Main

The discovery of Sedna1 has initiated an ongoing debate about the formation and evolution of the distant regions of our Solar System. Typical trans-Neptunian objects (TNOs) have perihelia q < 40 au, and their orbits are strongly influenced by Neptune’s gravitational perturbations. On the other hand, Sedna-like objects with large semi-major axes (a > 200 au)2 and large perihelia (q > 60 au)3 appear to evolve in stable orbits that have remained largely unchanged and not altered by the gravity of Neptune since the formation of the Solar System4. No viable transfer mechanisms to raise their perihelia exist with the current configuration of planets. Their stability suggests that an external gravitational influence beyond those of the currently known Solar System planets is required to form their orbits.

Several scenarios have been proposed to explain the orbits of Sedna-like objects: (1) interactions with a rogue planet-sized body or solar-mass star5,6,7,8, (2) interactions with a hypothetical distant planet9,10,11,12,13,14, (3) solar migration within the Milky Way15, (4) stellar encounters that took place while the Sun was still a member of its natal star cluster16,17,18,19 and (5) the capture of interstellar objects from low-mass stars during the early evolution of the Solar System5,20. So far, only three Sedna-like objects are known, so this population remains poorly understood. However, the discovery of additional objects is particularly challenging owing to their great distance from the Sun, resulting in very faint apparent magnitudes. Increasing the sample of Sedna-like objects is of great interest to better understand the history of the Solar System and place stronger constraints on the aforementioned scenarios.

Phase II of the ‘Formation of the Outer Solar System: an Icy Legacy’ project (FOSSIL II) is an extension of the original FOSSIL I project, using the 8.2 m Subaru Telescope on Maunakea (https://www.fossil-survey.org/). FOSSIL I utilized a long-term observational cadence repeating on two to five pointings to obtain light curves of Solar System objects21,22,23. FOSSIL II surveys ~25 deg2 to a limiting magnitude of mr ≃ 25.2, focusing on (1) the detection of high-perihelion TNOs, and (2) the dynamical classification and size distribution of resonant TNOs. Building on the foundations of shallower surveys such as the Canada–France Ecliptic Plane Survey (Petit et al.24, mg of 23.5–24.4), the Outer Solar System Origins Survey (Bannister et al.25, mr of 24.1–25.2) and the Dark Energy Survey (Bernardinelli et al.26, mr of 23.8), FOSSIL II aims to increase the inventory of small and distant outer Solar System objects, enabling better population modelling and exploration of their properties.

Within the first year of FOSSIL II observations, we detected an object 2023 KQ14 (nicknamed ‘Ammonite’) with a remarkably high perihelion. The barycentric orbit fit, based on a 10.16 year arc (from our observations and archival data; Methods) in the J2000.0 reference frame is a = 251.9 ± 0.3 au, e = 0.7383 ± 0.0003, i = 10.98° (error <0.01°), Ω = 72.104 ± 0.001°, ω = 198.71 ± 0.03° and q = 65.9 ± 0.2 au, with a heliocentric distance of approximately 71.0 au at the time of discovery (2023 May 16.5 Universal Time). At q = 66 au, it has the third largest perihelion among International Astronomical Union Minor Planet Center (MPC)-listed objects with a semi-major axis larger than 200 au and multi-opposition observations, following 2012 VP113 (q = 80.6 au)13 and (90377) Sedna (q = 76.3 au)1, and preceding (541132) Leleākūhonua (q = 65.0 au)27, 2021 RR205 (q = 55.6 au) and 2013 SY99 (q = 50.0 au)28. The median magnitude of mr = 24.6 corresponds to a diameter of 220–380 km for albedos of p = 0.15–0.05. This object fills the ‘perihelion gap’ of TNO discoveries with 50 au ≲ q ≲ 75 au (see the discussion in ref. 29), signalling the importance of distant TNO discoveries to map out the structure of the distant Solar System (Fig. 1). Meanwhile, Ammonite’s longitude of perihelion is in the opposite direction of the other Sedna-like objects (Fig. 2). Its high perihelion suggests the potential for long-term orbital stability, making it valuable for testing the recent hypotheses of primordial clustering of Sedna-like objects30 and the existence of a distant massive planet31,32,33 by analysing its orbital elements and overall dynamical behaviour.

Fig. 1: Orbital distribution of known distant TNOs.
figure 1

Objects with observed arcs ≥1 opposition (opp), a > 150 au and q > 30 au based on the MPC database as of February 2025 are plotted. This is an updated version of fig. 3.1 from ref. 29. The large perihelia objects mentioned in this study are indicated with their names. The location of Ammonite in this plot is marked with a star in both panels. Left, the semi-major axis versus perihelion distribution, where the dashed vertical red line marks the approximate region where galactic tides and passing stars become substantial perturbations on the TNO orbits, while the curved dot-dashed and dotted lines illustrate the upper boundary of chaotic diffusion and gravitational scattering by Neptune, respectively28,61. The hatched box indicates a region currently lacking any detections, as defined in ref. 29. Right, the distribution of Δϖ = ϖ − 254°, defined as the difference between the perihelion longitude ϖ (ϖ = ω + Ω) of each TNO and that of the hypothetical planet proposed previously32; Ammonite falls outside the proposed ϖ clustering of large-q objects.

Full size image
Fig. 2: Orbits of the four Sedna-like TNOs projected onto the J2000 ecliptic plane.
figure 2

Orbits of 2012 VP113, Sedna, Leleākūhonua and Ammonite, with Neptune’s orbit around the Sun shown for comparison. The orbital elements used are from the same MPC dataset as those in Fig. 1.

Full size image

Results

The a and q boundaries that define whether an object exhibits Sedna-like characteristics vary, particularly regarding long-term orbital stability over 4.5 Gyr. Dynamical studies suggest that some objects with similar large q, such as 2013 SY99 and 2021 RR205, may experience gradual orbital migration or diffusion owing to minor perturbations from Neptune and the influence of galactic tides/passing stars28,30. Although objects with large semi-major axis and q > 45 au tend to remain detached from the influence of the giant planets, their long-term stability needs to be verified through numerical integration34,35,36. A possible way to define the current diffusion boundary analytically is by applying the resonance overlap criterion, as discussed in refs. 37,38, which distinguish between chaotic and non-chaotic regions in a–q space.

Ammonite’s semi-major axis is close to that of the hypothetical planet proposed by Batygin and Brown14 and Lykawka and Ito31, as well as to the boundary of the Neptunian mean motion resonances in which an object may experience perihelion-raising through von Zeipel–Lidov–Kozai dynamics or mean motion resonances2,39,40,41,42. In addition, Ammonite provides a good test case for the primordial clustering of Sedna-like objects generated by a transient planetary body. Therefore, we performed the following numerical simulations to verify the orbital evolution of Ammonite.

Long-term stability

The results of both forwards and backwards simulations using two independent codes indicate similar stability (Methods), with mean variations of the semi-major axis and eccentricity remaining under 1% for the best fit a, possible highest a and possible lowest a orbits, as shown in Supplementary Fig. 1. The inclination of the clones of Ammonite oscillates between 8° and 11° throughout the simulations. None of the 1,000 clones show evidence of resonant behaviour in their orbital history; the closest major resonance (the 24:1 at a = 250.1 au) is about five sigma away from the best fit. This orbital evolution is consistent with studies of this region of orbital parameter space, which indicate that objects with a > 200 au and q > 60 au experience minimal orbital evolution in the timescale of 109 years (refs. 40,42). The orbital evolutions clearly demonstrate this similarity of the object to other Sedna-like objects. On the basis of this, we conclude that it can be identified as the fourth Sedna-like object discovered so far, with the third largest perihelion.

TNO discoveries so far suggest the presence of a ‘perihelion gap’: an apparent underpopulated region among the orbital parameters of TNOs with 150 < a ≲ 600 au and perihelia between roughly 50 and 75 au (refs. 27,29,43). Crucially, the gap is not generated by the limitation of sensitivities of the surveys, as more distant discoveries do occur (Fig. 1). If this gap does exist, it could be considered as a structural feature of the population, with implications for distinguishing between orbital formation and evolutionary mechanisms, such as the semi-major axis diffusion seen in larger a orbits. Ammonite is the first TNO with 150 < a < 600 au to have a perihelion in this gap. As demonstrated in Supplementary Fig. 1, Ammonite’s a is sufficiently small that it remains stable so that diffusion does not explain its orbit28,29. Therefore, a formation mechanism is still undecided but is required to populate orbits throughout the perihelia range. Future surveys with more detections are necessary to determine whether there is any distribution gap associated with the population.

Exploring a possible primordial orbital alignment

Recently, Huang and Gladman30 examined the orbital histories of the three previously known Sedna-like objects: Sedna, 2012 VP113 and Leleākūhonua. That study revealed an intriguing result where their longitudes of perihelion (ϖ) converged to a narrow cluster around 200° 4.5 Ga. Although this clustering hints at a primordial event that elevated their perihelia, additional discoveries and analyses are needed to solidify this picture. Here, we extend the analysis by including Ammonite and performing a similar backwards integration using the same parameters as in ref. 30.

Our results indicate a comparable clustering event around 4.2 Ga, roughly 300 Myr after the formation of the Solar System, with a confidence level (measured by the Rayleigh test of uniformity) exceeding 97% (Fig. 3). In addition, we performed 10,000 Monte Carlo simulations to test the robustness of this early clustering. Our results reveal that fewer than 7.7% of randomly shifted orbital histories exhibit a stronger ϖ clustering (between 4.55 and 4.16 Ga, as detailed in the Methods) than what we observe among the four Sedna-like objects.

Fig. 3: Time evolution and statistical analysis of the perihelion longitudes of four Sedna-like objects.
figure 3

Top, past evolutions of longitudes of perihelion (ϖ) for Ammonite (red) and the other three Sedna-like objects (black). Bottom, the circular s.d. of the four angles (black) and the statistical confidence (P value, shaded red) that they are generated from a uniform distribution. The addition of Ammonite suggests a late primordial clustering around 4.2 Ga compared with ref. 30, approximately 300 Myr after the formation of the Solar System, with over 97% confidence.

Source data.

Full size image

However, we note that this level of significance corresponds to slightly less than 2σ, and the inclusion of Ammonite results in a somewhat looser and delayed clustering relative to the findings of Huang and Gladman30. If future observations confirm a more pronounced and statistically robust clustering, it could indicate that a transient planetary perturber (for example, refs. 6,44) played a role early in the history of the Solar System. Following such an event, the clustering might have gradually dispersed due to the differential precession of the apsidal lines driven by the four giant planets. Stellar flyby models (for example, refs. 5,17,19), on the other hand, do not produce a clustered ϖ30. Further observations, particularly a more precise refinement of Leleākūhonua’s orbit and the discovery of new Sedna-like TNOs, will be essential to confirm or challenge this tentative primordial alignment and to better constrain the formation history of the early Solar System.

Interaction with a hypothetical planet

It is important to note that the ϖ and Ω of Ammonite do not align or cluster with those of Sedna, 2012 VP113 and Leleākūhonua (Figs. 1, right, and 2). A present-day planet has been proposed as a mechanism for gravitationally influencing and clustering the orbits of distant TNOs14. If this massive body indeed exists in this region, the stability of Sedna-like objects could serve as a test. In other words, the presence of Sedna, 2012 VP113, Leleākūhonua and Ammonite should indicate either negligible or strong dynamical interactions with the putative planet. Therefore, we employed the Mercury integrator, using the same clone generation as for the orbital stability analysis (Methods), to simulate three clones of each of the four Sedna-like objects for 1 Gyr, applying an a > 10,000 au criterion for ejection.

We incorporate planetary orbits proposed in previous studies into these simulations to investigate the influence of the hypothetical planet on the stability of the four TNOs. Brown and Batygin32 estimated the mass and orbit for the hypothetical planet of \(M=6.{2}_{-1.3}^{+2.2}\,{M}_{\oplus }\), \(a=38{0}_{-80}^{+140} {\mathrm{au}}\), i = 16° ± 5° and \(q=30{0}_{-60}^{+85} {\mathrm{au}}\). We selected four sets of orbital elements of the hypothetical planet in this investigation: (1) maximum likelihood, (2) maximum perihelion distance, (3) minimum perihelion distance from table 2 of ref. 32 and (4) nominal values from fig. 8 of ref. 32 (Table 1). As the mean anomaly of the hypothetical planet is not well constrained in previous studies, we selected mean anomaly values of 0°, 60°, 120°, 180°, 240° and 300° in our simulations. As presented in Table 1, the survival rates of Sedna, 2012 VP113 and Leleākūhonua were relatively high. Only 4 out of 216 clones (one Sedna clone and three 2012 VP113 clones) were ejected before the end of the 1 Gyr simulation. By contrast, most Ammonite clones (47 out of 54), except those in Table 1, case (b), experienced orbital instability and were ejected in the simulation. This is expected, as the nominal orbit of Ammonite has a higher probability of close encounters with the hypothetical planet of a similar orbit.

Table 1 The parameters of hypothetical planet and simulation results
Full size table

Brown et al.33 later updated their estimates of planet’s orbital properties to \(a = 50{0}_{-120}^{+170} \, {\mathrm{ au}}\), \(M = 6.{6}_{-1.1}^{+2.6}{M}_{\oplus }\) and aphelion distance of \(63{0}_{-170}^{+290} \, {\mathrm{au}}\). We also performed simulations with the updated nominal orbit (Table 1, case (e)), resulting in the ejection of only one 2012 VP113 clone by the end of 1 Gyr. The nominal orbit elements used here are similar to those in Table 1, case (b), indicating that a planet’s orbit with larger a and q has a lower likelihood of close encounters with Ammonite. Figure 4 illustrates the stability of nominal orbits of the four Sedna-like objects from a representative simulation. The simulation results indicate that Sedna, 2012 VP113 and Leleākūhonua experience strong gravitational interactions with the hypothetical planet, consistent with ref. 32, and exhibit notable clustering of their ϖ. This suggests that these Sedna-like objects would be gravitationally shepherded by such a planet, maintaining relatively stable configurations over the 1 Gyr.

Fig. 4: Orbital stability of Sedna-like objects under the influence of a hypothetical planet.
figure 4

The stability of four Sedna-like objects using the nominal hypothetical planet parameters from ref. 33 (M = 6.6 M⊕, a = 500 au, Q = 630 au (e = 0.26) and i = 15.6°). a–e, The time evolution of a (a), e (b), i (c), q (d) and Δϖ = ϖ − ϖHP (the difference between each object’s longitude of perihelion and that of the hypothetical planet) (e), for all four objects. The results suggest that Sedna, 2012 VP113 and Leleākūhonua are strongly influenced and clustered in longitude of perihelion (ϖ) with respect to this hypothetical planet, whereas Ammonite behaves differently. In d, the dotted and dashed grey lines represent q = 40 au and q = 30 au, respectively.

Source data.

Full size image

By contrast, Ammonite shows different behaviour with the Brown et al.33 planet compared with the other three Senda-like objects. Some Ammonite clones with different mean anomaly values exhibited only temporary clustering, suggesting that slight variations in the initial orbital parameters could affect the simulation results. However, the different stability of Ammonite compared with the other three objects suggests that the definition of Sedna-like objects should consider subdynamical populations if this hypothetical planet exists. It is worth noting that the semi-major axis of Ammonite (252 au) lies near the transition ‘wall’ suggested by Brown and Batygin32 between the nearby uniformly Δϖ-distributed population and the distant clustered population. This proximity to the transition region may explain why some clones of Ammonite’s orbit still experience temporary clustering due to the gravitational influence of the hypothetical planet. Additionally, the orbital pole positions of the four Sedna-like objects show a generally random distribution, rather than the notable clustering of pole positions seen in the sample of objects with 150 < a < 1,000 au and q > 42 au, as illustrated in figure 2 of Brown et al.32. The different orbital influence by the hypothetical planet on Ammonite provides a valuable contrast, emphasizing the range of dynamical behaviours that Sedna-like objects may exhibit in response to the presence of this hypothetical planet. It is important to highlight that primordial alignment (calculated through the perturbations of the four giant planets) and the current presence of a distant planet are mutually exclusive. Further discoveries of Sedna-like objects will clarify which external gravitational influence raised the perihelion of these objects.

Conclusions

The discovery of ‘Ammonite’, the first anti-cluster Sedna-like object with the third largest q among all TNOs, offers a valuable opportunity to evaluate current models of outer Solar System formation and evolution. With a perihelion of ~66 au, Ammonite’s confirmed stable orbit through simulations provides constraints on the possible orbital parameter range of a hypothesized distant and currently undetected planet. Meanwhile, simulations including all four Sedna-like object shows they may have experienced a primordial clustering of perihelion longitudes around 4.2 Ga.

These findings highlight the diversity of orbital properties and dynamical behaviours among distant Solar System objects. Future large surveys will be the key to increasing the number of large-q objects and refining our understanding of the dynamical processes shaping the outer Solar System.

Methods

Observations and orbit fit

The FOSSIL II survey is designed for the pre-discovery (known as precovery), discovery and recovery of TNOs, using the Hyper Suprime-Cam45 on the Subaru Telescope on Maunakea. With 16 closely packed pointings of Hyper Suprime-Cam, each imaged with 270-s exposures, FOSSIL II covers 25 deg2 to an expected limiting magnitude of mr ≃ 25.6. The observation plan was structured into three epochs, each separated by an interval between 60 and 90 days, with three exposures per night spaced by ~80 min (a ‘triplet’) to confirm the on-sky motion of Solar System objects. The initial orbit determination required at least two triplets separated by a few days. The time allocation included three half-nights for precovery and discovery in March and May 2023, and an additional two half-nights for recovery observations in June, with the goal of obtaining at least one triplet in each month. Precovery observations were successfully carried out using the new ‘EB-gri’ broad-band filter46 with 270-s exposures from 18 to 20 March 2023, over three half-nights (acquiring three triplets). However, owing to the weather, the discovery observations in May were limited to only one triplet on 16 May 2023, achieving an average limiting magnitude of mr ≃ 25.2. Unfortunately, all the recovery time in June 2023 was lost due to technical issues. Additional recovery observations in the r-band with longer exposures at 380 s were awarded for three quarter-nights of observation in August 2023.

In the preliminary FOSSIL II TNO candidate list, we identified Ammonite as an object with an extraordinarily high perihelion and barycentric distance. However, due to the slow movement of TNOs, the 5-month observation arc in 2023 was insufficient to accurately determine the perihelion distance. Therefore, a Canada–France–Hawaii Telescope (CFHT) Director’s Discretionary Time (DDT) proposal was submitted for two triplets of observations in July 2024. CFHT secured triplets on two different nights with 380 s exposures in the w-band, which improve the orbit determination of Ammonite. We then explored archived data through the Solar System Object Image Search47,48. The point-spreading functions of the moving object were identified within a one-sigma error prediction ellipse after incorporating the CFHT DDT measurements. We note that one of the precoveries in the 2021 DECam archive was identified in the raw (unreduced) image because the pixels within the prediction ellipse in the publicly calibrated image were resampled due to a bright star streak. After adding the 2021 measurements, even earlier DECam precoveries were identifiable in 2014, near the centre of the one-sigma error ellipse. Overall, these result in a total arc-length of 10.16 years for Ammonite. Supplementary Table 1 summarizes the FOSSIL II and CFHT observations as well as the archival data we used in this study.

The barycentric orbit fit, based on refs. 49,50, to the observations (Supplementary Table 1) in the J2000.0 celestial reference frame is as follows: semi-major axis a = 251.9 ± 0.3 au, eccentricity e = 0.7383 ± 0.0003, inclination i = 10.98° (with error <0.01°), ascending node Ω = 72.104 ± 0.001°, argument of periapsis ω = 198.71 ± 0.03° and perihelion distance q= 65.9 ± 0.2 au, determined from a 10.16-year arc with mean residual of 0.1″.

Long-term stability

As chaotic diffusion51 and minor perturbations may behave differently with various integrators, we performed N-body simulations using two codes: hybrid symplectic/Bulirsch-Stoer in Mercury52 and WHFast in Rebound53. The simulations were conducted with a time step of 180 days, integrating the orbit both forwards and backwards over 4.5 Gyr. To improve computational efficiency, the mass of the terrestrial planets was incorporated into the Sun, leaving only the four giant planets as massive perturbers in our nominal simulations. In addition to the nominal orbit, we selected another two clones with the highest and lowest values of a from a set of 1,000 clones generated using a covariance matrix and a Gaussian random distribution within three sigma of the orbital element errors to accurately account for orbital uncertainties, following the classification method of Gladman et al.54.

Compatibility with a primordial orbital alignment

On the basis of different hypotheses for the formation of Sedna-like objects, the timing when they were primordially implanted to their current orbits varies. For the stellar encounter model, Nesvorný et al.55 argue that the Sedna-like objects were implanted ~10 Myr after the gas disk was dispersed, whereas a rogue planet model generally requires ~100–300 Myr (refs. 6,44) for the planet to continuously lift planetesimals out of the primordial scattering disk. Assuming the rogue planet was initial formed in the giant planet region, it was most likely scattered to large-a orbit right after the giant planet instability56,57, which also triggered the migration of Neptune into the outer planetesimal disk and supplied icy bodies to be implanted into the region of Sedna-like objects. Previous studies58,59,60 showed that the instability must have occurred within 100 Myr after the dispersal of the gas disk. Therefore, in a rogue planet hypothesis, the timing of implantation (which corresponds to the timing when the postulate primordial clustering is tightest) varies from ~100 to ~400 Myr.

Therefore, we argue that t = 10–400 Myr after the disk gas dispersal should be treated as the potential ‘time of interest’ for the primordial alignment. This time interval corresponds to −4.55 to −4.16 Gyr if one assumes the gas disk disperse 10 Myr after the formation of the Solar System 4.57 Ga. We thus evaluated the significance of the clustering by assuming random current values of ϖ for each object, while accounting for the precession rates of the four Sedna-like objects: 610° Gyr−1 for Ammonite, 136° Gyr−1 for Sedna, 284° Gyr−1 for 2012 VP113 and 51° Gyr−1 for Leleākūhonua. Notably, these precession rates reflect the gradual evolution of their orbits due to the perturbations from the giant planets. However, it is important to recognize that this analysis does not account for the orbital uncertainties of these objects, which could affect the precise timing and extent of the clustering.