A Mars’ volatile and climate history
Knowing that Mars had a substantial global magnetic field billions of years ago, the atmosphere may have been protected against the solar wind for a geologically significant period. In the dynamo era, Mars may have retained a warm and dense atmosphere. When the dynamo stopped, there was no more protection against the solar wind and the planet was permanently subjected to atmospheric erosion.
Atmospheric loss may have contributed significantly to the evolution of the Martian climate over time. At the first times of the Martian history, the main mechanisms of atmospheric erosion were [Brain and Jakosky, 1998]:
Catastrophic degassing by meteoritic bombing: impact erosion.Hydrodynamic escaping: it occurs when a light species escapes in abundance to space (usually enabled by high solar EUV flux or another form of heating) and drags heavier species along with it through collisions.
Absorption and storage in the regolith of the surface and sub-surface.
Escape into space: it is the result of a set of physical processes that provide particles of the thermosphere, ionosphere and exosphere with sufficient energy to escape the planet.
Hydrodynamic escape should have been significant for Mars during the first few hundred million years after its formation [Zahnle and Kasting, 1986]. When it ceased, impact erosion was a dominant loss mechanism, then decreasing as the impact flux declined over time [Melosh and Vickery, 1989]. The impact of these two escape processes on the atmosphere and climate of Mars are reviewed by Jakosky and Phillips, . The two remaining loss processes were then exchange with the surface (as polar caps) and sub-surface (as carbonate deposits within the crust) and escape to space. As the exchange with the regolith is largely reversible, escape to space permanently removes particles from the atmosphere. Escape to space is likely to have been the dominant permanent loss process over the last 3.8 billion years, and therefore a main contributor in the change of climate inferred to have occurred between present and 3.5-4.0 billion years ago.
Escape to space removes both neutral and ionized species from three atmospheric reservoirs: the thermosphere, the exosphere and the ionosphere (previously described in section 184.108.40.206). Most of the escape occurs between the homopause and the exobase, typically between 190 and 210 km [Krasnopolsky et al., 1993]. A neutral or an ion escapes from Mars when it acquired the required escape velocity. The escape velocity is a ratio between the gravitational and electric potential energy of a given particle and the kinetic energy required to overcome it: is the universal gravitational constant, the mass of the planet, the elementary charge, the number of electrical charges, the electrostatic potential difference in the case of charged particles, the mass of the particle and the radial distance of the particle from the center of the planet.
Figure 14. Schematic of contemporary escape processes relevant for atmospheric neutrals and ions. The escape processes are in the rectangular boxes and presented in a ‘decision tree’. Solar inputs shown at the lower right contribute to the energization of atmospheric particles [Brain et al., 2017].
The different escape processes for atmospheric neutrals and ions are gathered in Figure 14. Three main escape processes have been identified for neutrals:
Jeans (or thermal) escape: a portion of the thermal distribution for an atmospheric species exceeds the energy necessary for escape. Only species with small mass (H, D and He) can escape significantly via this mechanism. Photodissociation of water combined to Jeans escape of H could explain the disappearance of water from the Martian atmosphere [Pierrard, 2003].
Photochemical escape: exospheric chemical reactions (such as dissociative recombination of an ionized molecule with a nearby electron, resulting in two fast neutral atoms) provide atmospheric species with sufficient velocity to escape. Photochemical loss can be significant for O, N and C.
Atmospheric sputtering: energetic incident particles (including ionospheric particles accelerated by the solar wind) can collide with particles of the thermosphere and ionosphere near the exobase. The momentum and energy exchange between particles carried out during the collision can lead to the ejection to space of the target particle. The incident particles can be solar wind protons or pickup ions.
When an atmospheric particle is ionized above the exobase, it can be accelerated, usually antisunward, by ambient electric fields. Some of the ions return to the atmosphere by precipitation while a certain amount is expected to escape. The main ion escape processes are:
Ion pickup: an ionized neutral particle is accelerated away from the planet by a motional electric field induced by the solar wind plasma (having a mean velocity of v) and its IMF B [Luhmann et al., 1991]: = − × . It occurs primarily for ionized exospheric neutrals.
Ion outflow: acceleration of low energy particles out of the ionosphere via plasma heating and outward directing charge separation electric field.
Bulk escape: any process for which coherent portions of the ionosphere are detached via magnetic and/or velocity shear processes and accelerated away from the planet.
MAVEN data enable the identification of three escaping planetary ion populations [Dong et al., 2015a]: strong plume fluxes over the MSE (Mars-Sun-Electric field) North Polar Region, strong antisunward ion fluxes in the tail region, and weak but energetic upstream pickup ion fluxes observed mostly on the dayside. The presence of such plumes has been conjectured before MAVEN, but previous measurements did not enable to set the permanent nature of this feature. The study of Dong et al. [2015a] illustrates the permanent presence of a substantial plume with strong ion fluxes widely distributed in the MSE northern hemisphere above the Magnetic Pile-up Boundary (see section 220.127.116.11), which gradually turn into tailward fluxes on the nightside without any significant boundary, as observed in Figure 15. In this figure are plotted the O+ fluxes and velocities for energies greater than 25 eV, projected in the MSE X-Z plane. At the time of this paper, the plume escape for O+ was estimated at 30% of the tailward escape, a number later revised at 50% when more data were available. This escape varies with the solar wind density, dynamic pressure and EUV fluxes. It is highest for low EUV flux and high solar wind pressure.
Figure 15. Observation of the plume fluxes over the MSE North Polar region [adapted from Dong et al., 2015a]. +fluxes and velocities for energies greater than 25 eV are projected on the MSE X-Z plane. The arrows show flux and velocity directions, while the colors label the magnitudes. The two grey dotted lines set the approximate position of the bow shock and of the Magnetic Pile-up Boundary.
Data recorded by the numerous spacecraft orbiting around Mars enable us to better understand the physical processes that drive escape today. This knowledge then allows us to extrapolate to earlier conditions, when the drivers were enhanced and escape was more rapid.
Venus, in the same manner of Mars, presents no internal magnetic field. The atmospheres of the two planets are nevertheless strongly different: the atmospheric pressure at the surface of Venus is ~90 bar while it is ~0.006 bar at the surface of Mars. As Mars, Venus is permanently subjected to atmospheric erosion. However, due to the mass of the planet, the escape velocity on Venus is about twice the one of Mars, making it more difficult for particles to escape [Pierrard, 2003]. Moreover, Venus has developed through its history a greenhouse effect far more import than those on Earth and on Mars. These different specificities are part of the explanation why the planet has managed to keep its dense atmosphere despite the erosion by the solar wind.
The interaction of the solar wind with Mars
The exact nature of the interaction of the solar wind with Mars has only been elucidated around twenty years ago when MGS revealed the absence of intrinsic magnetic field at Mars but also highlighted the presence of crustal magnetic field sources on its surface. Before MGS, the bow shock and the magnetosheath had already been explored by Mariner 4, Mars 2, 3 and 5 in the 70s while Phobos 2 observed the magnetotail in the 80s (see Figure 26 and Table 2).
Two models for the Martian interaction with the solar wind were opposed at that time. On one hand, a model of magnetospheric interaction, proposed by Dolginov et al.,  and Gringauz et al., , which claimed that the magnetic field of Mars controls the interaction of the solar wind with the planet. Their arguments were that they found more similitudes in the interaction of the solar wind with the Earth and Mars than between Mars and Venus, which was already known for not having any intrinsic magnetic field. Those conclusions were opposed to those of Vaisberg et al.,  and Russell et al.,  who proposed that the interaction of the solar wind with Mars is controlled by the Martian atmosphere, like the unmagnetized planet Venus. The different arguments of both camps are summarized by Vaisberg .
The MGS magnetic measurements finally supported the proposition that the solar wind interacts with the extended atmosphere and ionosphere of Mars, creating a magnetospheric cavity, but also with the small-scale crustal magnetic fields. This interaction hence contrasts with the plasma interactions at any other solar system bodies. The obstacle that faces the solar wind is a combination of a global atmospheric obstacle (like for Venus or comets), punctuated by many smaller-scale obstacles formed by strong crustal magnetic fields (somehow similar to the Earth or the Moon to a lesser extend).
The main features of the Martian global plasma interaction with the solar wind are summarized in Figure 16 [Brain et al., 2017]. Solar wind particles, coming from the left side, and the associated IMF, represented in yellow, induce magnetic fields in the conducting Martian ionosphere. This induced magnetosphere-like interaction forms a variety of plasma regimes and boundaries that can be distinguished using particle and field measurements obtained by the spacecraft orbiting around the planet.
Figure 16. Schematic of the Martian plasma interaction regions [Brain et al., 2017].
The solar wind carries with it the IMF (yellow) as it streams (dashed lines) toward the bow shock (green) upstream of Mars. The ionosphere (orange) is delimited by the ionopause (outer limit of the orange area).
In order to identify the different regions and boundaries of the Martian environment from plasma instruments observations, the data recorded by several instruments of the particle and field package of the MAVEN spacecraft during one orbit in October 2016 are plotted in Figure 17 (see section 2.2.3 for more details about MAVEN instruments). During this orbit, the spacecraft entered in the Martian induced magnetosphere on the dayside southern hemisphere, then traveled toward the northern hemisphere and the nightside it reached at ~17:15 UT. On the first panel is plotted the energy-time spectrogram of omnidirectional electron energy flux (JE) measured by SWEA. On the second and third panel are plotted the energy-time and mass-time spectrogram of omnidirectional ion energy flux (JE), respectively, measured by STATIC. On the fourth panel are the electron (red) and ion (black) density measured by SWEA and SWIA, respectively. Note that the density calculated by SWIA is not trustworthy below the magnetosheath as it does not record ions with energy below a dozen of eV. On the fifth panel is plotted the in-situ magnitude of magnetic field (black) measured by MAG, superimposed with the calculated magnitude of the crustal magnetic field (red), from the model of Morschhauser et al.  (see section 2.6 for more details). Finally, the altitude and the position of the spacecraft regarding Mars are plotted on panel 6 and 7 (see section 2.5 for more details about the frames). The approximate location of the different boundaries discussed in next subsections are represented by the brown vertical lines.
In the subsequent sections we first describe the solar wind interaction with Mars as a steady-state interaction (section 1.3.1), using together Figure 16 and Figure 17. We then discuss time-dependent effects modifying the location of the different regions of the Martian environment (section 1.3.2). We finish on a focus on the nightside ionosphere (section 1.3.3), which is the main region that has been studied throughout my PhD.
Figure 17. Example of a MAVEN passage in the plasma environment of Mars, with a periapsis on the dayside. Panel 1: SWEA energy-time spectrogram of omnidirectional electron energy flux. Panel 2: STATIC energy-time spectrogram of omnidirectional ion energy flux (C0 mode). Panel 3: STATIC mass-time spectrogram of omnidirectional ion energy flux (C6 mode). Panel 4: Electron density measured by SWEA (red) and ion density measured by SWIA (black). Panel 5: Magnetic field intensity (measured by MAG in black and calculated from the model of Morschhauser et al.  in red) versus time. Panel 6: Altitude versus time. Panel 7: Coordinates of the spacecraft in the MSO frame (see section 2.5). The vertical lines highlight the main boundaries of the Martian environment described in section 1.3.1.
Models are also important tools for placing spacecraft measurements in context, and for probing causes and effects in the physics underlying the plasma interaction. Most of the models employ either magneto-hydrodynamic (MHD) or hybrid assumptions. MHD models implement equations for the motion of electrically conducting fluids subject to electromagnetic forces, while hybrid models strive to include ion kinetic effects by considering the numerical particle motion (still treating the electrons as a neutralizing massless fluid), where each macro-particle represents a cloud of physical ions. The MHD models are particularly well suited for reproducing characteristics of the interaction with the solar wind, while the hybrid models may better describe kinetic effects. Comparison of several models have been made by Brain et al., , Kallio et al., , and Ledvina et al., . Some individual model results are discussed throughout this section, where appropriate.
The steady-state interaction
The characteristics of the solar wind interactions with weakly magnetized, or unmagnetized bodies, are in some regards similar to the flow around a magnetized planet [Luhmann et al., 1992, Brain, 2006], but for the lack of a global scale magnetosphere within which the motion of charged particles is governed by an intrinsic planetary magnetic field. At Mars, the disturbance caused by the presence of the planet is much smaller compared to the size of the planet than at the Earth (see Figure 3). The distance of the obstacle “nose” at the Earth is ~10 Earth radii upstream while the nose of the obstacle at Mars is only a few hundred kilometers above the surface.
Due to the absence of magnetic obstacle at Mars, the solar wind interacts directly with the upper atmosphere and ionosphere and induces a magnetosphere by the pile up of the IMF. The solar wind decelerates to become subsonic as it crosses the bow shock (section 18.104.22.168) and enters the magnetosheath (section 22.214.171.124). Few solar wind protons are observed downstream a boundary sometimes called the Magnetic Pile-up Boundary (MPB) (section 126.96.36.199). Below the MPB is the Magnetic Pileup Region (MPR). The ionopause/Photoelectron Boundary (section 188.8.131.52) separates the planetary ionosphere (section 184.108.40.206) from the MPR. A two-lobed induced magnetotail (section 220.127.116.11) forms on the nightside, with a current sheet carrying planetary ions between the two lobes (see Nagy et al., 2004 for a more complete review on the distinct plasma boundaries and regions resulting from the interaction of Mars with the solar wind).
The bow shock and the upstream region
When the solar wind, which is supersonic, encounters on its way a conductive obstacle such as Mars, a bow shock forms upstream of the planet (green line in Figure 16). The solar wind is then heated, deflected, and slowed down when it passes the shock so that it becomes subsonic. In Figure 17 the bow shock can be observed at ~16:20 UT on the inbound part of the orbit. This first boundary separates two regions:
The upstream one which is characterized by solar wind plasma (mainly protons with a drift energy greater than 1000 eV and a temperature (width of the beam) of the order of 10 eV), a low plasma density and temperature (the ion beam is rather narrow on panel 2), a low magnetic field (at Mars < 5nT), and a supersonic flow (not shown). The typical shape of the electron flux spectrogram in the solar wind is plotted in red in Figure 18. The left part for energies below 10 eV should not be taken into account as it is due to the potential of the spacecraft (see section 2.2.4).
Figure 18. Electron energy spectra measured by SWEA in different regions observed in Figure 17.
(the solar wind, the magnetosheath, the ionosphere and the lobes). The vertical red line corresponds to the value of the spacecraft potential in the solar wind.
The downstream one, called magnetosheath, which is characterized by a more intense magnetic field and density, a subsonic flow (not shown) and a higher temperature (the ion beam is larger on panel 2). The typical shape of the electron flux spectrogram in the magnetosheath is plotted in dark red in Figure 18 and will be discussed in the next section.
Due to the low gravity of Mars and the absence of intrinsic magnetic field, the bow shock is close to the surface of the planet (~2000 km at the subsolar point, ~5500 km at the terminator), so that an important part of the exosphere is located well out of the shock and can interact directly with the incident solar wind. The interaction of Mars with the solar wind hence starts several planetary radii away from the planet, where the exospheric neutrals are ionized, mostly by photoionization and charge exchange. The resulting ions keep the same temperature as their neutral parents (a few eVs), but gain a small amount of energy during the process of photoionization. The new born ions, heavier but colder than the native ions of the solar wind, are accelerated by the electrical fields associated to the incoming flow, which tend to restore the thermodynamic equilibrium. These ions are called ‘pickup ions’. As the solar flow approaches Mars, an increasing number of cold exospheric newborn ions are created and lead to a significant massloading of the solar wind [Bertucci et al., 2011], which decelerates the flow by ~5% [Kotova et al., 1997]. The term ‘mass-loading’ refers to a situation where slow-moving mass is added to a flowing plasma, decelerating it via conservation of momentum.
Thanks to the several shock crossings observed by MGS, the altitude of the shock has been observed to increase from the subsolar point away to the tail, roughly forming a hyperboloid of revolution. Calculations of the bow shock shape have been made by Vignes et al. [2000, 2002] based on MGS crossings, by Trotignon et al. , based on MGS and Phobos data, and by Edberg et al. , based on MEX data. Comparison of the results of these three models are presented in Figure 19 [Edberg et al., 2008]. The bow shock corresponds to the outer boundary plotted and the three fits can be observed to be globally in good agreement, slightly less at the subsolar point.
Figure 19. Comparison of the location of the bow shock and of the MPB calculated by three models:
Edberg et al.,  (plain line), Vignes et al.,  (dashed line) and Trotignon et al.  (dashed-dotted). All MPB and bow shock crossings found from the pre-mapping phase of MGS are plotted as dots and plus signs, respectively, in aberrated cylindrical MSO coordinates (taking into account the aberration of the solar wind flow direction by the planetary orbital motion, see section 2.5.1) [Edberg et al., 2008].
After the crossing of the bow shock, we can see in Figure 17 that the spacecraft enters into a region with the same ion population as in the solar wind (mainly protons and ++ ions), at approximately the same energy, but with a much higher temperature. The electron and ion densities are larger than in the solar wind, so as the amplitude of the magnetic field which however shows much more fluctuations. Moreover, if we observe the electron energy spectra recorded in this region in Figure 18 (dark red one), we can see that the electron population is at higher energy than in the solar wind, and is warmer. These characteristics are typical of the magnetosheath.
The magnetosheath is the region standing between the solar wind and the effective obstacle (dark blue region in Figure 16). It acts like a buffer between a region dominated by the solar wind dynamic pressure and the planetary induced magnetosphere dominated by the planetary plasma pressure. It is populated by shock-heated, dense and turbulent solar wind plasma. Indeed, the solar wind plasma is slowed to velocities lower than 100 km. s−1, compressed by up to several times its original density, and heated to temperature up to ~4 times higher than upstream. The flow is also deflected from its original antisunward motion. As it travels deeper into the magnetosheath, the incoming flow continues to slow down.
Table of contents :
1. The Martian environment
1.1. Interaction of the solar wind with the different bodies of the Solar System
1.1.1. The solar wind
1.1.2. Four different classes of interaction
1.2. The Martian obstacle
1.2.1. Mars today
18.104.22.168. Atmosphere – Exosphere – Ionosphere: who is who?
22.214.171.124. The Martian magnetic field
1.2.2. Back to the history of Mars
126.96.36.199. A magnetic field history
188.8.131.52. A Mars’ volatile and climate history
1.3. The interaction of the solar wind with Mars
1.3.1. The steady-state interaction
184.108.40.206. The bow shock and the upstream region
220.127.116.11. The magnetosheath
18.104.22.168. The Magnetic Pile-up Boundary and the Magnetic Pile-up Region
22.214.171.124. The ionopause and the PhotoElectron Boundary
126.96.36.199. The ionosphere
188.8.131.52. The wake and the magnetotail
1.3.2. Dynamics of the Martian interaction with the Sun
184.108.40.206. Martian magnetic topology
220.127.116.11. Pressure balance
18.104.22.168. Variability of the boundaries
1.3.3. Focus on the nightside ionosphere
2. Instrumentation, data and analysis tools used
2.1. Exploration of Mars
2.1.1. Mars Global Surveyor
22.214.171.124. Scientific objectives
126.96.36.199. Main discoveries
2.1.2. Mars Express
188.8.131.52. Scientific objectives
184.108.40.206. Main discoveries
220.127.116.11. Scientific objectives
18.104.22.168. Main discoveries
2.2.1. Mars Global Surveyor
22.214.171.124. The magnetometer: MAG
126.96.36.199. The Electron Reflectometer: ER
2.2.2. Mars Express
188.8.131.52. Electron Spectrometer: ELS
184.108.40.206. The ion spectrometer: IMA
220.127.116.11. The ion spectrometer: STATIC
18.104.22.168. The Magnetometer: MAG
22.214.171.124. The Electron spectrometer: SWEA
126.96.36.199. The Langmuir probe: LPW
2.3. Data coverage
2.4. Analysis tools
2.4.1. AMDA and 3D view
2.5.1. The Mars-centric Solar Orbital (MSO) frame
2.5.2. The IAUMars frame
2.5.3. Definition of the altitude
2.5.4. Definition of the nightside
2.6. Model of crustal magnetic field: the model of Morschhauser et al. 
3. Identification of suprathermal electron depletions in the nightside ionosphere
3.1. A story of depletions
3.1.1. Discovery of electron depletions
3.1.2. On the origin of plasma voids
3.1.3. Global properties of the plasma voids observed by MGS and MEX
3.2. General properties of electron depletions observed with MAVEN
3.2.1. Plasma voids or suprathermal electron depletions?
3.2.2. Plasma composition
188.8.131.52. Ions characteristics
184.108.40.206. Electrons characteristics
3.2.3. An overview of the variety of the flux spikes
3.2.4. Are electron depletions really related to crustal fields?
3.3. Automatic detection of suprathermal electron depletions: definition of the criteria
3.4. Application of the criteria
3.4.1. Application to MGS
3.4.2. Application to MAVEN
220.127.116.11. Application of criterion (1)
18.104.22.168. MAVEN coverage
3.4.3. Application to MEX
22.214.171.124. Unrestricted application
126.96.36.199. Restricted application
4. On the processes at the origin of suprathermal electron depletions
4.1. Altitude dependence of the distribution of suprathermal electron depletions
4.1.1. Altitude distribution of the electron depletions observed by MAVEN
4.1.2. Geographical distribution of suprathermal electron depletions: a common vision from above 250 km
188.8.131.52. Geographical distribution at 400 km
184.108.40.206. Geographical distributions from 250 km to 900 km
4.1.3. Going down to 125 km altitudes with MAVEN
220.127.116.11. From 250 to 170 km
18.104.22.168. Below 170 km
4.2. A competition between two main loss processes
4.2.1. Plasma composition of suprathermal electron depletions
4.2.2. The role of crustal magnetic sources
22.214.171.124. Comparison between the northern and southern hemispheres with both MAVEN and MEX
126.96.36.199. Evolution of the altitude distribution of electron depletions with crustal magnetic field amplitude
188.8.131.52. Pressure balance
4.3. Discussion on the altitude of the electron exobase
4.3.1. Updated scenario of creation of suprathermal electron depletions
4.3.2. Evolution of the altitude of the exobase with the Solar Zenith Angle
5. Around the holes: the dynamics of the nightside ionosphere
5.1. Where the electron depletions stop: the flux spikes
5.1.1. Injection of ionospheric plasma
5.1.2. Energy-time dispersed electron signature
5.1.3. Current sheet crossing at low altitudes
5.2. Unexpected (non-)observations of suprathermal electron depletions
5.2.1. Observation of electron depletions on the dayside
184.108.40.206. An altitude issue
220.127.116.11. A spacecraft charging issue
5.2.2. Non-observation of electron depletions at low altitudes
18.104.22.168. Different types of orbits with no electron depletion
22.214.171.124. Distribution of the 61 events in the Martian environment
126.96.36.199. Focus on the Tharsis region
5.3. Where suprathermal electron depletions reveal the UV terminator
5.3.1. Observation of the UV terminator
188.8.131.52. Distribution of electron depletions as a function of SZA
184.108.40.206. Review of the nightside definition
220.127.116.11. Observation of the UV terminator with LPW and SWEA
18.104.22.168. Back to suprathermal electron depletions
5.3.2. Determination of the average altitude of the UV terminator
22.214.171.124. Distribution of electron depletions over one Martian year
126.96.36.199. Results and comparison with model
5.3.3. Evolution of the UV terminator with seasons: the dawn- dusk asymmetry
188.8.131.52. Predictions from the model of Robert Lillis
184.108.40.206. Results obtained with electron depletions
220.127.116.11. Focus on the equinox of 2016
18.104.22.168. The mystery of the reversal at the aphelion and perihelion
Conclusions and Perspectives