Semimajor-axis Jumps as the Activity Trigger in Centaurs and High-perihelion Jupiter-family Comets

Following Jewitt (2009), we define Centaurs as objects with perihelia between Jupiter and Neptune (i.e., 5.2 au > q > 30.0 au) and semimajor axes of a < 30 au, while excluding any Jupiter Trojans that happen to fall in these ranges. We also follow the methods of Schambeau et al. (2018) in defining high-perihelion JFCs as those with q ≥ 4.5 au, because comets with such orbital parameters are likely to be the newest and least processed members of the group that have left the Centaur region only a few orbital periods ago (Sarid et al. 2019; Guilbert-Lepoutre et al. 2023). Using these definitions, our target sample includes all known Centaurs (including both active and inactive) and high-perihelion JFCs as of 2022 December.

We created 50 dynamical clones for each target using multivariate normal distributions defined by orbital covariance matrices provided by the JPL Small Bodies Database, ¹⁰ in order to assess the level of divergence due to chaotic effects introduced by the object's orbital element uncertainties as we track its orbital history. Because Centaur orbits evolve in the giant planet region, they are strongly affected by dynamical chaos (e.g., Tiscareno & Malhotra 2003), and numerical integrations of a given object can typically only be considered reliable until a close encounter with a planet occurs, after which orbital clones offer only probabilistic indications of an object's long-term dynamical evolution. Over longer timescales, even these probabilistic indications cease to be meaningful, as the divergence of dynamical clones simply grows too large. Given these limitations, we only consider short integration times over which we can interpret our results with high confidence, namely where there are no close planetary encounters and the evolutionary paths of all clones for a given object are essentially identical.

We conducted backward numerical integrations for the nominal orbit and all clone orbits of each Centaur and JFC in our sample for 200 yr into the past using the Bulirsch-Stöer integrator in the Mercury N-body integration package (Chambers & Migliorini 1997), which has the ability to adapt time-step sizes when shorter time steps may be needed to properly capture an object's dynamical behavior, e.g., during close planetary encounters. We use barycentric elements to display the final orbital evolution. To study the longer-term dynamical histories of our target objects and further investigate active Centaurs and JFCs with less prominent orbital changes, we also conducted backward integrations for 5000 yr. All integrations accounted for gravitational perturbations from the Sun and the eight major planets and used an initial time step of 2 days for 200 yr integrations and a larger time step of 30 days for 5000 yr integrations, respectively, in order to optimize CPU usage. Both of these time-step intervals are sufficient to track the orbital evolution of Centaurs for which the gravitational influence of the inner terrestrial planets is negligible. Our initial test with shorter time steps yielded essentially identical results to integrations using larger time steps over the considered timescales. We have also tested the reversibility of our integration code on the case of LD2, which underwent an extreme orbit-altering deep encounter with Jupiter (see left panels of Figure 1 by conducting forward integration from the end point of the backward integration at t −200 yr. During such a close encounter, the semimajor axis changes, which induces chaos. While our integrations are not perfectly, fully reversible, due to imperfect energy conservation and machine precision roundoff error, on the timescales of the reported a-jumps, our integrations are reversible to ∼1 part in 10⁹ in semimajor axis and ∼1 part in 10⁷ in mean anomaly (where mean anomaly will be the worst-conserved orbital element).

Zoom In Zoom Out Reset image size

Nongravitational accelerations, such as those due to the Yarkovsky effect and asymmetric mass loss from cometary activity, were not included because their effect on JFCs and active Centaurs is expected to be minimal over the time span covered by our integrations.

We performed numerical integrations as described above for all active and inactive Centaurs known as of 2022 December (39 and 268 objects, respectively) and for 17 high-perihelion JFCs. The list of the active objects we analyzed is summarized in Table 1. Upon completion of the integrations, we analyzed semimajor axis, eccentricity, and inclination evolution as functions of time and searched for large deviations from average orbital fluctuations. When such changes were found, we noted both the magnitudes of the changes and their durations. For two selected cases—167P and LD2—integration output was used as input for our thermal model in order to characterize the effect of these bodies' orbital histories on the thermal processing of their nuclei (Section 3).

Our numerical integrations show that all active Centaurs and high-perihelion JFCs analyzed in our work underwent recent rapid orbit-reshaping events in the form of sudden decreases in semimajor-axis distances and eccentricities that we refer to as a-jumps. The majority of a-jumps occurred less than 250 yr ago—only ∼12% of the Centaurs we see active today underwent their most recent a-jump between 400 and 1000 yr ago. The orbital changes were induced by close encounters with either Jupiter or Saturn and have been extremely fast—on the order of several months, in the most extreme cases, to several years. All noted a-jumps occurred before the divergence of the orbital clones when their orbital evolution was essentially identical.

The change in semimajor axis Δa was larger than ∼0.5 au in 74.5% of the active Centaurs and JFCs we investigated (Table 1). The most extreme a-jump occurred in the case of P/2004 A1 (LONEOS), which is now designated as 450P/LONEOS, where the Centaur's semimajor axis decreased by 5.2 au (more than 40% of the original value of 12.7 au) over several months after a close encounter with Saturn (encounter distance of < 0.03 au) in 1992. The remaining Centaurs underwent milder a-jumps by about 0.2–0.4 au inward. We hypothesize that these gentler orbital changes might be sufficient to trigger activity if they occur for objects with higher perihelia that are likely dynamically new in the region. The top left panel of Figure 2 suggests the dependence of the a-jump severity necessary to trigger activity on an object's perihelion distance.

Zoom In Zoom Out Reset image size

The change in semimajor axis in active Centaurs is typically accompanied by a change in eccentricity of similar magnitude, as seen in Figure 2, which shows parameters and the distributions of the element changes for the most recent a-jumps for the objects we investigated. The majority of active Centaurs also had corresponding changes in perihelion distance and inclination, which were typically very small—less than 1% of their original values. Only a handful of Centaurs, including 450P and LD2, with most extreme a-jumps (Δa > ∼10%) are exceptions to this rule. The median value of Δa_Cen = −0.62 au (or −5% of the pre- a-jump value), and the median value of Δq_Cen = −0.08 au (or −0.8% of the pre- a-jump value). JFCs generally appear to undergo more pronounced changes in both elements than active Centaurs, with Δa_JFC = −1.2 au (or −13.9% of original value) and Δq_JFC = −0.4 au (or −7.4% of the pre- a-jump value). In this way a-jumps cause sudden orbital reshaping from eccentric to more circular, forcing the Centaur to experience more frequent perihelion passages than it would have before the orbital change and, most importantly, to receive higher mean energy flux per orbit, which translates into higher mean per-orbit temperature and/or peak temperature (Prialnik & Rosenberg 2009; Gkotsinas et al. 2023). Figure 2 shows a single case—Centaur 166P/Neat, where the semimajor axis in fact increased, but further examination shows that this increase was a result of an earlier a-jump after which the object in question spent several decades on orbits with smaller a, followed by additional close encounters with Jupiter that increased a to the values we see today. This is yet another example of the rapid and complex orbital evolution of objects on the edge of Jupiter's orbit, which can switch back and forth between JFC and Centaur region in a matter of several orbital periods.

Remarkably, a-jumps are not present in the recent orbital histories of the vast majority of known inactive Centaurs over the timescales that we considered here. Our numerical integrations of 268 inactive Centaurs show that they typically undergo limited changes in a and q, with a median maximum absolute change of 0.5% ± 1.0% in semimajor axis over our integration periods. The Kolmogorov–Smirnov (K-S) test comparing the maximum amplitudes in the variations of semimajor axis, eccentricity, and perihelion over the inspected time period between the active and inactive Centaur populations suggests that we can reject the null hypothesis that these two samples were drawn from the same populations at 95% confidence level, meaning that the two groups underwent different dynamical evolution paths. Panels (a) and (b) in Figure 1 show some of the prominent a-jump cases we have identified compared to the orbital evolution of two inactive Centaurs without identified a-jumps in their dynamical histories: 2013 JX₁₄ and 2003 QD₁₁₂. Note that all our plots use barycentric orbital elements to avoid sinusoidal patterns due to Jupiter's orbital period and the related movement of the Sun around the Jupiter–Sun barycenter (e.g., Gladman & Volk 2021), which do not produce meaningful differences in the thermal evolution of distant objects.

167P has a radius of r ∼ 33 km (Bauer et al. 2013), making it a medium-sized Centaur, and is one of the two most distant known to be active: with q = 11.784 au, it was observed to be active at a heliocentric distance of r_h = 12.23 au (Jewitt 2009), which is close to the outer boundary of the AWI crystallization zone (Guilbert-Lepoutre 2012). It was discovered on 2004 August 10 by the Campo Imperatore Near Earth Objects Survey (CINEOS) and appeared asteroidal at the time (Boattini et al. 2007), but a year later it was reported to have developed an asymmetric coma. ¹¹ This object underwent its most recent a-jump 139 yr (or about two orbital periods) ago owing to a distant ∼4 Hill radius encounter with Saturn, as seen in Figure 3. This a-jump was one of the mildest found in our integrations for active Centaurs: the semimajor axis decreased by only 0.24 au (less than 1.5% of the original value), but long-term numerical integrations show that 167P has been slowly drifting inward: its semimajor axis has been slowly decreasing for the past ∼2000 yr from 17 to 16.2 au, where the a-jump we identify occurred at the very end of this drift period. The perihelion distance of 167P gently has only oscillated around a value of 11.8 au with no major changes. The object's orbital clones diverge owing to chaos prior to this point in time, making the the object's orbital history prior to ∼2000 impossible to determine with much certainty. In about 200 yr the semimajor axis will shift again to the pre-jump value, possibly quenching the activity. The a-jump we study here was the latest sudden reshaping of the orbit and could have been the final impulse to heat the nucleus just enough to enable the onset of activity.

The left set of panels of Figure 4 shows the results from our thermal model for 167P: the surface and radial profile temperature of 167P during three different perihelion passages (in 1804 before its 1884 a-jump, in 1938, and during its last perihelion passage in 2003 April). The temperature is seen to increase by no more than a couple of degrees both at the surface and in the interior to a depth of ∼10 m. Below this depth, the radial profiles for each epoch of perihelion passage converge to a temperature of ∼85 K. Interestingly, the hottest temperatures obtained for 167P through the modeling are too cold for AWI-crystallization-driven activity (Prialnik et al. 2004) but are too hot for the survival of subsurface CO or CO₂ ice deposits to be present to drive activity through sublimation (∼50 and ∼80 K for CO and CO₂, respectively). The thermal modeling does show that the thermal wave penetrates deeper after the a-jump; however, at the subsolar point used for the simple thermal model included here the subsurface region below this point is depleted of both CO and CO₂ pure ice species. This is not an unexpected result given the recent works in the literature focused on more detailed analyses of TNOs and Centaur interiors and the survival of pure supervolatile species over the age of the solar system (Davidsson 2021; Lisse et al. 2022; Parhi & Prialnik 2023). A possible explanation of 167P's observed activity could be the presence of subsurface CO₂ ice deposits at higher effective latitudes away from the subsolar point being activated by the deeper penetration of a thermal wave at those locations.

Zoom In Zoom Out Reset image size

In 167P's case, the a-jump may have caused temperatures to spike just enough either to set off sublimation of subsurface CO₂ deposits or to trigger AWI crystallization in layers that were previously intact, as discussed in Guilbert-Lepoutre (2012). Given the time frame over which we modeled the object's temperature evolution and its past dynamical evolution at high perihelion distances bracketed by clone evolution, we can rule out H₂O-sublimation-driven activity. The work of Guilbert-Lepoutre (2012) argues that at such large heliocentric distances (167P was at r_h ∼ 12.3 au when discovered) AWI crystallization should be minimal, and beyond ∼12 au this activity driver can be ruled out for 167P as well. However, there is a possibility that it could have been weakly active even before the a-jump, with activity sustained by CO₂ sublimation, since 167P is very likely a dynamically new, ice-rich Centaur and the temperatures of the nucleus from our model were high enough (just above 80 K) to allow its sublimation. In addition, Davidsson et al. (2021) have also shown that the process of CO segregation, where supervolatiles like CO and CH₄ trapped and preserved in CO2 may be released during sublimation of CO₂ at the front, or from even larger depths, could dominate beyond ∼10 au, where crystallization rates drop. Interestingly, recent James Web Space Telescope (JWST) observations indeed confirm the presence of CO₂ ices on surfaces of other Centaurs at similar heliocentric distances (Licandro et al. 2023), providing observational evidence that the volatile species necessary to produce coma activity are present in this region. However, the identification of 167P's activity driver is beyond the scope of this manuscript and indeed may not be identifiable until gas coma measurements are made of this body in the distant future.

LD2 is a small outer solar system object with a radius of r ≲ 1.2 km (Schambeau et al. 2020; Bolin et al. 2021) that is currently undergoing a much anticipated transition to a JFC via the JFC Gateway (Steckloff et al. 2020), orbiting just inside the Centaur region border with q = 4.578 au. It is therefore outside our formal definition of a Centaur, but given the fact that it experienced an a-jump that carried it inside Jupiter's orbit only 2 yr ago, it is a perfect example of the dramatic orbital changes, and therefore intense thermal environment changes, that a-jumps can produce. LD2 was discovered in 2019 but has exhibited activity in precovery images dating back to 2018 (Schambeau et al. 2020). Our integration results, which are in line with Steckloff et al. (2020), show that LD2 has experienced two rapid a-jumps. The first one occurred ∼175 yr ago, lowering the semimajor axis from ∼9 to 7.5 au. The most recent a-jump occurred in 2017 February and pushed the semimajor axis down to 5.3 au within the next 6 months (see Figure 3). LD2 is also one of the very few active Centaurs where the close encounter led also to a major change in perihelion distance—by almost 1 au. As LD2 was not detected near its expected position in archival images from 2017 March, we can assume that it was not active at the time, suggesting that the onset of activity on LD2 was relatively sudden and took place between mid-2017 and early 2018 (Schambeau et al. 2020). Its orbital clones started to diverge ∼400 yr ago, when it experienced another close encounter with Jupiter, limiting analysis of its dynamical history prior to that point to statistical assessments only. We can only infer that this is likely not the first significant heating episode for LD2 given the proximity to Jupiter: the object's perihelion distance was between 4 and 6 au in the past 5000 yr, with its semimajor axis mostly constrained between 4 and 12 au, and a future a-jump around 2060 will push both its semimajor axis and perihelion distance toward JFC-like values.

The right panels of Figure 4 display the results of the thermal modeling of its most recent a-jump (Section 3), specifically the evolution of the radial temperature profile at the subsolar point over the past ∼100 yr until the year 2030. As can be seen from the figure, a thermal wave penetrates deeper into the interior of LD2's nucleus owing to the a-jump, increasing the temperature by more than 20 K, almost an order of magnitude larger change than in the case of 167P owing to LD2's more dramatic orbital reshaping and also perihelion decrease. We discuss the thermal implications of both the orbit reshaping and a perihelion decrease in Section 5.2. Over the past ∼175 yr, LD2 was in a mostly stable orbit where repeated cycles of interior heating and cooling occurred, as seen by the repeated patterns to the left of Figure 4's top panel. The period following LD2's first a-jump 175 yr ago likely depleted its near-surface volatiles, leaving the nucleus largely inactive for most of the past century. The following close encounter with Jupiter in 2017 February caused the inward journey for LD2 by means of the a-jump and forced the interior heat wave to propagate deeper, triggering phase transitions in the volatiles that have been demonstrated as plausible drivers of the cometary activity we observe today.

Figure 4 (right panels) shows the evolution of surface temperatures, the propagation of heat into the interior, and the interior temperature profiles for three more recent peak periods of surface heating before, during, and after the most recent a-jumps. The intense increase in both surface and interior temperatures is evident, explaining the onset of vigorous activity. The bottom right panel of Figure 4 displays the interior's radial temperature profile for the peak surface temperatures received by LD2 during its pre−a-jump last perihelion passage in 1995 October (blue curve), the peak temperature achieved during 2018 August when the first observational evidence for activity was observed (green dashed curve; Schambeau et al. 2020), and finally for the peak surface insolation received by LD2 post−a-jump (red curve). The transition of LD2's interior's peak temperature from being too cold for efficient water-ice sublimation to being hot enough for vigorous water-ice sublimation provides a plausible explanation for the active behavior displayed shortly after its discovery during its most recent apparition (Bolin et al. 2021; Kareta et al. 2021). LD2's near-surface water-ice reservoir may have undergone a period of vigorous sublimation, while in the deeper interior (layers between 35 and 100 cm) AWI could also still potentially be present and could be heated sufficiently to contribute to observed outgassing due to crystallization.

Our numerical integrations show that there is usually a time lag of varying length between the most recent a-jump and observed activity onset on Centaurs. In several notable cases of strong a-jumps, such as in the cases of 450P/LONEOS, LD2, C/2020 Q2 (Pan-STARRS), and others, activity onsets were almost instantaneous, e.g., within a year, while for other objects lags between a-jumps and observed activity could be several decades to hundreds of years long, with the average value of 109.6 ± 129.9 yr. It appears that this time lag is a real feature considering that several Centaurs with deep prediscovery images still show point-source-like nuclei taken months after a-jumps (e.g., 167P, LD2, 2013 UL₁₀; Mazzotta Epifani et al. 2018; Sarid et al. 2019), but before the objects in question were flagged as active bodies.

In principle, for well-documented cases when activity onset can be identified with reasonable accuracy and precision (e.g., within a few years), the length of a time lag could help to identify the volatiles and processes driving the gas and dust loss and the inner structure of the Centaur nuclei. This is because the length of the lag should depend on the time needed for a thermal wave to travel through nonvolatile surface material before reaching subsurface ice, thus setting off the phase transition of those ices (Gkotsinas et al. 2022).

However, we need to exercise caution and take into account observational bias when estimating the time lag duration on Centaurs with long orbital periods. Our integration results suggest that about half of the active Centaurs underwent a-jumps more than 100 yr ago, but it is difficult to determine when the activity started after that. These objects were discovered and recognized as cometary bodies mostly after the early 2000s, with the exception of Chiron. After all, significant advances in telescope sensitivity and survey coverage have happened in the past few decades, meaning that we have really only had the observing capabilities to routinely detect activity on such distant and faint bodies (V-band magnitudes of <21) in the past 50 yr or so, particularly compared to the length of typical Centaur orbital periods (e.g., Mazzotta Epifani et al. 2017). In addition, Centaurs as a population have also been overlooked in the past decades by all-sky surveys, with their apparent motion falling in between fast-moving near-Earth objects (NEOs) that are the focus of surveys like Pan-STARRS (e.g., Chambers et al. 2016) and slow-moving TNOs that are the focus of deep-imaging surveys like OSSOS (Bannister et al. 2018). It is therefore entirely possible that some Centaurs might have been active long before current observations flagged them as such and the decades- to centuries-long time lag preceding the activity onset is a signature of observational bias.

It is also notable that a-jumps only seem to activate Centaurs within a certain heliocentric distance range: all active Centaurs and JFCs have been observed at heliocentric distances of r_h < 14 au, possibly pointing to the volatile species driving the observed activity. Several dedicated observing surveys of distant Centaurs (Jewitt 2009; Cabral et al. 2019; Li et al. 2020; Lilly et al. 2021) suggest that this observed lack of active objects beyond 14 au is not due to observational biases, but is instead consistent with the heliocentric range where the crystallization of AWI and/or CO₂ sublimation occur, implying that these two processes are the main drivers of activity in the Centaur region (Jewitt 2009; Prialnik & Jewitt 2022), as Centaurs mostly orbit outside the heliocentric range where H₂O sublimation is possible. Indeed, the most recent JWST spectral data show the presence of CO₂ ices on surfaces of Centaurs and also CO₂ emission lines on an active Centaur 2014 OG₃₉₂ that appears stellar (Licandro et al. 2023), providing observational evidence that the volatile species necessary to produce coma activity are present in this region and that the activity could still be ongoing even if it is below our visual detection limit.

Considering the fact that recent a-jumps are almost exclusively present in the orbital histories of active bodies in the Centaur and JFC region and that there are usually only short time lags between the most recent a-jumps and observed activity, our results suggest that the a-jumps may act as triggers of cometary activity in Centaurs when the right conditions are met, i.e., when the a-jump shifts Centaur's orbit into the range of heliocentric distances where either the AWI crystallization or CO₂/H₂O sublimation can take place, and, as discussed in the following sections, the orbital reshaping and/or perihelion change increases the per-orbit insolation of the Centaur. However, how long it takes for the thermal wave to travel through a Centaur nucleus and take effect remains an open question and most likely depends on the size of the object and its physical properties.

Thermally, orbital reshaping via a-jumps modifies the average total energy received throughout a Centaur's orbit, leading to the gradual increase of surface and subsurface temperatures over a larger portion of the orbit. The thermal wave is then capable of reaching deeper layers of the nucleus, potentially producing the hottest environment the Centaurs have ever experienced, as we show in our case studies in Section 4. Moreover, if the a-jump is also accompanied by a decrease in perihelion distance, such as we see only in the few cases of the most pronounced a-jumps, it could lead to a temperature surge during the perihelion passages high enough to allow for sublimation of different volatile species than before. Such "double whammy" a-jumps then change both the energy per orbit and seasonal thermal processes on the Centaur. Perhaps the best examples of this process are Centaurs LD2, 450P, and also 29P. However, for most Centaurs the perihelion distance value remains essentially fixed even if eccentricity and semimajor axis change after a close encounter.

We have calculated the change in the average per-orbit insolation that was caused by the largest variation in a – e over the inspected period (for active Centaurs this change was the a-jump) using Equation (4) in Prialnik & Rosenberg (2009). This work has shown that eccentric orbits can be modeled as circular, receiving the same total energy over an orbital period with the equivalent average circular orbit radius a_c = a ·(1 − e²), where a is semimajor axis and e is the eccentricity of the original eccentric orbit. For comparison, we have also used a time-averaged effective thermal radius ${r}_{T}=a\cdot (1+\displaystyle \frac{1}{8}\ {e}^{2}+\tfrac{21}{512}\ {e}^{4}+{ \mathcal O }({e}^{6}))$ in the calculation, which works better for estimating the energy received at distant parts of the orbit for more eccentric orbital configurations (Gkotsinas et al. 2023). We have found that the resulting pre− and post−a-jump average isolation change differed by only a few percent between both methods, and we conclude that our results are not strongly influenced by the choice of an orbit-averaging method for the type of orbits and the timescales we have investigated. Panel (d) in Figure 1 shows that the inactive Centaurs have typically undergone minimal and mostly negative changes in the average insolation, while active Centaurs (panel (c)) can have an increase in average insolation up to 34% (in case of LD2). The K-S test confirms that, regarding the changes in insolation per orbit, the null hypothesis can be rejected at the 95% confidence level, suggesting that the two groups have undergone statistically different thermodynamical evolution paths in the near past.

Interestingly, Figure 2 shows that several currently inactive Centaurs have experienced a-jumps of similar magnitudes to active Centaurs and have perihelion distances in the range expected for active Centaurs. Our calculations show that for the majority of these objects the orbital changes in fact did not lead to a substantial increase of per-orbit insolation; quite the opposite. There are only five notable objects that experienced both an increase in the average energy per orbit and a (gentle) decrease in both the perihelion and semimajor axis. All five Centaurs are discussed in Section 5.3. It is also possible that these seemingly inactive objects could have spent more time in the region and experienced prolonged periods of heating and thermal processing in the past, leading to volatile depletion in the shallow subsurface reservoirs, which essentially meant that they entered a period of dormancy (Gkotsinas et al. 2022). Or the differences are simply due to physical parameters such as the nucleus size, the position and depth of subsurface layers, or the object's obliquity and thermal inertia, which play an important role in determining whether and when a given object will be activated (e.g., Prialnik et al. 2004; Guilbert-Lepoutre 2012; Davidsson et al. 2021; Lisse et al. 2022).

Centaur residence times vary by several orders of magnitude as their orbits evolve from near-Neptune parts of the solar system toward Jupiter (e.g., Sarid et al. 2019; Gkotsinas et al. 2022). They spend the majority of the time in the outskirts of the Centaur region, but once a Centaur passes the orbit of Saturn, its evolution speeds up significantly and it typically takes only several orbits before it leaves the region either via the JFC pathway or some other way. Our numerical integration results show that the magnitude of an a-jump is indeed dependent on the perihelion distance of the given Centaur, as can be seen in Figure 2.

This is because Centaurs with smaller perihelia have a higher chance to undergo a series of close encounters with Saturn or Jupiter that drive the a-jumps. This relationship between the a-jump magnitude and perihelion distance also indicates that the more distant Centaurs need a smaller impulse to be activated, and this is very likely because these bodies have just emerged from the "freezer" and have undergone very little, if any, thermal processing yet. These distant Centaurs most likely still contain ice deposits on the surface (Licandro et al. 2023) or in shallow subsurface layers (e.g., Barucci et al. 2002) and are able to undergo phase transition even with small temperature change, which appears to be the case with 167P and a handful of other active Centaurs with high perihelia and mild a-jumps (Δa < 0.3 au). With these objects, their a-jumps occur at the end of a drift period, during which semimajor axis and eccentricity slowly decrease over several millennia. Mild a-jumps are then produced by distant close encounters with Saturn, typically within several Hill radii. We speculate that once these "drifters" reach sufficiently small semimajor axes, more pronounced close encounters will start occurring, propelling them onto the fast evolutionary track toward JFC orbits. Apart from 167P, there are two other active drifter Centaurs with similar orbital history: 2003 QD₁₁₂ and (248835) 2006 SX₃₆₈.

Our results are in line with thermal modeling work by Gkotsinas et al. (2022), who show that Centaur progenitors of the JFCs undergo multiple periodic heating episodes during their transition through the Centaur region that warm up consequently deeper and deeper layers of the nuclei. Centaurs gradually lose volatiles during every episode until the reservoirs of the thermally pristine or unprocessed material are deep enough that even at perihelion distances typical for JFC region the thermal wave does not reach them, or smaller Centaur nuclei will be thermally processed all the way through. In fact, the time spent in transient orbits is critical: Gkotsinas et al. (2022) have shown that JFCs spending long periods of time on transient orbits (i.e., with longer lifetime) are more susceptible to be heated up at greater depths and are prone to losing all free condensed hypervolatiles down to their core. This would explain why some inactive Centaurs have orbits very similar to active objects—either these bodies are completely processed and dormant or, on the other hand, they could be "dynamically younger" and could still have pockets of volatiles in the deeper layers with the possibility of reactivation if a future a-jump can reshape the orbit again and thermal wave could propagate deeper.

Our case study of 167P (Section 4) further underscores that the thermal processing the nucleus had undergone and the depth of its ice deposits might play an important role—dynamically new objects with volatile pockets in shallow underground layers can be reached by thermal waves that penetrate only several centimeters to meters deep and is triggered by a mild a-jump. In contrast, bodies that have undergone multiple periods of activity and have their upper layers depleted of volatiles likely need a major a-jump that allows for more pronounced heating and a subsequent thermal wave reaching deeper parts of the nucleus's interior.

Does this mean that every new Centaur has the potential to be active under certain circumstances? It is likely, since they have formed beyond snowlines of several volatile species, and even though primordial CO ice deposits have most likely already sublimated since their formation time (Lisse et al. 2021; Licandro et al. 2023), the majority of Centaurs could be expected to contain volatile pockets hidden deeper in the nucleus that have remained untouched until present day. As the Centaurs are making their way through the region by means of planetary perturbations, the sudden change in thermal conditions induced by a-jumps could eventually reach the sublimation temperatures for different volatile species, or the threshold at which AWI crystallizes, and set in motion a chain of events leading to an outburst. This way, a Centaur will transition between different activity drivers throughout its dynamical evolution. Furthermore, depending on the type of volatiles present and the size of the deposits, Centaurs activated this way could sustain gas and dust production for months, or even for several orbits depending on the speed at which the thermal wave propagates through the body until the activity-driving material is depleted. For example, Lilly et al. (2021) proposed that the propagation of the AWI crystallization front and sudden pressure from escaped volatile molecules could burst out and create an opening on the surface, or a landslide and expose the volatiles hidden under several skin depths, which could sublimate and sustain activity. Such regional events and large pits have been observed in situ on comet 67P by the Rosetta spacecraft (Steckloff & Samarasinha 2018; Jindal et al. 2022).

We use our a-jump hypothesis to identify Centaurs that have the potential to become candidates for outbursts in the near future. Jewitt (2009), Li et al. (2020), and Lilly et al. (2021) have shown that the Centaur activity is associated with orbits that have perihelion distance less than ∼14 au, because the vast majority of active Centaurs were at the heliocentric distances r_h < 14 au when they were detected, with the sole exception of Chiron. However, Chiron is an unusual object, the second largest Centaur known (r ∼ 120 km), and apart from its activity, there is evidence of a ring system surrounding it (Ortiz et al. 2015).

Based on our results, a good future outburst candidate would be (a) a seemingly inactive object from the low-perihelion group (q < 10 au), (b) one that is currently at heliocentric distance less than r_h < 14 au, and (c) one that has experienced an a-jump deeper than 0.5 au in the past 200 yr that yielded a positive change in the average per-orbit insolation >2%. Another favorable condition for possible observable activity would be if the Centaur is also nearing its next perihelion passage. A particularly interesting indicator of past activity could also be photometric surface colors, as active Centaurs have almost exclusively photometrically neutral surfaces akin to JFCs, which were proposed to be the results of blanketing by ejecta (Jewitt 2015).

There are currently only 12 Centaurs in the MPC database that fulfill the first three criteria, but only three of them are approaching perihelion in the next 15 yr or so: 534251 (2014 SW223), a small, D ∼ 10 km Centaur that experienced a large, 1.7 au a-jump 360 yr ago; (31824) Elatus, a medium-size Centaur with a diameter of r ∼ 28 km that experienced a 1 au a-jump 10 yr ago; and (32532) Thereus, a large Centaur with r ∼ 43 km that underwent a 0.9 au a-jump 100 yr ago. All three objects should be considered high-priority targets for observational monitoring to search for activity. Of the three, only Thereus and Elatus have measured photometric colors. Thereus falls into the gray group (Tegler & Romanishin 2003; DeMeo et al. 2009), which possibly suggests that it could have undergone activity periods in the past, while photometric measurements place the color of Elatus to the outskirts of the red group hinting on a primordial red surface (Peixinho et al. 2003; Tegler & Romanishin 2003). The three Centaurs are currently at heliocentric distances of 10.3, 15.5, and 11.6 au, with apparent brightness of V = 23.3, 22.6, and 20.2 mag, respectively. The remaining nine Centaurs from the group (2013 CJ₁₁₈, 2015 KH₁₇₂, 2013 AS₁₀₅, 2014 ON₆, 2008 RG₁₆₇, 2004 VP₁₁₂, 2014 LR₁₄, 2010 LJ₁₀₉, 2001 XA₂₅₅) are currently moving away from perihelion, fading rapidly past V ∼ 25, and as they have orbital periods comparable to a human life span, it will be up to the next generation of astronomers to search them for possible future activity.