuum level was removed by subtracting a linear slope under the lines. Modelled spectra were then corrected for the beam-convolved planet rotation and co-added following the same method as for the observations, on the same latitudinal range.

Fig. 2 (Top) Observed and modelled spectra of the H13CN line at 345.3398 GHz. The observed spectrum is depicted by the black line and corresponds to a latitudinal average on the 50°S-50°N range from both limbs. The modelled spectrum that best fits the observations is represented by the solid red line. The pink area shows the ±2σ uncertainties on the modelled line. (Bottom) Residuals corresponding to the difference between the observed spectrum and the model’s best fit. The horizontal dotted black lines represent the ±1σ noise level in the averaged spectrum.

Fig. 3 Same as Fig. 2 but for the HC15N line at 344.2003 GHz.

3.2 Thermal field

The latitude-pressure thermal field used in this paper is presented in more detail in Cavalié et al. (2023). The data were retrieved from IRTF/TEXES observations (Cosentino et al. 2017; Sinclair et al. 2023), which were taken near-simultaneously to the ALMA dataset. The thermal vertical profiles inverted from these data span 1 μbar to 1 bar with a 2 K uncertainty (Cosentino et al. 2017). The final thermal field consists of two 2D latitude-pressure fields (one at each planetary limb) re-gridded on a 1° latitude grid (see Extended Data Fig. 1 in Cavalié et al. 2023).

Fig. 4 Selection of H12C14N vertical profiles used in our radiative-transfer modelling at different latitudes and all (arbitrarily) taken from the eastern limb. They illustrate the limited variety of profiles retrieved at all low-to-mid latitudes on both limbs, as shown in Cavalié et al. (2023). The filled areas show the 1σ envelope resulting from the 1σ uncertainties on the retrieved profile parameters. A typical temperature profile extracted at the equator is also displayed on the top x-axis as an example (dashed black line). The full 3D temperature field can be found in Cavalié et al. (2023).

3.3 H¹²C¹⁴N abundance field

We used the H¹²C¹⁴N latitude-pressure abundance field determined by Cavalié et al. (2023) from the same ALMA dataset of March 22, 2017, at both limbs. Their retrievals were performed at 557 positions, which were extracted at the planetary limb from a beam oversampling of a factor of four. The HCN observations benefit from a high spectral resolving power of 3 × 10⁶ that enables us to resolve the true line shape, which in turn allows us to retrieve the vertical distribution of HCN. Each of their 557 HCN spectra were fitted with an inversion algorithm coupled to the radiative-transfer model presented above, in which the HCN vertical profile was parametrised with three parameters, a₁, a₂, and a₃, and the following relationship: $$y_{\mathrm{H}\mathrm{C}\mathrm{N}}(z)=a_1\times (1+\tanh \left(\frac{z-a_2}{2a_3}\right)),$$

where y_HCN is the HCN abundance or volume mixing ratio (VMR), z is the altitude in km above the 1-bar level, a₁ is half the asymptotical high-altitude HCN VMR, a₂ is a cut-off level between the upper and lower stratosphere, and a₃ indicates the steepness of the slope at the cut-off level, with a stronger slope for larger a₃ values. Examples of these retrieved HCN vertical profiles are presented in Fig. 4. The uncertainties on the retrieved vertical profiles come from the combination of several sources: (i) flux-calibration uncertainties, which are reported at a level of 5%; (ii) retrieval uncertainties on the a₁, a₂, and a₃ parameters resulting from the spectral noise and which translate to an approximately 5% uncertainty on the retrieved abundances. It should be noted that, even though the HCN line has an opacity of 1.6 at the line centre in the low-to-mid latitudes, any line saturation is accounted for when retrieving the uncertainties on the vertical profile parameters; (iii) uncertainties resulting from the 2 K uncertainty on the temperature field, which translate into a 1% uncertainty on the abundances; (iv) the uncertainties resulting from the continuum subtraction in the uv-plane during the data analysis. This was reported by Cavalié et al. (2023) to be the largest source of uncertainties on the retrieved abundances, causing a 30% uncertainty on the abundances. Considering all four contributions, there is an overall 40% uncertainty on the HCN vertical profiles.

With the individual HCN vertical profiles of Cavalié et al. (2023), the final HCN abundance field used in our radiative-transfer calculations consists on two 2D latitude-pressure fields (one at each limb), re-gridded with a 1° latitude step, as for the thermal field. This was done by co-adding vertical profiles located within each 1° latitude step. Because the latitudinal resolution is 2–3° in one beam in these observations, and because the beam was oversampled by a factor of four in the HCN profile retrieval process, it means that each 1° latitude step of our grid averages two retrieved profiles on average.

To ensure that the re-gridded HCN field presented above was valid, we ran radiative-transfer calculations for the HCN line at each of the 557 positions. As an example, Fig. 1 shows a model-observation comparison for the east-equatorial spectrum. After applying the spectrum-averaging method described in Section 2, we obtained the result shown in Fig. 1 right. The co-addition of individual spectra results in a significant decrease of the noise level, from 0.043 mJy/beam on a single spectrum, to 0.004 mJy/beam on the averaged one. As a consequence, the mismatches that lied within the noise on individual spectra result in differences after the averaging stage that then become more significant (see the residual on Fig. 1, right). However, those differences remain very limited. This demonstrates that the HCN abundance field can be used as a reference to derive the isotopologue abundances and consequently compute the ¹²C/¹³C and ¹⁴N/¹⁵N isotopic ratios.

3.4 H¹³CN and HC¹⁵N abundance field

We assume H¹³CN and HC¹⁵N have the same vertical and horizontal distribution as the 3D H¹²C¹⁴N abundance field detailed above, but rescaled by a constant factor representing the isotopic ratio. Fractionation from condensation with altitude was excluded for two reasons: (1) the HCN vapour pressure curve (see Jancso & Van Hook 1974; Fray & Schmitt 2009; Hudson & Gerakines 2023) indicates that the molecule is far from condensing under Jovian stratospheric temperature and pressure conditions (about 170 K at 0.1 mbar), with mole fractions of the order of 10⁻¹¹ − 10⁻⁹; and (2) vapour pressure curves between isotopologues are typically very similar. Saturation curves from Jancso & Van Hook (1974) (and references therein) indeed indicate a difference of the order of 10% between the HCN and HC¹⁵N curves, which is still very far from the Jovian atmospheric conditions. The same is expected for H¹³CN, of which the vapour pressure is unknown. Other isotopic fractionation processes (photolysis and neutral or ionic chemistry) can change the isotopic distribution as a function of altitude. It is the case for the N-bearing molecules in the N₂ rich atmosphere of Titan, but the fractionation occurs at very high altitudes, where isotopic ratios are sensitive to magnetospheric electrons (Nosowitz et al. 2025; Dobrijevic & Loison 2018). In the stratosphere of giant planets, these effects in the C- and N-bearing molecules, such as HCN, are poorly known and more unlikely in the 50°S-50°N latitude range adopted here, compared to auroral regions. Here, in the absence of observational or theoretical evidence, and to limit the number of parameters to adjust, we did not consider any isotopic fractionation effects as a function of altitude.

Therefore, we derived the H¹²C¹⁴N/H¹³C¹⁴N (resp. H¹²C¹⁴N/H¹²C¹⁵N) ratios by rescaling the H¹²C¹⁴N field until the H¹³C¹⁴N (resp. H¹²C¹⁵N) line was fitted. The isotopic ratios of carbon and nitrogen in HCN are then simply the following: $$\frac{y_{{\mathrm{H}}^{12}{\mathrm{C}}^{14}\mathrm{N}}}{y_{{\mathrm{H}}^{13}{\mathrm{C}}^{14}\mathrm{N}}}=\frac{{}^{12}\mathrm{C}}{{}^{13}\mathrm{C}}\text{\hspace{0.17em}and\hspace{0.17em}}\frac{y_{{\mathrm{H}}^{12}{\mathrm{C}}^{14}\mathrm{N}}}{y_{{\mathrm{H}}^{12}{\mathrm{C}}^{15}\mathrm{N}}}=\frac{{}^{14}\mathrm{N}}{{}^{15}\mathrm{N}}.$$

4 Results

We ran radiative-transfer calculations to find the isotopic ratios that best fit the observations. The results for ¹²C/¹³C and ¹⁴N/¹⁵N are the following: $$\begin{array}{rl}& \begin{array}{rl}\frac{{}^{12}\mathrm{C}}{{}^{13}\mathrm{C}}& =73\pm 5\\ & =(0.76-0.87)\times 89.0\text{\hspace{0.17em}(Earth ref.)\hspace{0.17em}}\\ & =(0.72-0.83)\times 93.5\text{\hspace{0.17em}(Sun)\hspace{0.17em}}\\ & =(0.69-0.87)\times 92.6\text{\hspace{0.17em}(Jupiter)},\end{array}\\ \\ & \begin{array}{rl}\frac{{}^{14}\mathrm{N}}{{}^{15}\mathrm{N}}& =245\pm 29\\ & =(0.80-1.00)\times 272.0\text{\hspace{0.17em}(Earth ref.)\hspace{0.17em}}\\ & =(0.48-0.63)\times 440.9\text{\hspace{0.17em}(Sun)\hspace{0.17em}}\\ & =(0.42-0.70)\times 434.8\text{\hspace{0.17em}(Jupiter)}.\end{array}\end{array}$$

The present isotopic ratios are compared to the reference terrestrial values of 89 for carbon and 272 for nitrogen (IAEA 1995; Coplen et al. 1992), to the solar values of 93.5 ± 0.7 and 440.9 ± 5.4 (Lyons et al. 2018; Marty et al. 2011), and to the Jovian bulk values from the Galileo entry probe of 92.6 ± 4.3 and 434.8 ± 56.7 (Niemann et al. 1998; Owen et al. 2001). In Jupiter, those ratios were measured in CH₄ and NH₃, respectively, as the major C- and N-bearing species in Jupiter’s troposphere. Additional references are found in Tables A.1 and A.3 for the Jovian bulk. The comparison factors encompass the uncertainties of our derived values in the Jovian exogenic HCN and the uncertainties on the object of comparison (the Earth, Sun, or Jupiter).

We took into account three types of uncertainties that we quadratically added: (1) the natural fitting uncertainty arising from the noise level in the spectrum (at ±1σ); (2) the error arising from the uncertainties on the individual HCN vertical profiles derived by Cavalié et al. (2023) (see Section 3.3); and (3) the uncertainty that arises from the choice of pressure broadening parameters γ and n (see Section 3.1). In order to avoid overestimating the second source of uncertainty when performing the average of the spectra on the 50°S-50°N latitudinal range, we divided by the square-root of the number of independent pointings in this range (=303/4).

For HC¹⁵N, the major source of uncertainty resides in the noise level (about 70% of the total error), followed by the effect of the broadening parameters and the uncertainty on the HCN profile (about 15%). In the case of H¹³CN, the uncertainty arising from the noise level accounts for ~10%, the rest being due to the broadening parameters and the uncertainty on the HCN profile (about 40% each). Averaging the spectra from 50°S to 50°N considerably reduces the impact of the individual HCN vertical profile uncertainties. The uncertainties on the thermal field are already accounted for in the HCN vertical profile uncertainties (Cavalié et al. 2023). The best fits to the data are shown in Figs. 2 and 3. Computations of χ²/N values around the line centre give 1.0 and 0.7, respectively, for the H¹³CN and HC¹⁵N lines.

The above results are representative of the ¹²C/¹³C and ¹⁴N/¹⁵N values in the carbon and nitrogen present in the HCN molecule in Jupiter at the observation date. Computations of the contribution functions, averaged on 50°S-50°N, indicate that for both isotopologues 60% of the line emission originates from the 0.03–1.80 mbar range, with a peak at 0.25 mbar. Implications of these results are presented in the next section.

5 Discussion

Stratospheric HCN is an exogenic species that was formed in extreme conditions from a Jovian air–cometary mixture (Gautier et al. 1995; Zahnle 1996) and has continuously evolved in the Jovian stratosphere ever since (e.g. Moreno et al. 2003; Lellouch et al. 2006; Cavalié et al. 2023). Therefore, the interpretation of the isotopic ratios in HCN is particular here; their values years after SL9 can give information on the Jovian–comet mixture at the origin of HCN or on the high pressure or temperature chemistry of the shock, but in principle they can also reflect the history of HCN and its isotopologues in the Jovian stratosphere. These two interpretations, i.e. origin and evolution, are discussed in this section.

In this context, we now have two data points for the ¹²C/¹³C and ¹⁴N/¹⁵N ratios in HCN, one from 1998 observations (Matthews et al. 2002), and one from 2017 data. No observations of the heavier isotopologues of HCN (or of any other exogenic species containing carbon or nitrogen) were made right after the collisions; hence, we do not have any direct constraints on the isotopic composition at that time and we can only rely on later measurements. A comparison of the 1998 and 2017 results shows a contradictory behaviour; i.e. while the Matthews et al. measurements suggest an atypically strong depletion in both ¹³C and ¹⁵N relative to the Jovian bulk, our new estimates instead indicate an enrichment compared to Jovian values. In what follows, we first discuss the validity of the 1998 results in Section 5.1. We explain why we eventually chose to discard them. We then give our interpretation of the 2017 results alone in Section 5.2.

5.1 Implications of the 1998 measurements

The first observations of H¹³CN and HC¹⁵N were taken in September 1998, four years after the SL9 impacts, by Matthews et al. (2002). They found the following ranges of values: $$\begin{array}{rl}& \begin{array}{rl}\frac{{}^{12}\mathrm{C}}{{}^{13}\mathrm{C}}& =176-409\\ & =(2.0-4.6)\times 89(\text{Earth ref.})\\ & =(1.9-4.4)\times 93.5(\text{Sun})\\ & =(1.8-4.6)\times 92.6(\text{Jupiter}),\end{array}\\ \\ & \begin{array}{rl}\frac{14\mathrm{N}}{15\mathrm{N}}& =1150-4500\\ & =(4.3-16.7)\times 272(\text{Earth ref.})\\ & =(2.5-10.3)\times 440.9(\text{Sun})\\ & =(1.8-11.2)\times 434.8(\text{Jupiter}).\end{array}\end{array}$$

Their results are compared to the Earth reference and to the solar and Jovian bulks, in the same way as in Section 4. Figs. 5 and 6 present determinations of the carbon and nitrogen ratios in various Solar System objects (more details are found in Tables A.1–A.3 and references therein). The Jovian atmosphere is well known to present solar values both for carbon and nitrogen (see the most accurate results from the in situ measurements of the Galileo entry probe (Niemann et al. 1998; Owen et al. 2001); a Jovian source was consequently ruled out. Matthews et al. (2002) interpreted the ratios to originate from an unusual cometary composition or an unknown isotopic fractionation process during the different phases of the impacts. We discuss the two hypotheses below and also give other interpretations.

Fig. 5 Review of 12C/13C ratio in the various Solar System objects. Detailed information on the instrument, the observation date, and the references are reported in Tables A.1 (giant planets only) and A.2 (the other Solar System determinations) and references therein.

Fig. 6 Review of 14N/15N ratio in the various Solar System objects. Detailed information on the instrument, the observation date, and the references are reported in Table A.3 and references therein.

5.1.1 An unknown isotopic fractionation during the impacts

Firstly, one possibility is that a fractionation mechanism in the shocks could have preferentially enriched HCN in the lighter isotopes (¹²C and ¹⁴N), probably through interactions with the dust, as suggested by Moreno et al. (2003). However, this hypothesis would need to be confirmed by laboratory experiments and/or plume modelling. Such work should explain why C, N, and S seem to be affected by this depletion process similarly, even if their isotopic ratios are usually very different. Until then, this possibility is still open.

5.1.2 An unusual comet composition

More than 20 years since Matthews et al. (2002), we now have access to tens of derivations in different comets and in several molecules. Among these, one can note the in situ data of the comet 67P/Churyumov–Gerasimenko in various molecules. All the measurements show Earth-like values for carbon, and the nitrogen ratio is systematically enriched in ¹⁵N (as measured in several N-bearing compounds) with respect to the terrestrial value with factors up to five, and this is, on average, with a factor of two (see Figs. 5 and 6). For the ¹²C/¹³C ratio, only Biver et al. (2016) nominally found a depletion in the comet C/2012 F6 (Lemmon) with respect to the terrestrial value, but their error bars are very large (124±64) compared to many other estimates, and they still encompass the terrestrial value. For the ¹⁴N/¹⁵N ratio, even the highest cometary values (e.g. 205 ± 70 in HCN in comet C/1995 01 (Hale–Bopp; Hily-Blant et al. 2017) are at least more enriched in ¹⁵N by a factor of five compared to those of Matthews et al. (2002). An unusual cometary source for the carbon and nitrogen content in HCN, as proposed by Matthews et al. (2002), is thus highly improbable with regard to all these comet measurements.

5.1.3 A temporal evolution of HCN between 1994 and 1998

Another potential explanation may derive from a temporal evolution of the total mass of HCN after the comet impacts of 1994, altering the isotopic composition of HCN. Continuous observations after the collisions indicated that the total mass of HCN increased by a factor up to six (Moreno et al. 2003) ten months after SL9, compared to the initial HCN mass derived by Bézard et al. (1997). The same amount was still observed in September 1998 by Moreno et al. (2003) and was consistent with the Cassini/CIRS observations at the end of 2000 (Lellouch et al. 2006 indicated an increase by a factor of ${5.5}_{-2.5}^{+5.0}$). As discussed extensively in Lellouch et al. (2006), the main explanation of the HCN enhancement is the conversion of Jovian NH₃ (Moses 1996) which was observed in the upper stratosphere after SL9 (Griffith et al. 1997).

This additional production of HCN from NH₃ should ultimately have modified the isotopic composition of the initial SL9-produced HCN. With very simple considerations, we can give an estimation of the isotopic ratios of the initial SL9-produced HCN, noted IR_i. The values of Matthews et al. (2002), denoted IR_f, are taken as the isotopic ratios of the total amount of HCN in 1998 (after the NH₃ conversion). The additional NH₃-produced HCN should present the typical ratios of the Jovian NH₃ and CH₄ denoted IR_a, taken here as the values measured by the Galileo entry probe (see Tables A.1 and A.3). The isotopic ratios of the initial SL9-produced HCN are written as follows (for carbon or nitrogen): $${\mathrm{I}\mathrm{R}}_i={(\frac{n}{{\mathrm{I}\mathrm{R}}_f}+\frac{1-n}{{\mathrm{I}\mathrm{R}}_a})}^{-1},$$

where n is the enrichment factor of HCN between the initial SL9-produced amount and the total amount of HCN several months after the impacts. The above formula is only valid for n ≥ 1, because no loss factor of HCN is considered. Values of n lie between 0.6 and six following the results of Moreno et al. (2003) and between three and 10.5 according to Lellouch et al. (2006). As a consequence, we considered n ranges between three and six.

Considering the uncertainties on IR_f and IR_a, IR_i presents only negative (i.e. non-physical) values for n > 1.9 for carbon and for n > 1.4 for nitrogen. Positive values can exist only if n is close to one; in this case, the initial ratios become even higher than those of Matthews et al. (2002), rendering them even more incompatible with values measured in comets. In addition, such values for n are incompatible with the range resulting from Moreno et al. (2003) and Lellouch et al. (2006). In conclusion, the changes in the HCN abundance in the years after the impacts cannot explain the abnormal isotopic ratios reported by Matthews et al. (2002).

5.1.4 A possible modelling issue

Matthews et al. (2002) analysed HCN (J=4–3) and CS (J=7–6) spectra obtained at the equatorial limb and at the planet’s disc centre with the James Clerk Maxwell Telescope (JCMT). The half power beam width (HPBW) was 13″.5, and the angular diameter of Jupiter was 50″, resulting in a covered latitudinal range of 10°S-10°N. At the time of their observations, HCN had already spread out latitudinally and was present up to northern mid-latitudes with a mixing ratio of about 10⁻⁸ (Moreno et al. 2003).

Independent temperature information was unavailable for the modelling of their observations. As a consequence, Matthews et al. (2002) simultaneously derived the temperature vertical profile and the HCN and CS abundance vertical profiles directly from the HCN and CS lines using the different observation geometries (limb vs. nadir). They used simple constant vertical profiles for P₀<0.2 mbar both for the temperature and the HCN abundance. Because the H¹²C¹⁴N line was optically thick (τ ~ 10 at the line centre), the H¹²C¹⁴N abundance that is derived is strongly tied to the retrieved temperature. In addition, the P₀ cut-off pressure derived from the line width also shows some dependency on the temperature. Thus, several solutions can be found simultaneously for the temperature, the HCN abundance, and P₀. To illustrate this point, Matthews et al. (2002) found an HCN abundance ranging from (3.0–4.5)×10⁻⁸ for a temperature of 164 ± 2 K at P₀<0.2 mbar, while a reanalysis in Moreno et al. (2003) indicated (0.8–2.5)×10⁻⁸ for a temperature of 167 ± 1 K at P₀<0.3 mbar for the same dataset. Our conclusion is that the abundance of H¹²C¹⁴N used to compute the isotopic ratios might not be well constrained, and this immediately impacts the isotopic ratios values. As a comparison, the CS line was less optically thick, and the subsequent sulphur ratio was closer to the terrestrial value, but still depleted in ³⁴S.

Contrary to H¹²C¹⁴N, the H¹³CN and HC¹⁵N lines were optically thin in 1998, and the derived abundances are more reliable. Therefore, a better estimator of the isotopic composition in HCN can reside in the H¹³CN/HC¹⁵N ratio. As temperature should affect the determination of the H¹³CN and HC¹⁵N abundances in a similar way, the H¹³CN/HC¹⁵N ratio should not be too temperature dependent. Matthews et al. (2002) found volume-mixing ratios of (1.4 ± 0.3) × 10⁻¹⁰ and (1.8 ± 0.8) × 10⁻¹¹, respectively, for H¹³CN and HC¹⁵N. From these values and their uncertainties, acceptable ratios are in the H¹³CN/HC¹⁵N=(2.7–12.9) range. As a comparison, the terrestrial and Jovian ratios give, respectively, 3.1 using the Earth references and (3.9–5.5) using the values of the Galileo entry probe. Our results lead to H¹³CN/HC¹⁵N=(2.8–4.0). Thus, when not considering the H¹²C¹⁴N abundance, they are now compatible with those of Matthews et al. (2002), within their large error bars.

For all the above reasons, we deem the isotopic ratios of Matthews et al. (2002) not to be sufficiently reliable, and useful information can only be found in their ¹³C/¹⁵N ratio. In the next section, we disregard them and only consider our new estimates to propose new scenarios on the origin and evolution of HCN.

5.2 Implications of the 2017 results

We derived new carbon and nitrogen isotopic ratios in HCN 23 years after the SL9 impacts. We find that HCN in 2017 is enriched in ¹³C and ¹⁵N compared to the solar–Jovian bulk values. The ranges of values defined by the uncertainties at 1σ are incompatible with the ratios in Jupiter for both ratios. For nitrogen, the ¹⁴N/¹⁵N ratio still encompasses the more ¹⁵N-enriched terrestrial value, but on the edge of the error bar. The nitrogen ratio appears to be at an intermediate value between the Jovian bulk and typical cometary values that are ¹⁵N-rich (see Fig. 6). For carbon, the comparison is less diagnostic as all the Solar System objects exhibit very similar values (see Fig. 5), but a value of 73 ± 5 is still unusual in the Solar System. Enrichments in ¹³C with respect to the terrestrial reference are observed in some comets within the error bars. Therefore, the enrichments seen here could constitute the direct signature of the comet.

However, during the 23 years separating our measurements and the impacts of SL9, we recall that HCN evolved in the Jovian atmosphere by spreading out horizontally and vertically (Moreno et al. 2003; Lellouch et al. 2006; Cavalié et al. 2023), but also chemically. Several chemical processes may have taken place or may still be taking place. (1) In the months following SL9, additional HCN was produced from stratospheric NH₃ conversion (Moreno et al. 2003; Lellouch et al. 2006); (2) monitoring years later in 2017 by Cavalié et al. (2023) suggested that the total quantity of HCN decreased by a factor of 5.0 ± 3.0 compared to the 1995–1998 data of Moreno et al. (2003) or 12.0 ± 3.5 compared to the 2000 values of Lellouch et al. (2006). This decrease of the HCN mass over time is mainly caused by auroral losses through adsorption on aerosols; (3) in addition, several isotopic fractionation processes could have occurred between 1994 and 2017 and could have altered the isotopic composition (neutral or ionic chemistry in the gas phase, chemistry on aerosols, photolysis, etc.). In the N₂-rich atmosphere of Titan, isotopic fractionation in N-bearing molecules exists and occurs as a function of altitude and depending on the molecule (Nosowitz et al. 2025; Dobrijevic & Loison 2018). However, little is known about the fractionation rates and the kinetics of these potential reactions with HCN at the ~0.1 mbar level in the Jovian atmosphere.

In what follows, for the sake of simplicity, we give an estimation of the origin of the elements in HCN, without any isotopic fractionation mechanism over time. Contrary to carbon, the nitrogen isotopic ratio shows very distinct values depending on the objects, and it is therefore a more reliable tracer of the origin. We can thus estimate the nitrogen contribution from the comet to HCN.

Following the method used in Matthews et al. (2002), the total number of an atom, X (¹⁴N or ¹⁵N), is expressed as ${\mathrm{N}}_{\text{tot\hspace{0.17em}}}^{\mathrm{X}}={\mathrm{N}}_{\mathrm{S}\mathrm{L}9}^{\mathrm{X}}+{\mathrm{N}}_{\mathrm{J}}^{\mathrm{X}}$, where ${\mathrm{N}}_{\mathrm{S}\mathrm{L}9}^{\mathrm{X}}$ is the number of the X atom coming from the comet, and similarly ${\mathrm{N}}_{\mathrm{J}}^{\mathrm{X}}$ from Jupiter. The nitrogen isotopic ratios in the mixture, in SL9, and in Jupiter are written as ${\mathrm{I}\mathrm{R}}_{\text{tot,SL9,J\hspace{0.17em}}}={\mathrm{N}}_{\text{tot,SL9,J\hspace{0.17em}}}^{{14}_{\mathrm{N}}}/{\mathrm{N}}_{\text{tot,SL9,J\hspace{0.17em}}}^{{15}_{\mathrm{N}}}$. The fraction of Jovian nitrogen in the mixture, $f=\frac{{\mathrm{N}}_{\mathrm{J}}^{{14}_{\mathrm{N}}}}{{\mathrm{N}}_{\text{tot\hspace{0.17em}}}^{{14}_{\mathrm{N}}}}$, is given by $$f=\frac{{\mathrm{I}\mathrm{R}}_{\mathrm{J}}\times ({\mathrm{I}\mathrm{R}}_{\mathrm{S}\mathrm{L}9}-{\mathrm{I}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}})}{{\mathrm{I}\mathrm{R}}_{\mathrm{t}\mathrm{o}\mathrm{t}}\times ({\mathrm{I}\mathrm{R}}_{\mathrm{S}\mathrm{L}9}-{\mathrm{I}\mathrm{R}}_{\mathrm{J}})},$$

where IR_tot is the value of the mixture that we derived here, IR_J is well known from Galileo entry probe measurements (Owen et al. 2001), and IR_SL9 is taken as the weighted average of the ~50 measurements presented in Fig. 6 and referenced in Table A.3, leading to (140 ± 4).

We find the fraction of cometary nitrogen to be ~(40±20)% of the total amount of nitrogen in HCN, taking into account the uncertainties on each ratio. The fraction of Jovian nitrogen is thus (60±20)%. The latter contains all the Jovian contributions, including the one resulting from the conversion of stratospheric NH₃ in the months that followed the impacts. Removing the latter from the total Jovian contribution depends on the HCN increase factor, n, which is not tightly constrained (see Moreno et al. 2003; Lellouch et al. 2006, and Section 5.1). For example, with n = 3 or n = 5, about (50–80)% or (60–90)% (respectively) of the nitrogen in the initially SL9-produced HCN would be cometary. These results then suggest at least an equal cometary and Jovian origin for nitrogen in the SL9-produced HCN. As discussed in Lellouch (1996), a predominant cometary origin for nitrogen is possible in the exogenic N-bearing molecules, which is consistent with our estimations.

New HCN, H¹³CN, and HC¹⁵N data are needed to monitor the temporal evolution of the isotopic ratios in HCN. A variation of the ratios after our 2017 measurements would definitively point towards a fractionation mechanism as a function of time, induced by various chemical reactions. If such a variation is seen, better constraints will be set on the poorly known chemistry between HCN, its isotopologues, and the rest of the Jovian atmosphere. If, within the error bars, no temporal evolution of the ratios is observed, it would either suggest that a chemical equilibrium has already been reached or that isotopic fractionation is negligible on a timescale of a decade. In the last case, the isotopic ratios could then directly represent the initial mixture at the origin of the exogenic molecules formed during the SL9 impacts. In the meantime, photochemical models are needed here to confront the enrichments we derived to potential fractionation processes and to understand how these processes alter the isotopic composition over time.

However, as time is passing by, the abundance of the SL9-produced HCN is expected to continue its decrease at mid-latitudes from photolysis, vertical transport, and horizontal transport/destruction towards the poles, making it more difficult to obtain good SNR measurements. The JUpiter ICy moons Explorer (Juice) mission will arrive at Jupiter in 2031, and monitoring of the isotopic ratios in HCN will be of great importance to understand the isotopologues’ evolution and chemistry (Fletcher et al. 2023). It will be especially interesting to look at the isotopologues of other well-known molecules to try to complete the picture of fractionation processes as a function of the species.

6 Conclusion

We present ALMA observations of Jupiter of H¹³CN and HC¹⁵N lines at 345.3398 and 344.2003 GHz, respectively, taken in 2017. To increase the SNR, we co-added limb spectra on the latitudinal range 50°S-50°N after correcting for the planet-induced Doppler shift of the lines observed at different latitudes on the planet limb. With a radiative-transfer model, we rescaled the HCN vertical profiles retrieved from the same ALMA dataset (Cavalié et al. 2023) to fit the isotopologue lines. The ¹²C/¹³C and ¹⁴N/¹⁵N isotopic ratios deduced from the fitting are 73 ± 5 and 245 ± 29. The values indicate that HCN is enriched in the heavier isotopes compared to the Jovian bulk values measured in CH₄ and NH₃. This behaviour is at odds with the 1998 measurements of Matthews et al. (2002) that suggested a very unusual depletion. Because of limitations in the retrieval of the H¹²C¹⁴N abundance, we chose to disregard their results when discussing our new measurements.

Using only our new derivations, we conclude that the enrichments we see here in both ¹³C and ¹⁵N can result from two, not mutually exclusive, scenarios: (1) the signature of the comet contribution to the formation of the Jovian stratospheric HCN and/or (2) the chemical evolution of HCN in the Jovian atmosphere. In the absence of models taking into account the different fractionation mechanisms for HCN, we can only give an estimation of the origin of the material in HCN, without considering fractionation over time. Since the carbon isotopic ratio does not seem to present any fractionation in Solar System objects compared to the terrestrial/solar values, it is more difficult to conclude on the cometary contribution to carbon, especially since our value is inconsistent with the standard terrestrial and solar values. On the contrary, comets are easily traceable by their highly ¹⁵N-enriched ratios compared to the Jovian value close to solar. With the knowledge of the bulk isotopic composition of Jupiter, the dozens of derivations in comets, our new measurement, and by taking into account the additional production of HCN from NH₃ in the months after the SL9 impacts, we estimate the cometary contribution to the nitrogen composing the initial SL9-produced HCN to be at least 50%.

Monitoring the isotopic ratios in HCN over time is essential to constrain the chemical evolution of the minor isotopologues and infer the possible fractionation processes. Future observations with the Juice spacecraft will be especially needed in that respect. Observations should be confronted with photochemical models that incorporate the isotopologue chemistry to give an idea of the strength and the timescale of the fractionation processes. However, one should keep in mind that the HCN abundance at mid-latitudes decreases over time because of transport and photolysis. Thus, achieving high SNRs of the minor isotopologue lines will become increasingly challenging, limiting our ability to determine the isotopic composition of HCN.

Acknowledgements

C.L. and T.C. acknowledge funding from the Centre National d’Études Spatiales (CNES) and the Programme National de Planétologie (PNP) of CNRS/INSU. This Paper makes use of the following ALMA data: ADS/JAO.ALMA#2016.1.01235.S. ALMA is a partnership of ESO (representing its member states), NSF (USA) and NINS (Japan), together with NRC (Canada), MOST and ASIAA (Taiwan), and KASI (Republic of Korea), in cooperation with the Republic of Chile. The Joint ALMA Observatory is operated by ESO, AUI/NRAO and NAOJ.

References

Appendix A Carbon and nitrogen isotopic ratios in the Solar System

Tables A.1 and A.2 summarise ¹²C/¹³C determinations in various Solar System objects (giant planets in Table A.1 and other objects in Table A.2). Table A.3 presents similar data but for the ¹⁴N/¹⁵N ratio. Details on the object, molecule, instrument, observation date, sensitive region, and the references from which we extracted the data are added. Figs. 5 and 6 display these values. In the case of comets and meteorites, as many measurements were performed, we do not detail them in the tables below, but values can be found in the references and are also illustrated in Figs. 5 and 6.

All Tables

All Figures

Fig. 1 HCN limb spectra. (left) HCN limb spectrum extracted at a single position on the eastern equatorial limb at 3°N. (right) HCN spectrum averaged over the two limbs and on the latitudinal range 50°S-50°N. The observed spectra are presented in black and the modelled spectra in red. The differences between the observed and the modelled spectra are shown below with the residuals. The 0 Jy/beam level is depicted by the thin solid black line. The dashed-dotted lines represent the 1σ noise level in the single spectrum on the left panel, and the dashed lines the 1σ noise level from the right panel’s averaged spectrum. For the single pointing spectra, the continuum level was subtracted and the lines were centred at the rest frequency. The averaged spectra follow the same averaging method described in Section 2. Observations come from the dataset presented in Cavalié et al. (2023). Modelled spectra were computed with the procedure detailed in Section 3.

In the text

Fig. 2 (Top) Observed and modelled spectra of the H13CN line at 345.3398 GHz. The observed spectrum is depicted by the black line and corresponds to a latitudinal average on the 50°S-50°N range from both limbs. The modelled spectrum that best fits the observations is represented by the solid red line. The pink area shows the ±2σ uncertainties on the modelled line. (Bottom) Residuals corresponding to the difference between the observed spectrum and the model’s best fit. The horizontal dotted black lines represent the ±1σ noise level in the averaged spectrum.

In the text

	Fig. 3 Same as Fig. 2 but for the HC15N line at 344.2003 GHz.
In the text

Fig. 4 Selection of H12C14N vertical profiles used in our radiative-transfer modelling at different latitudes and all (arbitrarily) taken from the eastern limb. They illustrate the limited variety of profiles retrieved at all low-to-mid latitudes on both limbs, as shown in Cavalié et al. (2023). The filled areas show the 1σ envelope resulting from the 1σ uncertainties on the retrieved profile parameters. A typical temperature profile extracted at the equator is also displayed on the top x-axis as an example (dashed black line). The full 3D temperature field can be found in Cavalié et al. (2023).

In the text

	Fig. 5 Review of 12C/13C ratio in the various Solar System objects. Detailed information on the instrument, the observation date, and the references are reported in Tables A.1 (giant planets only) and A.2 (the other Solar System determinations) and references therein.
In the text

	Fig. 6 Review of 14N/15N ratio in the various Solar System objects. Detailed information on the instrument, the observation date, and the references are reported in Table A.3 and references therein.
In the text