Disappearing Solar Wind at Mars: Changes in the Mars‐Solar Wind Interaction

On 26 December 2022 the solar wind density dropped by over an order of magnitude and remained low for about a day. We have utilized in‐situ plasma measurements made by the Mars Atmosphere and Volatile EvolutioN mission to determine how this change affected the Mars‐solar wind interaction. During this time period, on inbound orbit segments, MAVEN sampled the terminator ionosphere, which switched from a magnetized to unmagnetized state immediately following the minimum in solar wind density. The magnetic field amplitude was typically 5–10 nT within the upper ionosphere prior to the event and consistently <1 nT after. During the event the magnetic pressure dominated immediately above the ionosphere while within the ionosphere the ionospheric plasma pressure dominated. The high altitude terminator ionosphere remained in this unmagnetized state throughout the event, suggesting that it was the new equilibrium state of the system. The terminator upper ionosphere returned to its original magnetized state once the solar wind density had recovered. The outbound orbit segments sampled the dayside subsolar region which remained magnetized throughout the event: the magnetization state of the ionosphere varied locally, dependent upon the solar zenith angle and corresponding incident solar wind dynamic pressure. Such conditions are different to the commonly reported unmagnetized ionospheric state at Venus during solar maximum conditions, where the interplanetary magnetic field is repelled from the entire dayside ionosphere. Drastic changes in the upstream solar wind are able to change the Mars‐solar wind interaction state on timescales less than one MAVEN orbit (∼3.5 hr).

For purely unmagnetized bodies such as Venus, the dayside ionospheric thermal plasma pressure can be large enough during solar maximum to stand off the solar wind dynamic pressure about 70% of the time (Luhmann et al., 1980).Under such conditions the solar wind and ionospheric plasmas are clearly separated on the dayside, with the upper ionosphere boundary characterized by a steep gradient in plasma density.Due to pressure balance between the ionospheric plasma and the overarching solar wind dynamic pressure, the solar wind IMF is unable to penetrate into the ionosphere, and the magnetic field strength drops close to zero within the upper ionosphere (Elphic et al., 1980;Phillips et al., 1985).This state is referred to as "unmagnetized".The photoionizing flux is lower during solar minimum leading to a smaller ionospheric peak density and an ionospheric plasma pressure that is not strong enough to withstand the solar wind dynamic pressure.In this case the IMF can diffuse into the ionosphere and large scale, strong (several 10s nT) magnetic fields are frequently observed there (Luhmann & Cravens, 1991).In this case the ionosphere is referred to as being in a "magnetized" state.
Mars is similar to Venus in that it does not possess a global dipole magnetic field, but it is a unique body in our solar system because it does possess localized crustal magnetic fields that act like mini-magnetospheres that rotate with the planet (Acuna et al., 1999).Mars' magnetosphere is thus termed to be "hybrid" because it possesses elements of both magnetized and unmagnetized magnetospheres.The Martian ionospheric thermal plasma pressure is large enough to stand off the solar wind dynamic pressure about 50% of the time and in these cases steep gradients in plasma density are observed at the dayside upper ionosphere, similar to the Venus-like unmagnetized state (e.g., Duru et al., 2009;Vogt et al., 2015).On a few percent of orbits localized regions of the ionosphere are observed to be unmagnetized but these features are thought to arise from the interaction between the IMF, crustal magnetic fields and ionosphere, rather than a Venus-like global interaction with the solar wind (Fowler, Ortiz, et al., 2022)).
For these unmagnetized bodies the state of the solar wind-ionosphere interaction (magnetized or unmagnetized) depends upon the "internal" and "external" conditions, that is, is the ionospheric plasma pressure large enough to balance the incident solar wind dynamic pressure?At Venus this condition is met at solar maximum when the peak ionospheric density is large and the solar wind dynamic pressure is weaker, but not at solar minimum when the peak ionospheric density is reduced while the solar wind dynamic pressure is larger (e.g., Richardson and Wang (1999)).For the case study at Mars presented in this manuscript, we focus on how the solar wind-ionosphere interaction changes when the internal conditions (ionospheric density and plasma pressure) are kept approximately constant but the "external driver" (solar wind dynamic pressure) dropped and then recovered by over an order of magnitude during the time span of 1-2 days.We show that at the terminator region the state of the solar wind-ionosphere interaction transitioned from magnetized to unmagnetized in response to this upstream variability, remaining in that state for at least two consecutive MAVEN orbits, until the solar wind density began to recover to pre-event values.
The remainder of this paper is organized as follows: the datasets utilized are described in Section 2; detailed in-situ plasma observations of the orbit immediately following the minimum in solar wind density are presented in Section 3. In-situ plasma observations spanning multiple orbits surrounding the event are present in Section 4. Our discussion is presented in Section 5 and we conclude in Section 6.

Data Sets
The disappearing solar wind event was observed at Mars by NASA's Mars Atmosphere and Volatile Evolu-tioN (MAVEN) mission on 26 December 2022 and data in this study are analyzed from that day and those surrounding it.During this time period MAVEN's elliptical orbit had periapsis and apoapsis altitudes of around 180 and 4,500 km, respectively, with periapsis sampling the nightside ionosphere.MAVEN's orbit period was roughly 3 hr and 39 min.This study utilizes observations from the Langmuir Probe and Waves (LPW, Andersson et al., 2015), Magnetometer (MAG, Connerney, Espley, Lawton, et al., 2015), SupraThermal And Thermal Ion Composition (STATIC, McFadden et al., 2015), Solar Wind Ion Analyzer (SWIA, Halekas et al., 2015), and The thermal electron density and temperature are provided by LPW via the analysis of measured current-voltage (I-V) curves using the methods described in Ergun et al. (2015Ergun et al. ( , 2021a,b),b).Values are provided in level 2 (L2) science data files.The 3D magnetic field is measured at 32 Hz by MAG and data are presented here in the Mars Solar Orbital (MSO) coordinate system, defined as X pointing Sunward along the Mars-Sun line; Z pointing north out of the ecliptic plane, and Y completing the right-handed coordinate system, pointing approximately opposite to Mars' orbital motion.STATIC is an electrostatic top hat analyzer that measures ions from 0.1 eV up to 30 keV in energy; a time of flight velocity analyzer is utilized to distinguish the major ion species at Mars (H + , He ++ , O + , O 2 + , CO 2 + ).The STATIC level 3 (L3) ion density and temperature data products are utilized in this study, described in Fowler, McFadden, et al. (2022) and Hanley et al. (2021), respectively.Observations of solar wind ions between energies of 25 eV and 20 keV, including density and temperature moments, are provided by SWIA, an electrostatic top hat analyzer.Observations of suprathermal electrons at energies 3 eV-4 keV are provided by SWEA, an electrostatic top hat analyzer.

Observations
An overview of the solar wind properties during the disappearing solar wind event, as observed at Mars and Earth, is given in Halekas et al. (2023).The time period of interest spans December 23rd to 29th 2022.On December 26th, over the course of a few hours, the solar wind density dropped by over an order of magnitude (from 2 to 3 cm −3 to 0.1-0.2cm −3 at Mars).This density minimum lasted for approximately one day and gradually recovered to its pre-drop value over the course of roughly a second day (Figure 1 in Halekas et al. (2023)).In this study we focus on MAVEN observations made during the orbit immediately following the minimum in solar wind density, to understand how the Mars-solar wind interaction changed during such an extreme and sudden drop in solar wind density.We also study neighboring orbits to determine how the Mars-solar wind interaction system evolved throughout this event.

Overview
An overview of the case study orbit immediately following the minimum in solar wind density ("the event") is shown in Figure 1, with detailed descriptions of the electron and ion plasma measurements following in Sections 3.2 to 3.4.The orbit trajectory is shown in the MSO coordinate system in Figure 1 O1-O3: apoapsis sampled post-noon in the northern dayside hemisphere and periapsis sampled the post-midnight ionosphere in the southern hemisphere.The inbound segment, which is studied in detail here, sampled the dawn terminator region over northern and equatorial latitudes.The outbound segment, which is not studied in detail here, passed close to the sub-solar point on the dayside.
Corresponding time series observations made by MAVEN are shown in panels a-h.While the main focus of this study is the inbound orbit segment we also show the outbound segment here to provide full context of the orbit.As noted by Halekas et al. (2023), MAVEN did not sample the undisturbed solar wind during this orbit, which is atypical for this time period and occurred due to the expansion of the Martian magnetospheric system in response to the disappearing solar wind (an overview of the response of the various plasma boundaries and regions is provided in Halekas et al. (2023)).In our Figure 1, prior to 11:30 MAVEN initially sampled the dawn flank magnetosheath, a region characterized by hot (high temperature) protons spanning energies of a few hundred to ∼1,000 eV (panels c, d, e).Contrary to typical sheath characteristics, the plasma environment was relatively stable: there was very little fluctuation in the magnetic field amplitude and vector (panels a and b), nor the proton energy spectra (panel d).This "calm" plasma environment in the sheath was a result of the low mach number solar wind during the solar wind density minimum: the low mach number significantly raised the proton temperature anisotropy threshold required for sheath proton distributions to become unstable and to produce electromagnetic waves (Halekas et al., 2023).
At 11:30 MAVEN exited the magnetosheath and encountered cold, low energy tenuous ionospheric plasma at an altitude of ∼3,500 km and a solar zenith angle (SZA) close to 90°.This was the upper ionosphere and it was composed of O + and O 2 + (panels d, e, f).At the same time that the upper ionosphere was first encountered the magnetic field magnitude clearly dropped from 6 to 7 nT to <1 nT (panel a), and remained consistently small until just after 12:10.The modeled crustal magnetic field strength (orange line, panel a) was negligible at such high altitudes.The focus of this study is to understand the physics governing this somewhat unique interaction on the inbound orbit segment between the solar wind and upper ionosphere (i.e., the magnetic field amplitude dropping to <1 nT in the upper ionosphere).Similar conditions were observed during the inbound segment of the following orbit (but not shown here).An interesting point to note here is that as MAVEN transitioned from the magnetosheath into the ionosphere it observed signatures that are both consistent and inconsistent with the photoelectron boundary (PEB), described by for example, Xu et al. (2023).Specifically, SWEA observed both the "CO 2 photoelectron peaks" at 22-27 eV, and high energy (greater than 500 eV) electrons, both traveling isotropically relative to the local magnetic field (Figure 1g summarizes this; the detailed energy spectra and pitch angle distributions are not shown here).Our current understanding requires that both ends of the magnetic field connect to both the ionosphere, and solar wind, in order to produce these features.It is not clear how these features arise, and a separate study is investigating this further.Within the tenuous upper ionosphere during the inbound segment, the initial ionospheric plasma density was a few 10s cm −3 and the spacecraft potential was relatively stable at around −0.68 V (the green line in panel e shows the (negative) spacecraft potential measured by STATIC), although the variability in this potential prior to 11:45 is likely instrumental, as discussed in Section 3.3.MAVEN was initially sunlit and crossed the sun-shadow boundary at the orange vertical line; at this point multiple discontinuities were observed by several of the plasma instruments.The spacecraft potential jumped to about −4 V as a direct result of the spacecraft no longer being sunlit and thus no longer emitting a photoelectron current.Significant and sudden enhancements in ion energy flux were observed by STATIC (panels e), thought to be caused through a combination of the jump in spacecraft potential and instrumental field of view effects (see Section 3.3).A corresponding decrease in suprathermal electron energy flux and density was observed by SWEA across the sun-shadow boundary (panel g), a result of the lowest energy suprathermal electrons no longer being able to overcome the larger negative spacecraft potential and reach the instrument (Section 3.2).The jump in spacecraft potential was observed in the LPW I-V sweeps as a shift in the voltage at which the zero-current crossing occurred (the black region of the I-V sweeps in panel h).These discontinuities and their significance on the interpretation of the MAVEN observations are discussed in detail in Sections 3.2 and 3.3.The environment within this tenuous upper ionosphere was collisionless: the neutral density did not rise above the noise threshold of the Neutral Gas and Ion Mass Spectrometer instrument carried by MAVEN (NGIMS, Mahaffy et al., 2015, data not shown here) until about 12:30.The NGIMS neutral density profiles were similar to those measured on the prior orbit; this is not surprising as the neutral density scale height is to first order controlled by solar EUV heating on the dayside, which was constant during this event.
Once MAVEN transitioned across the sun-shadow boundary, dayside ionospheric plasma was still observed until about 12:20, identified by the He-II photoelectron peaks in the SWEA energy spectrum (panel g, enhanced energy flux at ∼22-27 eV, Frahm et al., 2006).This dayside plasma had likely been transported to the nightside and many studies have investigated the processes involved (e.g., Adams et al., 2018;Fox et al., 1993;Girazian et al., 2017).Within this region the magnetic field magnitude returned to values of ∼10 nT, which may have been driven by the presence of crustal magnetic fields, as denoted by the orange line in panel a.MAVEN was sampling the deep nightside ionosphere at this time, at an SZA of ∼170°.MAVEN reached periapsis just before 12:40 (black vertical line) and was still sampling the nightside ionosphere here (SZA ∼125°); the variability in the plasma environment (particularly STATIC and SWEA observations, panels e, f, and g) was typical for the nightside ionosphere (e.g., Fillingim et al., 2007;Lillis et al., 2009).The outbound segment of the orbit sampled the dayside ionosphere, flying close to the sub-solar point as MAVEN encountered the Mars-solar wind interaction region at around 13:30.The various plasma boundaries and regions were significantly expanded, with the dayside ionosphere being observed outside of the typical bowshock location (see Halekas et al., 2023 for more details).

Electron Observations
Detailed observations of the thermal and suprathermal electrons during the inbound segment of the orbit are shown in Figure 2. The time range of this figure focuses on the period when MAVEN exited the magnetosheath and encountered the cold thermal plasma of the extended ionosphere.When the ionosphere was first encountered and the magnetic field amplitude dropped to <1 nT (about 11:35 UTC, panel a), the suprathermal electron density N e rose sharply to 10-12 cm −3 (panel e) while the suprathermal electron temperature T e dropped from 15 to 20 eV to ∼5 eV (panel f), consistent with MAVEN transitioning from the magnetosheath to upper ionosphere.Note that within the upper ionosphere there were two "lines" of suprathermal N e values; the lower one is a result of instrumental effects and should be ignored.Within the ionosphere the low energy (<50 eV) suprathermal electrons were comprised of ionospheric dayside photoelectrons, again identified by the He-II peak at 22-27 eV.A higher energy population >100 eV was also present within the ionosphere that appears to be sheath or solar wind in origin.Detailed analysis of the suprathermal electron populations are beyond the scope of this study and are not discussed further here.
The absolute value of the spacecraft potential as measured by STATIC is shown as the black line in panel c.In this tenuous plasma region there was just enough ambient plasma to drive it slightly negative (−0.68 V) despite the spacecraft being sunlit.The spacecraft potential as measured by STATIC is stable close to the sun-shadow boundary (orange dashed vertical line) but becomes more variable at earlier times, around 11:55 and before.The spacecraft potential is determined via the lowest energy that protons are observed by STATIC (after accounting for spacecraft velocity ram effects and assuming no significant proton flows).The variability in spacecraft potential prior to 11:55 was likely instrumental due to field of view issues and low counting statistics (see Section 3.3).The blue line in panel c shows the relative spacecraft potential as measured by LPW: when one sensor performs an I-V sweep, the other sits in a passive mode and measures the change in relative potential between it and the spacecraft, which is a measure of the relative change in spacecraft potential.LPW did not suffer from field of view issues here and this measure of relative change in the spacecraft potential is more reliable in these tenuous plasma conditions.LPW observed a constant negative spacecraft potential in the sunlit region.
The thermal N e and T e are shown in panels h and i and the black crosses represent the L2 values returned from the standard I-V curve fitting techniques.There are discontinuities in both variables at the sun-shadow boundary and these are instrumental in nature.To first order thermal N e is proportional to the maximum current collected during an I-V sweep and one can see that the maximum current collected was roughly constant across the sun-shadow boundary (orange colors at +40 V in panel g).The discontinuities in reported N e and T e across the boundary are due to the low density plasma that MAVEN is sampling: the instrument and fitting algorithm are designed to operate in the dense, cold ionosphere where the thermal electron population dominates and the plasma is characterized by densities 3-4 orders of magnitude larger, and temperatures 10-20 times lower, than those observed here.When these conditions are not met the LPW L2 fitting algorithm is not always able to obtain reliable results.When this is the case empirical estimates of thermal N e and T e can still be obtained, and some discussion of MAVEN instrument operation modes is prudent before diving into these details.
MAVEN instrument operation modes are defined based on the average locations of the various plasma boundaries and regions at Mars, and the extreme expansion of the ionosphere here is far outside of these typical bounds.When the upper ionosphere was first encountered MAVEN was at such high altitude that the spacecraft and instruments were still operating in high altitude apoapsis modes (to observe hot, energetic low density plasma) rather than ionosphere mode (to observe cold, low energy dense plasma).LPW was thus operating with large ±40 V sweep ranges when it first encountered the ionospheric plasma (y axis on panel g).The changes in LPW operation modes are visible in panel g when the sweep range reduces to ±10 V and then ±5 V at 12:22 and 12:33, respectively.For context, 12:22 UTC is where MAVEN was typically expected to first encounter the ionosphere.The voltage sweep ranges reduce through periapsis primarily to improve the resolution of the I-V sweep in voltage space, which is needed to derive the colder T e encountered in the ionosphere (e.g., Ergun et al., 2015).In addition, smaller sweep ranges reduce the amount that LPW swings the spacecraft potential, which can adversely impact the other plasma instruments on MAVEN, particularly when sampling the cold low energy plasma in the ionosphere.
In this case the large voltage sweep range of LPW was actually beneficial for two reasons: (a) changes in spacecraft potential of a few V across the sun-shadow boundary were negligible compared to the full sweep range and were easily captured; (b) the "electron saturation region" of the I-V curve was still sampled.The electron saturation region is the portion of an I-V sweep where the probe potential is positive and large enough such that the collected current is dominated by electrons.Generally speaking, during plasma conditions when the spacecraft potential and thermal T e are constant, the current measured in the electron saturation region (at constant sweep voltage) is proportional to the thermal N e (see e.g., Ergun et al. (2015) and references therein).As T e increases, the electron saturation region is encountered at larger positive sweep voltages and so larger I-V voltage sweep ranges are required to capture it.The ±40 V sweep range in this event captured the electron saturation region despite T e being large (10,000 K or more compared to ∼400-500 K in the dense ionosphere), allowing us to derive an empirical estimate of N e independently to using the full curve fitting algorithm, which is designed for use in higher density, colder plasma.
The empirical estimate scales the measured maximum currents on the sunlit side (most tenuous plasma density) of the shadow boundary with those measured on the shadow side (larger plasma density and measured current), where N e derived from the full fitting algorithm are reliable.The details and assumptions of this empirical method are described in Section 3.2.1.The results of this empirical derivation are shown as the blue line in Figure 2h, where thermal N e is now continuous across the sun-shadow boundary giving us confidence in this empirical method.Further support for this method is provided by the sporadic agreement with several L2 N e values (black crosses) in the sunlit region, where every so often the fitting algorithm was able to obtain an accurate solution.The resulting thermal N e in this tenuous region of the upper ionosphere were a few 10s cm −3 when first encountered, gradually rising to ∼100 cm −3 at the sun-shadow boundary.The thermal N e was thus about 70%-90% of the total N e , with the remainder contributed by the suprathermal population.This empirical method of deriving thermal N e was only achievable with such high confidence because the electron saturation region is unambiguously sampled in the measured I-V sweeps, and changes in spacecraft potential are small compared to the large ±40 V sweep range.
The thermal T e values (black crosses in panel i) also showed a discontinuity at the sun-shadow boundary which is again instrumental and a result of the low density hot plasma that affects the derivation of thermal N e .The effect of T e on measured I-V curves is more complex than N e and here we have used a simple empirical scaling method to provide estimates of "corrected thermal T e " in the sunlit region.The discontinuity in T e straddled a value of ∼10,000 K and as such, any derived thermal T e (black crosses) during sunlight, that lay below a value of 10,000 K, were scaled by a factor of three, so that the estimated values are roughly continuous across the sun-shadow line.These estimates are shown as the blue line in panel i.While this method is not as rigorous as our estimates for thermal N e , it provides order of magnitude estimates of thermal T e that are appropriate for our purposes, primarily to understand the dominant physics controlling the Mars-solar wind interaction during this event (discussed in Section 5).These empirical methods require several assumptions that are noted in Section 3.2.1 and they can only provide estimates of N e and T e ; full curve fitting is thus preferred when possible to remove reliance on these assumptions.

Empirical Derivation of Thermal Electron Density
The LPW I-V curves analyzed to produce the empirical values of thermal N e (shown in Figure 2h, blue line) are shown in Figure 3.The full I-V curves are shown in panel a, spanning when MAVEN first encountered the tenuous upper ionosphere (dark purple lines) through till just after the sun-shadow boundary (light blue lines).The line colors match the colorbar at the bottom of Figure 2.
The I-V sweeps are generated by placing the sensor (which is immersed in the local plasma) at a known "sweep voltage" and measuring the corresponding electric current collected by it.The sensor voltage is swept through a range of voltage values to produce the I-V curve (LPW I-V curves consist of 128 voltage measurement points).When the instrument operates nominally, a measured I-V curve consists of three main regions: (a) the "ion saturation region", which corresponds to the lowest ∼25% of sweep values (most negative measured currents) and is where the collected current is dominated by ions; (b) the "electron saturation region", which corresponds to the highest ∼25% of sweep values (most positive measured currents) and is where the collected current is dominated by electrons; (c) the "electron temperature region", which corresponds to the "knee" in the I-V curve where the measured current switches from negative to positive, and marks where the current collection transitions between ion and electron dominated.These regions are labeled in Figure 3. Generally speaking, during plasma conditions when the spacecraft potential and thermal T e are constant, the current measured in the electron saturation region within an I-V sweep (at constant sweep voltage) is proportional to the thermal N e (see e.g., Ergun et al. (2015) and references therein).When MAVEN was sampling the tenuous sunlit upper ionosphere both assumptions are reasonable: the change in spacecraft potential of ∼3.5 V across the sun-shadow boundary was small compared to the I-V sweep range of ±40 V, and the thermal T e changed slowly in this region.
The process to empirically calculate N e is as follows.
1.The largest four voltage sweep values from a single I-V sweep were extracted.The last measurement point (highest voltage) was discarded as this can sometimes be contaminated by instrumental effects.2. The average current was calculated from the remaining three measurement points.An average was used to reduce the effects of any spurious data points.3. The average maximum currents calculated in step 2 were linearly scaled to the average maximum current and corresponding average absolute value of L2 thermal N e (black crosses), from a time period of high confidence in the full fitting algorithm.Here this was taken to be between 12:10:00 and 12:12:40, spanning 10 measurement points.The thermal N e was fairly constant over this 10 measurement point period, with the mean thermal N e being 125 cm −3 and the standard deviation being 14 cm −3 .
Panel a in Figure 3 shows that the maximum current at each I-V sweep gradually decreased as one moves backward in time, from the sun-shadow boundary crossing (lightest blue colors) to when the ionospheric thermal plasma was first encountered (dark purple colors).There was no discontinuous jump in maximum current measured as MAVEN crossed the sun-shadow boundary, and this supports the continuous empirical N e presented as the blue line in Figure 2h.
A zoom in of the same I-V sweeps is shown in Figure 3b, focusing on the knee region of the I-V curves where the measured current transitions from ion dominated (negative sweep voltages) to electron dominated (positive voltages).The voltage at which this knee is observed is determined by the spacecraft potential, and relative shifts in the horizontal direction between I-V sweeps represent changes in the spacecraft potential.The figure highlights a clear discontinuity in the voltage location of the knee between darker blue colors ("knee 1") and lighter blue colors ("knee 2"), marking the jump of ∼3.5 V as MAVEN crossed the sun-shadow boundary.There are no other obvious discontinuities and significant changes in the location of the knee while MAVEN samples the ionospheric plasma (dark and light blue colors), signifying that the spacecraft potential otherwise remained approximately constant throughout this time period, supporting the assumptions made during the empirical derivation of thermal N e .At the darkest purple colors in Figure 3b the knee extends down to ∼−20 V; this location is where MAVEN transitioned out of the magnetosheath plasma and was yet to encounter the thermal plasma of the ionosphere; the spacecraft potential was large (10s V) and positive here.

Ion Observations
Detailed ion observations made over the same time period as Figure 2 are shown in Figure 4. Magnetosheath ions are observed at the earliest times (before 11:30) by SWIA and STATIC (panels c, e, and f).The proton energy fluxes drop off just prior to MAVEN encountering the ionosphere; the ionosphere is encountered coincident with the drop in magnetic field strength (panel a).STATIC observations show that the ionospheric composition is initially dominated by O + , with O 2 + becoming dominant at around 11:55 (panels f and g).The heavy ion density is initially about 1 cm −3 in the upper ionosphere, about 40-50 times smaller than the combined thermal and suprathermal electron densities measured by LPW and SWEA (Figures 2e and 2h).Such a large charge imbalance would appear aphysical and it is much more likely that STATIC suffered from field of view issues that prevented it observing the full ion distribution (discussed below).
As MAVEN crossed the sun-shadow boundary the energy flux measured by STATIC abruptly increased (panels e, f) with a corresponding (relatively small) discontinuity in ion density observed (panel g).These observations support the hypothesis that field of view issues prevented STATIC from observing the full ion distribution while MAVEN was sunlit.The observed jump in spacecraft potential to more negative values across the sun-shadow boundary resulted in low energy ionospheric ions being accelerated into STATICs field of view.Further evidence of this is seen in the STATIC direction information shown in panels i and j: sudden enhancements in energy flux were observed across a wide range of angles (both phi and theta) at the sun-shadow boundary, suggesting that these ions were accelerated into STATIC's field of view by the larger (more negative) spacecraft potential.
10.1029/2023JA031910 10 of 23 STATIC was also not operating in periapsis mode at such high altitudes but by chance happened to be pointing in the spacecraft ram direction and so observed the bulk of the thermal ionospheric plasma once MAVEN crossed the sun-shadow line.The ram direction is marked by the green lines in panels i and j.The large swings in the Within the tenuous upper ionosphere at such high altitudes, MAVEN's velocity was about 2 kms −1 which was about half its periapsis velocity.While the ionospheric plasma was cold, MAVEN was still flying subsonically in this region relative to even the heavy O + and O 2 + ions.This means that the major ion species were able to enter STATIC from outside of the ram direction.This was observed, particularly as MAVEN crossed the sun-shadow boundary.At that time the ion energy flux was enhanced in the ram direction, and additional ions also entered the field of view from off ram look directions (panels i and j).These ions (both ram and off-ram) were likely accelerated into the field of view by the more negative spacecraft potential.
The fully derived (black crosses) and empirically derived (blue line) thermal electron and O 2 + ion (green line) temperatures are shown in panel h.At periapsis (right hand side of the figure) the O 2 + T i was equal to or slightly less than the thermal T e , while at higher altitudes on the inbound segment (times between ∼12:27 and ∼12:35) the O 2 + T i was larger than the thermal T e .These trends are consistent with those reported for midnight periapses by Hanley et al. (2021).At times prior to ∼12:27 (corresponding to altitudes greater than about 500 km) O 2 + T i are not derivable from the STATIC data.This is because (a) the tenuous plasma is no longer "beam-like" and enters the STATIC field of view from off-ram directions, but (b) is still low energy (<25 eV).The latter means that an instrumental effect known as "cold ion suppression", which reduces the instrument sensitivity at low energies, is active.The combination of these conditions invalidate the assumptions needed to derive reliable values of thermal O 2 + T i , as described in Hanley et al. (2021).As we will show, accurate knowledge of T i is not crucial for our interpretations and further analysis utilizes assumptions based on the observations, as noted below.

Pressure Balance at the Magnetosheath-Ionosphere Interface
Various pressure terms during the inbound segment of the orbit have been calculated using the MAVEN observations, and these are shown in Figure 5. Magnetic pressure is calculated via and plasma pressure via n s k b T s , where u 0 is the permeability of free space, k b Boltzmann's constant and subscript s is species (electrons or ions).The total plasma pressure is the sum of the suprathermal electron, thermal electron, and thermal ion pressures.
To calculate this value, pressure data are paired closest-in-time to each other, with a maximum time difference of 20 s allowed to take into account the different measurement cadences of SWEA, LPW, and STATIC.
Panel a uses the original L2 LPW N e and T e values (the black crosses in Figures 2h and 2i) to calculate the thermal electron pressure, while panel b uses the empirical values (the blue lines in Figures 2h and 2i).All other pressure terms and line colors are identical between panels.Prior to 11:30 in Figure 5 MAVEN was sampling the magnetosheath and the magnetic pressure was larger than the SWIA plasma pressure (black vs. brown lines; these two terms were equal in value at times earlier in the orbit not covered by the figure).The suprathermal electron plasma pressure measured by SWEA (green line) was about an order of magnitude smaller than the magnetic pressure in the sheath.The SWIA plasma pressure started to decrease at around 11:15 as MAVEN exited the magnetosheath (e.g., Figure 1d).The magnetic pressure abruptly dropped by roughly two orders of magnitude at around 11:35, marking when the magnetic field strength dropped to <1 nT and the upper ionosphere was first encountered.We call the magnetic pressure just prior to this drop the "incident magnetic pressure" (i.e., the pressure incident on the upper ionosphere), which had a value of around 2 × 10 −11 Pa.At the time of the drop there were coincident enhancements in all plasma pressure terms (suprathermal and thermal electron plasma pressures (green and purple lines), and thermal ion pressure (blue line, discussed below)).The suprathermal electron plasma pressure within the ionosphere was about half the incident magnetic pressure.The thermal electron plasma pressure calculated using the original data points (panel a, purple line) was about an order of magnitude smaller than the incident magnetic pressure.
The thermal ion pressure was assumed to be equal to the plasma pressure of thermal O 2 + ions, which was the dominant ion throughout most of the ionosphere (Figure 4g).Because the O 2 + T i was only derivable at lower altitudes during this event (green line, Figure 4h) we made assumptions to extrapolate this to higher altitudes, in particular to the region where MAVEN first encountered the upper ionosphere.We have assumed that T i = 10,000 K throughout the inbound orbit segment, based on Figure 4h: O 2 + T i was derivable after about 12:22 but not at prior times.At this time T i was about 10,000 K; based on the ion energy spectrum in Figure 4e, the ion energy did not increase significantly and so we assumed T i was 10,000 K for the entire inbound segment.While this assumed T i was too large at periapsis (times after 12:30) we are interested primarily in the solar wind-ionosphere interaction region which was encountered at times prior to 12:20, and in particular prior to the sun-shadow boundary crossing (∼12:08).This assumption provides additional context but ultimately does not affect our conclusions because STATIC observed only a fraction of the full ion distribution function while sunlit.The total observed ion pressure was thus typically 1-2 orders of magnitude smaller than the suprathermal and thermal electron pressure terms in this region (Figure 5b, prior to orange vertical dashed line).Our assumed T i may be incorrect by a factor of 2 or 3 and in this case the electron pressures still dominate.When STATIC observed the full ion density immediately after MAVEN crossed the sun-shadow boundary, the ion and thermal electron plasma pressures were about equal and at that time both greater than the suprathermal electron pressure.
The total plasma pressure (sum of the suprathermal electron, thermal electron, and ion pressure terms) is shown as the red lines in Figure 5.In panel a the total plasma pressure in the sunlit ionosphere (between ∼11:40 and 12:05) was about half of the incident magnetic pressure.The abrupt jump in thermal electron plasma pressure (and corresponding total plasma pressure) at the sun-shadow boundary in panel a were a result of the non-physical jump in thermal N e discussed in Section 3.2.
Using the empirical values of thermal N e and T e in the sunlit ionosphere region produced a significantly larger thermal electron plasma pressure almost equal to (and eventually greater than) the suprathermal electron plasma pressure (panel b).In this case the total plasma pressure in the sunlit ionosphere was roughly equal to the incident magnetic pressure.The total plasma pressure was constant across the sun-shadow boundary which is more physical than the instrumental discontinuity observed in panel a.

Evolution of Pressure Terms
To understand the evolution of the Mars-solar wind interaction during the disappearing solar wind time period, the various pressure terms evaluated in Section 3.4 are shown for the inbound segments of five MAVEN orbits throughout this event in Figure 6.We have used the same method described in Section 3.2.1 to derive thermal N e and T e within the tenuous upper ionosphere.As per our analysis above, spacecraft potential effects impede calculations of ion temperature in the upper ionosphere regions and so for simplicity we have assumed that T i = T e /2 throughout these orbits.
This assumption is based on (a) the statistical trends of T e and T i observed at the dawn and dusk terminator (Figure 6 in Hanley et al., 2021) and (b) that within the tenuous upper ionosphere in Figure 4h, T e is about 20,000K compared to our assumed value of T i = 10,000k.As before the thermal and/or suprathermal electron pressure terms typically dominated in the ionosphere and modifying this assumption by factors of 2-3 does not affect our conclusions.
The orbit preceding the minimum in solar wind density is shown in panel a and the four orbits following this are shown in panels b-e.The SZA of MAVEN corresponding to panel a is shown in panel f; MAVENs trajectory is essentially identical on these five consecutive orbits and so the SZA information is accurate for the remaining orbits to within 1° at any given point.Panel b is the orbit analyzed in detail in Figures 1-5.We have performed similar analysis on the other neighboring orbits to ensure correct pressure terms here; while we do not show this analysis we comment on the results below.The horizontal axis shows MAVENs altitude with respect to periapsis, where negative values denote the inbound orbit segment.
During the orbit preceding the minimum in solar wind density (Figure 6a) the Mars-solar wind interaction appears more typical: the ionosphere was not particularly inflated and was first encountered at more nominal inbound altitudes of ∼−1,500 km.The upper ionosphere was identified by the sharp increase (by two orders of magnitude) in thermal ion plasma pressure (blue line) at ∼−1,500 km.Some care must be taken in comparing this set of orbits because the spacecraft trajectory is such that the upper ionosphere is first encountered at an SZA of ∼130° in panel a, quite far from the terminator region of SZA ∼90° in panel b.We can however state with confidence that the ionosphere in panel a did not extend to as high an altitude at the terminator as it does in panel b.As MAVEN first encountered the upper ionosphere in panel a there was a slight decrease in the magnetic pressure, however, the magnetic field strength lay between 2 and 7 nT in this region and did not approach zero.The ionospheric plasma pressure (red line) dominated within the ionosphere below about −1,000 km, indicating a magnetized ionosphere.At about −600 km MAVEN encountered a strong crustal magnetic field and the magnetic pressure clearly dominated.
For the orbit shown in panel b the ionosphere had expanded drastically, and was first encountered at ∼-3,500 km (at an SZA of ∼90°).Within the upper ionosphere the dominant pressure terms were clearly different compared with panel a; in panel b there was a clear switch from magnetic pressure dominance outside of the ionosphere to plasma pressure dominance within, corresponding to the region where the magnetic field amplitude fell to <1 nT.The upper ionosphere was thus unmagnetized and this state was present down to about −1,200 km altitude, where the plasma and magnetic pressures became approximately equal.Below about −800 km altitude the magnetic pressure dominated and the ionosphere became magnetized.
The second orbit following the minimum in solar wind density (panel c) shows similar characteristics to those in panel b: an expanded ionosphere that was first encountered at the terminator at an altitude of about −3,500 km, where a corresponding clear shift in dominant pressure terms occurred (from magnetic outside to plasma within the ionosphere).The solar wind density was still at its minimum during this time period and the observations suggest that the unmagnetized nature of the upper ionosphere during the inbound segment is the steady state configuration of the upper terminator ionosphere during these conditions.Our detailed analysis of this orbit (not shown) confirmed that most of the field of view and instrumental effects described for the orbit in panel b were also present during this orbit, and have been corrected for here.
For the orbits shown in panels d and e the solar wind density had begun to return to more normal values, although it was still somewhat lower than usual (see Figure 7).The ionosphere appears to have shrunk back closer to its normal extent, being first encountered at larger SZA on the nightside at altitudes between −1,000 and −1,500 km (SZA around 155°).In panel d the magnetic field strength still approached zero nT and the plasma pressure still dominated within the upper ionosphere, suggesting that despite the ionosphere shrinking in extent the upper portion was still unmagnetized in nature.In panel e the plasma pressure still dominated within the upper ionosphere but the magnetic field strength was about 2.5 nT, suggesting that the ionosphere was magnetized and that the type of ionosphere-solar wind interaction was shifting back to the prior magnetized case.It may be that the solar wind dynamic pressure had recovered enough that the IMF was able to penetrate across the full dayside ionosphere including the terminator region (e.g., Cravens et al., 2023;Hamil et al., 2022), although without direct observations of the upstream solar wind it is difficult to determine this conclusively.In both panels strong crustal magnetic fields were encountered at periapsis where the magnetic pressure then dominated.

Evolution of the Mars-Solar Wind Interaction Region in Response to Changes in Solar Wind Density
To place the observations and results from Section 3 into context with the bigger picture we also determine how the ionosphere-solar wind interaction system responds to the changes in solar wind density over the disappearing solar wind time period.Our goal here is to provide context for the observed changes in interaction states (magnetized vs. unmagnetized upper ionosphere) and we do not focus on the specific locations of the various Martian plasma boundaries and regions, which are covered in detail by other studies (see Halekas et al., 2023).
The evolution of various parameters spanning 26 MAVEN periapsis passes (labeled orbits 0 to 25, centered on the disappearing solar wind event) are shown in Figure 7.The green line in panel a shows solar wind dynamic pressure measured at Earth by the WIND spacecraft, time shifted to Mars' radial distance from the Sun.The time shifting method used is described in Halekas et al. (2023).Note that the WIND pressure values have not been scaled by the factor  1  2 (where r is the Sun-planet distance) to account for the radial evolution of the solar wind between Earth and Mars.The black line in panel a shows the maximum magnetosheath ion plasma pressure observed by SWIA during the inbound segment of each periapsis pass, calculated using the same method as for Figure 5.We use the maximum plasma pressure because during this time period the plasma boundaries were expanded to such large degree that MAVEN did not sample the full extent of the magnetosheath and so we simply use the largest magnetosheath pressure observed on each inbound segment.While MAVEN may not have sampled the absolute largest plasma pressures within the magnetosheath (due to sampling bias), this method provides results consistent with the rest of our analysis and interpretations.The blue asterisks denote the periapsis passes analyzed in Figure 6, with the second blue asterisk being the orbit analyzed in detail in Figures 1-5.
The drop in solar wind density by over an order of magnitude causes a corresponding similar amplitude drop in solar wind dynamic pressure, shown by the green line in panel a.The solar wind velocity dropped to about 60% of it's pre event value (Halekas et al., 2023), also contributing to the reduction in dynamic pressure.The maximum magnetosheath pressures observed by MAVEN mirror this trend, dropping by nearly two orders of magnitude during the minimum in solar wind density.The smallest magnetosheath pressures are observed coincident with the two orbits immediately following the minimum in solar wind density and dynamic pressure.
The altitude at which the upper ionosphere was first encountered on the inbound segment of each periapsis pass is shown in panel b.This location was identified using a "quick and simple" by eye method: when (a) low energy (<10 eV) thermal plasma density (N e with LPW or N i with STATIC) was consistently above 20 cm −3 and (b) He-II ionospheric photoelectrons were observed in SWEA at 22-27 eV.These criteria were defined to maximize the likelihood that ionospheric thermal plasma was being identified, as opposed to cold ion outflow that is common in the magnetotail region and close to periapsis for this general time range.Cold ion outflow is typically characterized by ion densities below 20 cm −3 and ion energies that can be greater than 10 eV (e.g., Inui et al., 2018Inui et al., , 2019)).These simplified criteria were used because our goal is to identify the overarching trends of how the interaction region responds to the changes in solar wind density, as opposed to a detailed study of the plasma boundaries and regions, that are being carried out by other authors as noted in Halekas et al. (2023).
The altitude extent of the upper ionosphere was anti-correlated with the solar wind dynamic and magnetosheath pressures: as the dynamic and magnetosheath pressures decreased, the ionosphere was encountered at higher altitudes.The ionosphere was encountered at the highest altitudes during the minimum in solar wind dynamic pressure and magnetosheath plasma pressure.The extent of the upper ionosphere also appeared enhanced for orbit number five.Closer inspection of the time series observations for orbit five (not shown here) suggest that this is a particularly dense cold ion outflow event (the ion density was about 100 cm −3 ).The plasma observations (including ion energy and mass spectra, and density) are similar to, for example, Figure 4 in Inui et al. (2018).Another study in preparation details the response of the various plasma boundaries during this time period, utilizing more robust boundary identification techniques.As such we have not attempted to refine our boundary identification method here.

Discussion
Our case study analysis of the orbit immediately following the minimum in solar wind density has highlighted several instrumental caveats that must first be addressed before a complete understanding of the system can be obtained.The greatly expanded nature of the ionosphere during this event means that when it is first encountered by MAVEN the instruments and spacecraft are still operating in high altitude modes that are not necessarily pointed and operating nominally to observe the low energy cold ionosphere.In addition, (a) in this case the upper ionosphere is tenuous (density 10's cm −3 ) and LPW is thus operating in an environment outside of its design specifications; (b) the relatively slow spacecraft apoapsis velocity (just over 2 km s −1 ) means that MAVEN travels subsonically relative to the tenuous thermal ions, leading to field of view complications for STATIC; (c) MAVEN happens to cross the sun-shadow boundary while within the tenuous upper ionosphere, leading to a large (3.5 V) and sudden jump in spacecraft potential that influences whether low energy ions and electrons can reach STATIC and SWEA, subsequently causing discontinuities in their derived plasma densities.

Instrumental Effects
LPW and the L2 fitting algorithm are designed to measure cold, high density plasma, and the instrument was thus operating at the limit of its capability in the tenuous conditions encountered in the upper ionosphere.As a result the fitting algorithm struggled to consistently derive accurate values of N e and T e there.The reported L2 values (black crosses in Figures 2h and 2i) were typically low by a factor of 4-5 and ∼3, respectively when compared to our empirical derivations (blue lines Figures 2h and 2i).The empirical derivation of thermal N e in particular was possible because LPW was still operating in high altitude mode when the ionosphere was first encountered.The I-V sweeps spanned a range of ±40 V, which is large compared to both the changes in spacecraft potential (including the ∼3.5 V jump across the sun-shadow boundary) and the temperature of the thermal electron population (about 10,000 K or an eV).In contrast, the jump in spacecraft potential across the sun-shadow boundary did impact the SWEA measurements of suprathermal electrons: the jump from −0.68 V to −4 V meant that low energy electrons could no longer overcome this potential and reach SWEA; there was a corresponding discontinuity and drop in measured suprathermal N e across the boundary.Within the tenuous and sunlit upper ionosphere the suprathermal electrons (densities 10-12 cm −3 ) made up a significant fraction of the total electron population, about 20% by density.Their large T e mean that they contributed to about half of the total electron pressure (with thermal electrons contributing the other half).
STATIC suffered from field of view effects because the spacecraft velocity was slower near apoapsis and during this time period MAVEN was moving subsonically relative to the tenuous ions (including heavier O + and O 2 + ).Instead of observing ions in the ram direction only as happens at periapsis when MAVEN travels supersonically, ions could enter STATIC from any direction, which may fall outside the instrument field of view (360° × 90°).This effect was most obvious when comparing derived total N i (∼1 cm −3 ) with total electron N e (∼50-100 cm −3 ) in the sunlit region.Such a large disparity between ion and electron densities is aphysical, and STATIC was almost certainly not observing the full ion distribution.This interpretation is supported by STATIC observations as MAVEN crossed the sun-shadow boundary within the tenuous upper ionosphere: the spacecraft potential jumped from −0.68 V to −4 V and low energy ions were thus accelerated toward the spacecraft and entered STATIC's field of view.There was a significant discontinuous enhancement of measured ion energy flux and corresponding derived N i across the sun-shadow boundary (Figures 4g-4i and 4j), which included ions suddenly entering STATIC's field of view from off-ram directions where none were prior to the crossing.These field of view issues and corresponding low count rates also cause problems when trying to calculate the spacecraft potential using STATIC observations.The low count rates in this region constitute low signal to noise and lead to instrumental variability in the STATIC derived spacecraft potential (black line in Figure 4d, prior to 11:55).Measurements of the relative change in the spacecraft potential made by LPW were more reliable in these tenuous plasma conditions and measured a constant spacecraft potential throughout the tenuous upper ionosphere, until the sun-shadow boundary was crossed (Figure 4d blue line).Note that this measurement by LPW provides an accurate measurement of the relative change in spacecraft potential when variability is relatively small (<a few V), but accuracy decreases across larger discontinuous jumps such as that observed at the sun-shadow boundary, and when the plasma temperature varies quickly.

Mars-Solar Wind Interaction
The total plasma pressure within the tenuous upper ionosphere calculated using original LPW L2 values of N e and T e was about half that of the incident magnetic pressure just outside of the ionosphere (red and black lines in Figure 5a).Such a situation suggests that the system was not in pressure equilibrium (i.e., dynamic in nature) and this may be acceptable, except that a similar result is also obtained on the following orbit, ∼3.5 hr later (Figures 6b and 6c).One would expect the system to come to equilibrium over such a long period of time (compared to, e.g., plasma transport timescales in the magnetosphere and ionosphere) and such an interpretation is difficult to support.
When the empirical values for thermal N e and T e are used, the total plasma pressure balances the incident magnetic pressure at the interaction region (red and black lines in Figure 5b).This is true for the following orbit as well, suggesting that the high altitude ionosphere near the terminator was in an unmagnetized state during this time period: the ionospheric plasma pressure was large enough to withstand the incident magnetic pressure and prevent the solar wind magnetic field from penetrating into it (leading to the observed values of magnetic field strength consistently <1 nT).We note here that generally speaking, plasma boundaries in a state of equilibrium should be under force balance but not necessarily pressure balance.The magnetic tension and gradient of magnetic pressure forces are two candidate contributors at this particular boundary.At the location of MAVEN when it encounters this boundary-the high altitude terminator-we expect the solar wind that has draped around the planet to be almost straight such that curvature and tension forces are negligible.In this case, the gradient in magnetic pressure would dominate, supporting our interpretations.Further discussion of this topic is given in Section 5.3.
The location of MAVEN during these observations is important in correctly interpreting the observations: previous studies have investigated pressure balance across the various Martian plasma boundaries and regions, and generally speaking have shown that for times when the Martian ionosphere is in an unmagnetized state, pressure balance occurs across the upstream solar wind dynamic pressure, the magnetosheath plasma pressure, the magnetic pressure within the magnetic pileup region, and the plasma pressure within the ionosphere (e.g., Holmberg et al., 2019;Lentz et al., 2021); Sánchez-Cano et al., 2020 These studies tend to focus on the low SZA dayside ionosphere where the dynamic pressure acts toward the planet.During this disappearing solar wind event MAVEN encountered the tenuous upper ionosphere at the terminator and so the solar wind dynamic pressure does not point Marsward.Our interpretations are that in the flanks of the magnetosheath, the magnetosheath plasma pressure (assumed to act roughly isotropically) was the primary Marsward pressure force acting against the ionosphere.As the magnetosheath pressure dropped off closer to the planet, the magnetic pressure associated with magnetic pileup became the dominant pressure term, impinging on the upper ionosphere.The ionospheric plasma pressure was large enough to withstand this incident pressure (at the terminator), preventing the IMF from penetrating into the ionosphere and leading to the unmagnetized state observed there.
An important characteristic of this unmagnetized state is that it appears to be somewhat localized in nature occurring on the inbound segment where MAVEN sampled the terminator region, but not on the outbound segment where MAVEN crossed close to the subsolar point.On the outbound segment the magnetic field strength was 10-15 nT within the upper ionosphere suggesting a magnetized state (Figure 1).This property distinguishes our case study from previous well known examples of unmagnetized ionospheres such as those reported at Venus.At Venus during solar maximum, the ionospheric plasma pressure is large enough to stand off the solar wind dynamic pressure and prevent the IMF penetrating into the ionosphere across the entire dayside ionosphere (e.g., Luhmann et al., 1980;Luhmann & Cravens, 1991).Specifically, on average the ionospheric plasma pressure is sufficient to hold off the solar wind at 400 km altitude at the subsolar point and 1000 km at the terminator (Russell et al., 2006).The change in altitude is driven primarily by the fact that the solar wind dynamic pressure points less Venusward at higher SZA.Transient enhancements in the solar wind dynamic pressure can temporarily depress the altitudes at which this pressure balance occurs (Elphic et al., 1980).Thus, an unmagnetized Venus ionosphere is global in nature spanning the entire dayside and "flaring" outward to higher altitudes at larger values of SZA, in response to the reduced Venus-directed solar wind dynamic pressure at these larger SZA.The underlying Venusian ionosphere determines the magnetization state, which in turn depends on the solar cycle phase and whether there is sufficient solar ionizing flux to produce a dense enough ionosphere (e.g., Russell et al., 2006).
In contrast, our case study shows that the Mars ionosphere was magnetized at the subsolar point and unmagnetized at the terminator (i.e., the magnetization state varied across the dayside).An important aspect when comparing to Venus is that the primary sources of the Martian ionosphere did not change significantly during the disappearing solar wind event: to first order the underlying neutral atmosphere and photoionizing EUV flux remained constant throughout this time period (i.e., changes related to solar cycle variability were negligible over the course of a few days), and so the outward pressure from the ionosphere was constant.The decrease in solar wind density and associated dynamic pressure meant that the solar wind forcing from above had however decreased significantly: while the solar wind IMF was still able to penetrate into the subsolar ionosphere this was no longer the case at the terminator region, due to the reduced Mars-ward directed dynamic pressure at such large SZA.In essence, such drastic changes in the solar wind forcing (a reduction in dynamic pressure of over an order of magnitude during the course of ∼ a day) led to the observed state change in the Mars-solar wind interaction.An alternative explanation for this event is that the unmagnetized nature of the high altitude terminator ionosphere is driven by rotations in the upstream solar wind IMF.Because the same magnetic signatures were observed at the same high altitude terminator location on two consecutive MAVEN orbits, this explanation would require two near identical rotations in the IMF to occur when MAVEN was sampling the same location on those two orbits.We deem this unlikely and favor the pressure balance argument.
One further noticeable difference for our case study at Mars compared to Venus is that the suprathermal electrons within the upper ionosphere appeared to play an important role, contributing a roughly equal amount to the total plasma pressure as the thermal electrons.In contrast at Venus, sharp gradients in thermal N e are commonly observed at the dayside ionosphere-solar wind interface (a feature known as an ionopause, e.g., Elphic et al., 1980), and the thermal electron plasma pressure is typically sufficient to stand off the solar wind.For the tenuous and expanded Martian ionosphere the hot suprathermal electrons were important in determining the interaction state with the solar wind.

Estimates of Magnetic Curvature-Versus Gradient in Magnetic Pressure-Force Terms
Thus far we have argued that during the inbound orbit segment the upper ionospheric plasma pressure was able to balance the impinging magnetic pressure, to produce the features observed by MAVEN.Under equilibrium one expects force balance, however, this does not necessarily equate to pressure balance.The Lorentz force, and more specifically the J × B force, will act at the boundary (or current layer) between the magnetosheath and upper ionosphere, where the gradient in magnetic field is observed.The J × B force is comprised of two terms: the curvature in magnetic field, and the gradient in magnetic pressure (Equation 1): where the first term on the right hand side represents the magnetic curvature force and the second term the gradient in magnetic pressure (and μ 0 is the permeability of free space).At the location when MAVEN crossed the boundary (high altitude at the terminator), intuitively one expects little curvature in the global draped magnetic field and thus the curvature force to be small.MAVEN observations at Mars support this (e.g., Figure 3 in Connerney, Espley, DiBraccio, et al., 2015).A separate study by coauthor Ma of this manuscript investigates the global magnetic field configuration during this event, via global magnetohydrodynamic (MHD) simulations, and those results also support this assumption.
A quantitative estimate to support this assumption was also made using the MAVEN observations: Minimum Variance Analysis (MVA, using the methods described in Sonnerup and Scheible (1998)) was performed across the boundary transition region, enclosed by dash-dot vertical lines in Figure 2, between 11:32:00 and 11:37:14.The MVA results are shown in Figure 8, which include the minimum (B 1 ), intermediate (B 2 ) and maximum (B 3 ) variance vectors, the magnitudes of which are shown in panels a-c.Hodograms of these components (panels e and f) show a ∼180° rotation in the field, as expected at the current layer.The measured magnetic field across the transition was rotated from the MSO coordinate system into the MVA frame, shown in panels AA and BB, respectively.The minimum variance direction is assumed to be normal to the boundary: the average values of this vector before and after the boundary were small, about −0.06 and 0.44 nT, respectively.
An estimate of the current density within this layer was made using a method similar to that in, for example, Goetz et al. (2016), via Ampères Law, Equation 2: which can be approximated by Equation 3: where B t is the discontinuity in the magnetic field tangential to the boundary and L is the characteristic length scale of the boundary crossing.In the MVA frame the vector tangential to the boundary is represented by the maximum variance vector, B 3 here.The change in B 3 across the boundary, B t , was about 6 nT.An estimate of L was obtained by assuming that the magnetic field vector in MSO coordinates prior to the transition represented the orientation of the boundary.The angle that the MAVEN spacecraft traveled relative to this was calculated; in combination with MAVENs orbit velocity across the transition (∼2.1 kms −1 ) and the time it took for MAVEN to cross the boundary (about 5 min), the boundary thickness L was estimated to be about 350 km.Use of Equation 3 provided an estimate of the current density running along the boundary, as approximately 0.014 μAm 2 , which is comparable to previous estimates of the current densities that comprise the current systems within the Martian magnetosphere (e.g., Ramstad et al., 2020).
The total J × B force was estimated by multiplying this current density (0.014 μAm 2 ) by the magnetic field strength prior to crossing the boundary (∼6 nT), to give an estimate of |J| ∼ 9 × 10 −17 Nm −3 .From Figure 5a the magnetic pressure prior to the transition is about 2 × 10 −11 Pa; dividing this by L gives an estimate of the gradient in magnetic pressure to be about 6 × 10 −17 Nm −3 .Thus, the magnetic curvature term is estimated to be about 3 × 10 −17 Nm −3 , about half that of the gradient in magnetic pressure term.Our assumption that the gradient in magnetic pressure dominates over the magnetic curvature force at this boundary is thus supported by these results.

Evolution of the Mars-Solar Wind Interaction
Our comparison of the various pressure terms demonstrates that the state of the Mars-solar wind interaction changed during the disappearing solar wind event.During the orbit prior to the minimum in solar wind density (Figure 5a) the dynamic pressure was large enough to overcome the ionospheric pressure, even at large SZA, and the ionosphere was encountered at low altitudes and in a magnetized state (i.e., the magnetic field amplitude is 2-7 nT in the upper ionosphere).The change in interaction state immediately following the minimum in solar wind density was clear (panels b and c): not only was the ionosphere greatly expanded but it was in an unmagnetized state where the magnetic field strength within the upper ionosphere was consistently <1 nT and the plasma pressure dominated.As the solar wind density recovered the interaction started to revert back to its prior state: panel d shows that while the ionosphere still appeared unmagnetized, its extent had greatly shrunk, and by the following orbit (panel e) the ionosphere was magnetized.
The Mars-solar wind interaction can change states over short timescales in response to sudden and significant changes in the upstream solar wind dynamic pressure.Figures 6 and 7 demonstrate that the Mars system responded on timescales of less than a MAVEN orbit (∼3.5 hr), with the system coming to a new state of equilibrium (unmagnetized ionosphere at the terminator) over three MAVEN orbits, before returning to its pre-event magnetized state once the solar wind density recovered.

Conclusions
On 26 December 2022 the solar wind density dropped by over an order of magnitude, observed at both Earth and Mars, and an overview of this "disappearing solar wind" event is given by Halekas et al. (2023).We have analyzed the time period surrounding this density minimum using MAVEN observations at Mars, with a particular focus on the orbit immediately following this event and the goal of understanding how such sudden and extreme changes in the solar wind affect the Mars-solar wind interaction.Our study is motivated in particular by observations made during the two orbits immediately following the density minimum, where the magnetic field strength fell to values consistently <1 nT within the greatly expanded upper ionosphere during the inbound orbit segments, which were located over the dusk terminator region north of the equator.
Our conclusions are summarized as follows: 1. Prior to the minimum in solar wind density the Martian ionosphere was in a magnetized state.On the MAVEN inbound orbit segment immediately following this minimum in solar wind density, the terminator upper ionosphere had transitioned to an unmagnetized state: the thermal plasma pressure within the tenuous upper ionosphere was large enough to prevent the solar wind IMF from penetrating into it, resulting in the observed magnetic field strength consistently <1 nT.This state transition appears localized to the terminator region where the solar wind dynamic pressure does not point planetward: the dynamic pressure was still large enough such that during the outbound segment of this orbit that sampled the dayside subsolar ionosphere, the ionosphere remained in a magnetized state.2. The terminator upper ionosphere remained in this expanded unmagnetized state for two consecutive orbits spanning the duration of the minimum in solar wind density.Following the minimum, as the solar wind density began recovering, the ionosphere shrank in extent but remained unmagnetized, before returning to a magnetized state on the fourth orbit after the density minimum.3. The underlying dayside neutral atmosphere and ionization sources were to first order unchanged over the course of a few hours between MAVEN orbits, and such large changes in the upstream solar wind density thus drive the changes in the Mars-ionosphere interaction state (magnetized to unmagnetized).The Mars-solar wind interaction responds to changes in the upstream solar wind density on timescales of less than a MAVEN orbit (∼3.5 hr). 4. While some aspects of this interaction appear analogous to when the Venus ionosphere is in an unmagnetized state, there are several important differences.The entire Venus dayside ionosphere is typically unmagnetized during solar maximum as a result of increased photoionization and enhanced thermal plasma pressure.In contrast, this case study demonstrates that only localized portions of the ionosphere (at the terminator) became unmagnetized, driven by changes in the dynamic pressure of the upstream solar wind (the "external" pressure force) rather than internal changes within the ionosphere (the "internal" pressure force).5.When MAVEN first encountered the greatly expanded ionosphere the spacecraft and several plasma instruments were still operating in high altitude observing modes, as opposed to low altitude ionospheric modes.
Several instrumental caveats arose related to the tenuous plasma environment, field of view issues, and sudden and significant changes in the spacecraft potential as MAVEN crossed the sun-shadow boundary.These caveats have been addressed and we demonstrate that they must be accounted for before the observations can be correctly interpreted: if the reported L2 data are used "as is" then incorrect conclusions are drawn.

Figure 1 .
Figure 1.Overview of MAVEN's orbit immediately following the minimum in solar wind density.(a): magnetic field magnitude (black) with the Morschhauser spherical harmonic crustal magnetic field model (Morschhauser et al., 2014) (orange); (b): magnetic field vector; (c): angle the magnetic field makes relative to the local zenith direction; (d): SWIA omni-directional ion energy flux; (e), (f): STATIC omni-directional ion energy flux and mass spectra.The green line in panel (e) marks the (negative of) spacecraft potential as measured by STATIC; (g): SWEA omni-directional suprathermal electron energy spectrum; (h): LPW I-V sweeps.The colorbar underneath panel h marks ellapsed time and maps to the colored orbit trajectories in panels O1-O3.The orange dashed vertical line marks the sun-shadow transition boundary (MAVEN is sunlit prior to and in darkness after the boundary).The black vertical line, and black crosses in O1-O3, mark periapsis.

Figure 2 .
Figure 2. MAVEN observations focusing on thermal and suprathermal electrons for the inbound orbit segment immediately following the minimum in solar wind density.(a): magnetic field magnitude (black) with the Morschhauser spherical harmonic crustal magnetic field model (Morschhauser et al., 2014) (orange); (b): magnetic field vector; (c) absolute spacecraft potential measured by STATIC (black) and relative spacecraft potential measured by LPW (blue); (d): SWEA omni-directional suprathermal electron energy spectrum; (e), (f) SWEA suprathermal electron density and temperature; (g): LPW I-V sweeps; (h), (i): LPW thermal electron density and temperature (black crosses) with instrumental corrections applied (blue lines).The colored time bar beneath panel (i) matches to Figure 1.The orange dashed vertical line marks the sun-shadow transition boundary; the vertical black dash-dot lines enclose the region over which MVA is undertaken (see Section 5).

Figure 3 .
Figure 3. Individual I-V sweeps measured by LPW when sampling the upper ionosphere and crossing the sun-shadow boundary.(a) shows the full I-V sweeps; (b) shows a zoom in on the knee region of the I-V curves.I-V sweeps span the time range 11:22:40 to 12:12:40, denoted by the colors that map to the timebar in Figure 2.

Figure 4 .
Figure 4. MAVEN observations focusing on ions for the inbound orbit segment immediately following the minimum in solar wind density.(a) magnetic field magnitude (black) with the Morschhauser spherical harmonic crustal magnetic field model (Morschhauser et al., 2014) (orange); (b): magnetic field vector; (c) SWIA omni-directional ion energy flux; (d) absolute spacecraft potential measured by STATIC (black) and relative spacecraft potential measured by LPW (blue); (e), (f): STATIC omni-directional ion energy flux and mass spectra; (g) STATIC L3 ion densities; (h): LPW original and corrected T e and STATIC O 2 + T i ; (i), (j): STATIC direction information in the toroidal and poloidal dimensions, respectively; the green lines mark the ram direction in the STATIC instrument frame.The timebar underneath (i) matches Figure 1.The orange dashed vertical line marks the sun-shadow transition boundary.

Figure 5 .
Figure 5. Various pressure terms calculated using MAVEN observations for the inbound orbit segment immediately following the minimum in solar wind density.(a) Original L2 LPW N e and T e are used in the calculation of thermal electron pressure.(b) Empirical LPW N e and empirically corrected T e are used in the calculation of thermal electron pressure.The timebar matches to Figure 1.MAG = magnetic pressure calculated via   2 2 0; STA, SWE, SWI and LPW = STATIC, SWEA, SWIA and LPW plasma pressure calculated via n s k b T s , where subscript s denotes ions or electrons; CF = magnetic pressure from the crustal magnetic field model; SWE + LPW + STA is the total plasma pressure, excluding sheath pressure from SWIA.Note, SWIA pressure is not shown when it falls below 5 × 10 −12 Pa, which is the background floor of the instrument due to galactic cosmic rays and radioactivity of the micro channel plates within the instrument.

Figure 6 .
Figure 6.Evolution of pressure terms for inbound orbit segments during the disappearing solar wind event.(a) the orbit proceeding the minimum in solar wind density; (b) the orbit immediately following and analyzed in this study; (c)-(e) the three subsequent orbits.(f) SZA of MAVEN.Line colors follow the same legend as Figure 5. Negative altitudes along the X axis mark that this is the inbound segment of the orbit.

Figure 8 .
Figure 8. Minimum Variance Analysis results across the transition region between the magnetosheath and upper ionosphere, during the inbound orbit segment from the orbit studied in detail in Figures 1-5.Panels show: A-C: the minimum (B 1 ), intermediate (B 2 ) and maximum (B 3 ) components; D: the magnetic field magnitude; E-F: hodograms resulting from the MVA.Text shows the eigenvalues and their ratios (e 1-3 ); average values of the magnetic field in each component (<B 1-3 >) and the minimum variance vectors (B 1-3 ).Panels AA and BB show the observed magnetic field in MSO coordinates, and rotated into the MVA frame, respectively.