Event overview
During this flyby, Mercury’s distance from the Sun was 0.315 astronomical units (AU). Figure 1a shows the location of the Earth, BepiColombo, Solar Orbiter and STEREO-A spacecraft at the time of the flyby. BepiColombo entered the Hermean environment through its northern magnetosphere, tailwards of the cusp, and exited through the dayside southern magnetosphere (Fig. 1b). The magnetic field is taken from the Korth–Tsyganenko–Heyner (KTH22) modular Mercury magnetic field model22,23,24, using a disturbance index (DI) value of 27. This value indicates moderately disturbed magnetospheric conditions and was obtained by optimizing the model to match the inbound magnetopause crossing observed by the MPO magnetometer (Methods; Extended Data Fig. 1).
Fig. 1: Overview of the BepiColombo 4th Mercury flyby.
a, Location of BepiColombo, STEREO-A, Solar Orbiter and Earth on 4 September 2024 compiled using Solar-MACH67. b, Trajectories of BepiColombo’s fourth Mercury flyby on 4 September 2024 in the XMSM–ZMSM (left) and YMSM–ZMSM (right) planes in units of Mercury radii (2,439.7 km). The thicker black line corresponds to BepiColombo’s trajectory with the black data points corresponding to the spacecraft’s location at intervals of 6 min, and the black arrow shows the direction of travel. The blue line in each panel represents a model magnetopause and the black lines correspond to the modelled average magnetic field calculated using the KTH22 model. c, BepiColombo SIXS-P proton (top) and electron (bottom) flux measurements for the selected energy channels. The proton channels combined in pairs of two consecutive channels to increase statistics. The flyby interval is bordered by a pair of dashed lines and shaded grey, to the right of the plot.
Figure 1c shows the selected channels representing energetic proton (~1–9 MeV) and electron (~70 keV to 2 MeV) fluxes, measured by the SIXS particle detector (SIXS-P)21 instrument onboard BepiColombo/MPO (Methods). These fluxes began to enhance on 1 September 2024, at approximately 15:00 UT, with further enhancements occurring on 3 September 2024. During this period, the Sun was close to its activity maximum, and multiple prominent fast CMEs were detected by the coronagraphs onboard the SOHO (located at the Lagrangian Point L1, about 1.5 million kilometres from the Earth towards the Sun) and STEREO-A spacecraft. In addition, several intense X-ray flares were observed by both the Geostationary Operational Environmental Satellites (GOES) and the Solar Orbiter/STIX instrument, which at the time was positioned almost on the opposite side of the Sun to Earth.
The energetic particle fluxes at BepiColombo remained elevated and steady as the spacecraft approached Mercury. A clear and abrupt decrease in both electron and proton fluxes was observed late on 4 September, coinciding with BepiColombo’s CA to the planet. After this flyby, fluxes recovered to their previous elevated levels, indicating that the decrease was caused by the flyby itself, rather than by subsequent solar activity variations.
The SEPs follow approximately interplanetary magnetic field lines, which generally have relatively low curvature. However, scattering from irregularities in the solar wind and magnetic field leads to cross-field transport, causing their directions to spread over a wide range of angles. As a result, Mercury is exposed to energetic particles arriving from multiple directions rather than from a single field-aligned beam.
Overview of energetic electron and proton variations
During the flyby, the SIXS-P instrument measured energetic particle fluxes using three of its five detectors, each having a viewing cone of approximately 25° wide (Methods). Side 0 was directed towards Mercury, while side 1 and side 2 were oriented upwards and downwards, respectively, and both viewing westwards (Fig. 2).
Fig. 2: SIXS-P detector side orientations.
a, Three-dimensional visualization of the orientation of the boresight vectors for SIXS-P detector sides 0, 1 and 2. The light-blue shaded region represents a magnetopause boundary modelled with the KTH22 model for a distance from the Sun of 0.315 AU and a DI of 27/100. The day and night sides of the planet are shown as well as the spacecraft’s position (black dots) at intervals of 6 min. The direction of the boresight vectors for sides 0, 1 and 2 detectors are shown by the brown, green and orange arrows, respectively. b, Plot of the BepiColombo trajectory in the X–Y plane showcasing the westward (clockwise) and eastward (anti-clockwise) directions, as well as the proton gyration direction in the Mercury’s magnetic field. Side 0 points towards the planet, and sides 1 and 2 point approximately tailwards.
A closer examination of the SIXS-P measurements around the time of the CA at 21:48 UT reveals a clear directionality in the particle flux response and notable differences in the electron and proton trends (Fig. 3; see Extended Data Figs. 2 and 3 for separate side 1 and side 2 electron and proton fluxes). The most dramatic decreases are observed by side 0.
Fig. 3: SIXS-P Proton and electron flux measurements.
a, Proton (combined channels P1 + P2, P3 + P4, P5 + P6 and P7 + P8) and electron fluxes (channels E1, E3, E5 and E7) during BepiColombo’s fourth Mercury flyby on 4 September 2024 with 30-s averaging of the data. b, Combined proton channels from P1 + P2 to P7 + P8 and electron channels from E1 to E7 as 10-s averages shown separately for side 0 (corresponding plots for sides 1 and 2 are given in Extended Data Figs. 2 and 3). The horizontal bars the one-count levels, representing the instrumental detection threshold. The grey shaded regions show the intervals when the lowest side 0 proton channels are contaminated by electrons (Methods). The dashed line shows the time of CA at 21:48 UT, and the dashed-dotted lines the inbound and outbound magnetopause (MP) crossing times from the KTH22 model with DI of 27.
The side 0 proton fluxes begin to decrease gradually around 21:43 UT. Near the CA, they drop sharply and finally disappear across all energy channels. The grey shaded enhancements at the lowest proton energies observed during the dropout is discussed in detail later. By contrast, on sides 1 and 2 the proton fluxes remain relatively stable throughout the flyby, showing only minor reductions before the CA.
The electron fluxes, on the other hand, exhibit an abrupt decrease on all investigated sides and energies starting around 21:46 UT, a few minutes before the CA. The fluxes recover sharply on side 0 approximately 2 min after the CA (~21:50:30 UT), although relatively large variations continue beyond the marked magnetopause crossing time. By contrast, the fluxes observed by sides 1 and 2 recover more gradually but stabilize relatively early to resemble profiles preceding the dropout.
Proton flux variations and planetary shadowing
The well-known east–west effect, in which charged particles are bent by the planetary magnetic field, can lead to directional differences in the detected fluxes, for example, refs. 25,26. In Mercury’s magnetic field, energetic protons have relatively large gyroradii, comparable to the gradient/curvature scales of its magnetosphere (Fig. 4b). For example, a 2-MeV proton in a 200-nT magnetic field would have a gyroradius of approximately 1,000 km. Therefore, these particles cannot be trapped within Mercury’s magnetosphere but instead impact to the surface in a process referred to as planetary shadowing. Shadowing may also arise from purely magnetic deflection, in which high-energy particle trajectories are bent away from the planet above the detector, for example, ref. 27. However, for ≳1-MeV protons, planetary shadowing is expected to dominate, because their paths are only weakly bent by the Mercury’s magnetic field.
Fig. 4: Particle motion in Mercury’s magnetic field.
a, Magnetic field magnitude and components estimated from the KTH22 model. B field, the magnetic field in general; |B|, the magnitude of the magnetic field, taken from the vector field. b,c, Proton and electron gyro radii calculated for SIXS-P energies (Fig. 3). d,e, Loss cone critical angle (black and grey circles) and the mirror ratio (blue squares) for the northern and southern hemispheres, calculated using the KTH22 model. The mirror ratios are defined as the ratio of the magnetic field strength along the local field line (towards north and south from the spacecraft location, respectively) and the field magnitude at the spacecraft’s position. f, Electron PAs with the shaded areas showing the 25° viewing cones and with the northern and southern loss cone widths shown by black circles and grey squares, respectively. The PAs are defined as the angle between the boresight unit vector for each detector and the magnetic field unit vector. They are obtained from the KTH22 model instead of direct measurements due to calibration issues during the cruise phase, when MPO and MMO are stacked with the Mercury Transfer Module (MTM) in a magnetically unclean configuration. The dashed thick vertical line indicates the CA and thinner vertical lines the strongest electron dropout.
On the dayside near the equatorial region, Mercury’s magnetic field points northwards (upwards), causing positively charged protons to gyrate clockwise when viewed above, as depicted in Fig. 2. Consequently, more protons are expected arriving from the west than from the east. The tailward-facing SIXS-P detectors (side 1 and side 2) should therefore capture protons with minimal obstruction. By contrast, Mercury-facing side 0 is expected to exhibit large dropout as protons may intercept the planet’s surface before their path curves enough to reach the detector. These expectations are in agreement with the observations presented in Fig. 3. The small flux reductions on sides 1 and 2 before the CA are probably due to partial obscuration of the detector by the planet combined with the wide angular width of their viewing cones (half-width of approximately 25°). The above described expectations and observations are also in agreement with the Toffoletto–Hill field modelling28 of allowed (yellow regions) and forbidden (blue regions) proton trajectories, shown in Fig. 5; side 0 corresponds largely forbidden trajectories that intersect the planet, whereas sides 1 and 2 are on allowed orbits meaning the protons can reach the detector sides without intersecting the planet.
Fig. 5: Modelled allowed and forbidden proton trajectories.
Contour plots of 0.99–1.25-MeV (SIXS-P P1 + P2) and 11.0–31.9-MeV (SIXS-P P7 + P8) protons that have been simulated using the Toffoletto–Hill field model with backwards tracing from the spacecraft’s position at 21:49, 21:55 and 22:01 UT. The coordinates are the local zenith/azimuth at the spacecraft position and the azimuth is measured from the local horizontal north towards the east. The yellow colour indicates allowed trajectories, that is, particles that could have accessed the detectors from interplanetary space, whereas the blue colour indicates forbidden trajectories, that is, particle trajectories that would intersect the planet. The planetary shadow and a penumbra (a chaotic region) between the allowed and the forbidden trajectories (greenish colour) is also seen. Three view cones (sides 0, 1 and 2) of SIXS-P are also indicated.
On side 0, higher-energy protons reach the one-count levels earlier, as their larger gyroradii allow them to intersect the planet sooner. The enhancement observed in the lowest energy proton channels between 21:51 and 21:53 UT (Fig. 3, grey shaded area) is due to electron contamination; during this interval, the P1 proton channel, when converted to an equivalent electron flux, closely matches the measured electron energy spectra (Methods; Extended Data Fig. 4). Detections before this interval are, however, real protons counts as they occur during a complete electron dropout.
The fluxes recover earlier for higher-energy than for lower-energy protons, whereas the Toffoletto–Hill simulation shows higher-energies to have a larger fraction of forbidden trajectories at this time. Although higher-energy protons are generally expected to impact the planet for longer, the delayed recovery observed for lower-energy protons reflects their greater sensitivity to the magnetic field; due to their smaller gyroradii, they are more easily bent by the field and prone to effects related to field curvature and small-scale magnetic variations, leading to enhanced drifts and pitch-angle scattering, for example, ref. 29. The discrepancy between the model and observations therefore probably arises from the limitations of the Toffoletto–Hill magnetic field formulation in accurately producing the magnetopause and field geometry near the dawn-flank.
Origin and variations of high-energy electrons
During the investigated period, the electrons observed by SIXS-P are expected to be primarily of solar origin that have become quasi-trapped within Mercury’s magnetic field, rather than being produced by internal magnetospheric processes such as substorms30,31,32,33,34. Due to their smaller gyroradii (Fig. 4c), electrons originating from the Sun/solar wind can only access Mercury’s magnetosphere either through the polar cusps or field lines opened by magnetic reconnection between the interplanetary magnetic field and Mercury’s magnetic field. Once electrons gain access to the magnetosphere, they start curvature drifting eastwards. Relatively constant electrons fluxes observed during most of the flyby are consistent with Mercury having been exposed to high fluxes of SEPs for several days (Fig. 1), suggesting sufficient time to diffuse, resulting in a stable flux signature.
Support for the solar origin of these electrons is given by the relatively hard energy spectra with spectral indices typical of SEP events, for example, refs. 35,36, as shown in Fig. 6b–d. The indices also remain at similar values for most of the flyby, except around the CA, where they become considerably harder, exhibiting a shallower slope. During this interval, the side 0 energy spectra (Fig. 6e; Methods) reveals an enhancement at energies peaking around ~200−400 keV. When the affected channels from E3 to E5 are excluded, the derived spectral index becomes consistent with the SEP spectra. The pitch angle (PA) range sampled by the side 0 detector coincidentally approaching 90° (Fig. 4f), suggests that the observed electron spectra represent a locally further energized SEP population. At this time, whistler waves were reported in Mercury’s magnetosphere37, which, in the Earth space environment, are known to efficiently accelerate electrons to relativistic energies, for example, ref. 38. Electrons with PAs close to 90° are expected to be preferentially enhanced by whistler waves, as they remain confined near the magnetic equator where these waves are most intense and frequently observed. Furthermore, the side 0 viewcone lies outside the calculated loss cone at this time (Fig. 4f). By contrast, electrons observed by sides 1 and 2 correspond to more field-aligned PAs, causing them to spend less time in the equatorial region and thus be less affected by whistler waves. Their viewcones also partially intersect the loss cone, allowing a fraction of these electrons to precipitate.
Fig. 6: Evidence for solar origin of energetic electrons.
a–d, SIXS-P (a) 60-s electron fluxes averaged over channels E1–E6 and spectral indices for sides 0 (b), 1 (c) and 2 (d), averaged from 1-s spectral indices. Error bars indicate the standard deviation of each 60-point sample. The dashed black line represents the time at CA. Side 0 spectral indices between 21:47 and 21:50 UT are excluded due to low counts. The shaded regions indicate intervals where spectral indices are probably affected by local energization (side 0; brown shaded area) and proton contamination (side 1 and 2; blue shaded area). e, SIXS-P side 0 electron spectra showing the local energization where the blue star is the P1 channel converted to equivalent electron channel (Methods). f, Total SIXS-P electron counts during the fourth Mercury flyby overlaid with a heat map of bursty energetic electron event observed during MESSENGER’s 8-h orbits from April 2012 to December 2013 plotted as a function of invariant latitude and magnetic local time (MLT), following ref. 39. The map includes electron events up to 200 keV, corresponding to SIXS-P channels E1–E4. The solid white line denotes the average open-closed field boundary.
However, the harder spectral indices on side 1 and 2 are the result of proton contamination (note that unlike on side 0, the proton fluxes on these sides did not drop). These detectors observed high proton fluxes throughout the event and converting the affected electron channels to equivalent proton fluxes shows that they align well with the proton energy spectra around the CA (Extended Data Fig. 5). The converted electron channels lying slightly above the proton energy spectra suggest the presence of a minor electron population, probably accelerated by whistler-mode waves, also detected on sides 1 and 2. The proton contamination also explains why the electron fluxes on sides 1 and 2 do not decrease as low as on side 0.
Further evidence of solar origin comes from a comparison of SIXS-P observations with a statistical 8-h orbit map of MESSENGER high-energy electron bursts, in which SEP events were excluded and only electrons accelerated during magnetospheric substorms were selected (Fig. 6f). During its fourth flyby, BepiColombo observed elevated electron counts over a considerably broader region than where MESSENGER detected electron bursts (with 12-h orbit map39 the overlap is even smaller).
Electrons cannot be shadowed by the planet through the same mechanism as protons, as they circle in the opposite direction (east to west) and have substantially smaller gyroradii. Instead, the dropout may occur when their PAs are within the local loss cone (Methods). In the Hermean magnetosphere, these conditions must be evaluated separately for the two hemispheres, as Mercury’s magnetic field is much weaker in the southern hemisphere than the northern hemisphere40,41.
The northern loss cone half-width reached a maximum of about 60° close to CA, whereas the southern loss cone spans 90° between 21:38 and 21:50 UT (Fig. 4d–e), implying that all PAs should precipitate. However, for most of the periods when PAs are predominantly within the loss cone, the electron fluxes remain relatively steady. A possible explanation is that precipitation does occur, but the field lines are opened by magnetic reconnection, allowing electrons to be continuously replenished by the ongoing SEP event. This interpretation is consistent with the observed SEP-like spectral indices and with reconnection at the Hermean magnetopause being a common process for nearly all interplanetary magnetic field orientations, for example, ref. 42. The onset of the electron dropout occurs slightly below the invariant latitudes at which MESSENGER ceased observing energetic electron bursts. This suggests that SEP electrons accessing along reconnection-opened field lines are also limited from the innermost regions and, if some do enter, they precipitate rapidly.