Impacts of ionospheric plasma on magnetic reconnection and Earth’s magnetosphere dynamics

Ionospheric ions (mainly H+, He+ and O+) escape from the ionosphere and populate the Earth’s magnetosphere. Their thermal energies are usually low when they first escape the ionosphere, typically a few eV to tens of eV, but are energized in their journey through the magnetosphere. The ionospheric population is variable, and it makes significant contributions to the magnetospheric mass density in key regions where magnetic reconnection is at work. Solar wind magnetosphere coupling occurs primarily via magnetic reconnection, a key plasma process that enables transfer of mass and energy into the near-Earth space environment. Reconnection leads to the triggering of magnetospheric storms, aurorae, energetic particle precipitation and a host of other magnetospheric phenomena. Several works in the last decades have attempted to statistically quantify the amount of ionospheric plasma supplied to the magnetosphere, including the two key regions where magnetic reconnection proceeds: the dayside magnetopause and the magnetotail. Recent in-situ observations by the Magnetospheric Multiscale spacecraft and associated modelling have advanced our current understanding of how ionospheric ions alter the magnetic reconnection process at mesoand small-scales, including its onset and efficiency. This article compiles the current understanding of the ionospheric plasma supply to the magnetosphere. It reviews both the quantification of these sources and their effects on the process of magnetic reconnection. It also provides a global description of how the ionospheric ion contribution modifies the way the solar wind couples to the Earth’s magnetosphere and how these ions modify the global dynamics of the near-Earth space environment.

In the dayside cusp, energy often comes from waves or accelerated particles originating from magnetic 166 reconnection or other processes at the dayside magnetopause. Wave-particle interactions seen in this depends on the convection path and the location of the magnetopause. The plasmaspheric plasma is 241 also heated as it expands in the magnetosphere (e.g., Genestreti et al., 2017), although the degree of 242 heating is variable and there are certainly times when very cold plasmaspheric material is observed at 243 the magnetopause. Finally, the density within the plasmaspheric material is quite variable. Detailed 244 density measurements across plumes show variations of an order of magnitude (e.g., Chappell, 1974; line that separates drift paths around the dawnside from those on the duskside in Figure 3, i.e., the 247 drainage region.

248
In addition to plasmasphere erosion by magnetospheric convection, there are other mechanisms that 249 facilitate ion escape from the plasmasphere to the outer magnetosphere: the plasmaspheric trough 250 (Chappell et al., 1971) and the plasmaspheric wind (Dandouras et al., 2013).

251
The plasmaspheric trough occurs at magnetic latitudes slightly above the plasmapause. The classical 252 polar wind lifts light ions (H + and He + ) and electrons in the same way as inside the plasmasphere, but this 253 plasma it located at open drift paths (see Figure 3), outside the corotating plasmasphere. Typical 254 densities of the plasmaspheric trough are ~10 cm -3 at L-shell = 4 (Chappell, 1971), which drop to few 255 tenths of cm -3 when they reach the magnetopause at L-shells of 10 -12, due to radial expansion.
For the same amount of parallel energization, lighter ions have larger parallel velocities than heavy ions.

285
On the other hand, magnetospheric convection acts in the same way for all species. As a consequence, a 286 mass filter effect arises when magnetospheric convection is non-negligible: lighter ions escaping the 287 polar cap region or nightside auroral zone travel further along the magnetic field line before reaching 288 the plasma sheet in the magnetotail than heavier ions (Figure 4). A velocity filter effect also applies 289 within a single species: the slowest, i.e., less energetic, ions being deposited close to Earth, and faster, 290 i.e., more energetic, ions further tailward. Some of the fast ions from this region will escape directly into  farther a particle travels down the tail, the more the magnetic field lines are distended in the center 360 plane of the tail and the more curvature drift the particle will encounter when it enters this region.

361
Particles which become the warm plasma cloak enter the tail earthward of those which become the 362 plasma sheet and subsequently the ring current.

367
Of direct significance to this review is the fact that the warm plasma cloak particles are convected past 368 dawn to the post-noon magnetopause where they enter the reconnection region at the nose of the 369 magnetosphere, changing the plasma characteristics both by energy and mass and thus affecting the 370 rate of reconnection on the dayside, see sections 3 and 4. A corollary to this, of course, is that the entry 371 of the ionospheric ions into the centerline of the magnetotail also affects the reconnection process 372 there. As the polar/lobal wind particles (10's to 100's of eV) enter the distended field lines of this region, 373 they will begin to be accelerated, thereby influencing the conditions conducive for magnetic 374 reconnection in the magnetotail.

375
In summary, the flow of ionization from the ionosphere through the lobes to the central plane of the 376 magnetosphere affects two different areas of reconnection, initially the neutral sheet area of the tail 377 and potentially, through the sunward flow of the warm plasma cloak to the dayside magnetopause.
Tracking ionospheric plasma as it flows through the magnetosphere requires two critical components: 1) Polar). Early estimates of the contributions of these initially low energy particles showed that they were 449 sufficient in terms of density to create the major observed plasma regions of the magnetosphere 450 (Chappell et al., 1987). Later ion trajectory studies of these up-flowing cold ions showed that they not 451 only moved through the different magnetospheric regions, but in so doing were energized to match the 452 observed energies in these regions (plasmasphere, plasma sheet, warm plasma cloak, and ring current)

492
There have been recent global modelling efforts including the ionospheric source, which clearly indicate 493 their relevance for populating the Earth's magnetosphere, as discussed in section 2.3. The main 494 drawback of these models is that they need to couple many different physical processes occurring at 495 very different spatial scales and plasma regimes, from the highly collisional ionosphere, including 496 chemical processes to assess the plasma density and composition, to the collisionless magnetosphere 497 and convection of the magnetic field lines.

498
In the following subsections, we first describe the techniques for detecting cold ions (up to ~10 eV), 499 corresponding to the initial energy of ionospheric ions when they escape to the magnetosphere. Then, 500 we review all the available statistical in-situ and remote observations near the two main reconnection 501 regions in the Earth's magnetosphere: the dayside magnetopause and the magnetotail. We describe and 502 put together the statistics of observations of cold ionospheric-originating ions in these two key regions.

503
As mentioned in section 2, not all ions of ionospheric origin are cold when they reach the reconnection 504 regions. However, from an observational perspective, it is not possible to distinguish the origin of hot 505 (keV) protons. The cold ions discussed in this section correspond to the young ionospheric plasma 506 supply, in the sense that they did not yet have time to be energized significantly, and correspond 507 unequivocally to the ionospheric source. with energies of less than ~10 eV, such as those directly originating from ionospheric outflow and the 512 plasmasphere, are often hard to detect in space plasmas. A main source of this difficulty arises from the 513 fact that a sunlit spacecraft in a low-density plasma becomes positively charged up to tens of volts 514 (Grard, 1973;Garrett, 1981;Whipple, 1981). Hence, positively charged ions at very low energies will not 515 reach the spacecraft and cannot directly be detected. Various techniques have been developed to overcome this challenge (André and Cully, 2012). determine the plasma density at a specific altitude (Benson, 2010). With ground-based radars and incoherent scatter radars, several plasma parameters of the ionospheric plasma populations can be estimated (Ogawa et al., 2009). In the magnetosphere, passive remote sensing with instruments on 522 spacecraft detect EUV solar photons resonantly scattered from He + ions (Spasojevic and Sandel, 2010).

527
Observing plasma in situ with detectors onboard a spacecraft allows for direct measurements of local 528 plasma properties, but adds uncertainties in the observations caused by interaction of the spacecraft 529 itself with the plasma. In the source region of ionospheric outflow, the plasma density can be so high 530 that the spacecraft potential becomes zero or slightly negative, due to many impacting electrons on the 531 spacecraft surface, allowing for low-energy populations to be measured. At altitudes of a few hundred 532 km ion detectors are used to study positive ions at low energies (Shen et al., 2018). Additionally,

533
Langmuir probes are used to determine electron density and temperature in dense plasmas (Brace,

535
At higher altitudes in a low-density plasma, low-energy ions are still able to be observed in situ, for 536 instance when a satellite is in eclipse (i.e., in the Earth's shadow) during short periods, and hence 537 become negatively charged (Seki et al., 2003). When a spacecraft is positively charged, an indirect 538 method for measuring the cold ion density is to estimate the total plasma density from observations of ). In addition, the total plasma density is estimated from the spacecraft 542 potential. This potential depends on the density and the electron temperature but can in many 543 magnetospheric plasmas be calibrated and used to estimate the total density (e.g, Grard, 1973; Laakso 544 and Pedersen, 1998; Lybekk et al., 2012; Jahn et al., 2020). To obtain particle distribution functions in 545 velocity space, the positive charging of the spacecraft that repels the positive ions must be reduced. One 546 method is to use a negatively charged aperture plane around the ion detector entrance, as was used for 547 the RIMS instrument on Dynamics Explorer (Chappell et al. 1980). An alternative approach is to 548 negatively bias the entire instrument or a large part of the spacecraft as done for the Magnetospheric 549 Plasma Analyzers (MPAs) on certain geosynchronous spacecraft (Borovsky et al., 1998). Yet another , but often a spacecraft potential of a few volts remains. We note that several studies concentrate 553 on initially cold ions that have been heated (i.e., larger thermal velocity than expected given the 554 plasmaspheric or ionospheric source) or are drifting, e.g., due to E x B motion, i.e., large enough bulk 555 velocity to overcome the spacecraft charging (e.g., Lee and Angelopoulos, 2014). In these situations ion

559
An alternative method for determining the presence and properties of a cold ion population utilizes the 560 fact that a supersonic flow of cold positive ions can create a large enhanced wake behind a positively 561 charged spacecraft. The wake will be filled with electrons with a thermal energy that is higher than the 562 ram kinetic energy, in contrast to that of the ions. This creates a local electric field which can be 563 observed and then used to detect the presence of cold ions. Using multiple instruments to measure the density, the cold ion flux can be deduced (Engwall et al., 2009). This method requires one technique to determine the local electric field, such as detecting the potential difference between probes on wire 567 booms in the spin-plane of a spinning spacecraft, and another to characterize the essentially 568 unperturbed geophysical electric field, such as detecting the drift of artificially emitted keV electrons gyrating back to the spacecraft, as is done with an instrument onboard the Cluster and Magnetospheric studies covering a major part of solar cycle (André et al., 2015). In addition, observations from the MMS 572 spacecraft have been used to show that charging of the individual wire booms affects observations, but 573 can also be used to obtain information on cold ions (Toledo-Redondo et al., 2019). observations can only be done from a statistical perspective, using from months to years of spacecraft 581 observations. Different missions have different orbits, including equatorial versus polar orbits, and 582 different or even varying apogee and perigee distances. In addition, the dayside magnetopause location 583 is dynamic, most of the time being located between 8 -12 R E from Earth. Another important difference 584 between studies is the instruments and associated techniques they use for inferring the plasma 585 properties, in particular density, composition and temperature. We decided to group the studies by the 586 main technique they use for cold ion detection. Since the studies reviewed in this section use different 587 spacecraft, different techniques, and even different definitions of ionospheric plasma, one needs to be 588 careful when comparing their results. We tried to enunciate the main points to consider for each of 589 these studies when discussed together. At the end of this sub-section, we provide a table with the main  The most straightforward technique to infer the properties of cold ions in space plasmas is by using the near the dawn-side magnetopause was 30% (see Figure 6), considering as detection any flux above the 610 noise level of the instrument at the low energy range. Their occurrence probabilities are higher for 611 larger L-shells, and this is because the ion detector requires that the bulk plasma velocity has higher near the magnetopause, where local motions of the boundary and ULF waves accelerate the cold 614 plasma to energies above the spacecraft potential. The dawn-dusk asymmetry is explained by the 615 location of the drainage region, which is predominantly in the dusk sector. Finally, they also compared the statistical occurrence of thermal ions with the orientation of the solar wind magnetic field, or
They found that plasmaspheric plumes are a persistent feature of geomagnetic storms, and that they 626 last for ~4 days. Their typical flow velocities are ~15 km/s towards dayside magnetopause. The average 627 mass flux is ~2 x 10 26 ions/s, and the average mass released per event is ~2 x 10 31 ions. These numbers 628 indicate that plumes constitute a primary escaping path of plasma. The plume plasma density, flow 629 velocity and width all decrease with the plume age. However, these observations are taken far away 630 from the Earth's magnetopause, which is typically situated at ~10 R E . Assuming an effective area of the 631 drainage plume region at the magnetopause of ~9 x 12 R E , as in André and Cully (2012), and an average 632 outflow velocity of ~15 km/s (Borovsky and Denton, 2008), the resulting ionospheric average density at 633 the magnetopause in the drainage region corresponds to ~3 cm -3 during storm times.

634
The previous studies discussed the presence of plasmaspheric material in the outer, dayside

661
19%. This number is lower than for other studies probably due to non-detections of cases with total 662 energies below 10 eV. With regards to the WPC, they find for the dawn sector 17 events out 221 663 featuring bidirectional jets for more than 2 min, corresponding to 8% occurrence. This value is again 664 lower than previous estimates by Nagai et al. (1983) and Chappell et al. (2008), and the reason is that 665 their threshold requirements for density and duration for considering detection were more restrictive 666 for this study. They estimate a median density of 5.4 cm -3 for plasmasphere-originating ions, and a 667 median density of 5.2 cm -3 for the WPC, indicating that they captured only very dense events. dayside magnetosphere (i.e., plasmaspheric plumes and dense WPC), at distances > 7 R E , and within 1.5 h in time from the magnetopause crossing. They exclude the magnetosheath and the low-latitude boundary layer by imposing the requirement that no significant He ++ is present. They distinguished the have n He+ /n O+ < 1 and plume intervals have n He+/ n O+ > 1. Outflow from the high-latitude ionosphere is 679 dominated by O + with much less He + (e.g., Collin et al., 1988); thus, it stands to reason that the WPC is 680 distinguishable from the plume by its O + content. Since their observations rely on particle instruments,   considering plume detection at 10 R E was 3.8 cm -3 and therefore looked only for high-density plumes.

695
The density was inferred from the average spacecraft potential, during 2 min of observations in the 696 magnetosphere adjacent to the crossing. In principle, inferring the density from the spacecraft potential 697 has the advantage of accounting for typically 'hidden' low-energy ions, but this method has to be 698 carefully calibrated by comparing with other observations (cf. Section 3.1). They found that 137 out of 699 520 crossings (26%) contained the high-density plasmaspheric plume adjacent to the magnetopause in 700 the dusk sector, with most densities greater than 5 cm -3 and up to more than 100 cm -3 .         spacecraft potential measurement rather than from particle measurements. However, they search only 803 for high-density plumes, and impose density thresholds of ~5 cm -3 (varying with radial distance to Earth).

804
They find an occurrence probability of 26% and average densities of 5 -10 cm -3 . Based on these results,

805
we note that studies relying on particle detectors cannot provide accurate occurrence probabilities,

806
because the cold ions often do not reach the detectors due to the positive spacecraft potential.

807
We conclude that cold ions are present at the dusk magnetopause > 80% of the time, with average H + 808 densities of at least a few times 0.1 cm -3 . This includes 20 -25 % of the time when the density is > 3 cm -

826
and André and Cully (2012). This is probably related to the higher upper energy limit they use (400 eV).

827
The WPC can be often be found at energies >100 eV, and these events would be missed by the other 828 studies. In addition, the ion detector used by Chappell et al. (2008) has a larger geometrical factor, 829 because it was specifically designed to measure cold to warm ion populations (few eV -400 eV).
For the dawn side, we conclude that the probability of finding the WPC at the magnetopause in the 831 dawn side is > 50% -70%, with average H + densities of few tenths of cm -3 to few cm -3 . For ~10% of the 832 time, the average H + density is ≥ 3 cm -3 . Similar to the duskside studies, there are too few studies using 833 composition to investigate composition differences with L-shell or local time.

895
although Lennartsson (1989) observed overall higher densities ( Figure 10a). This is likely because the 896 radial range extended to lower radial distances, where the density is higher (as will be discussed). Figure   897 10b compares the O + /H + density ratio from the Cluster study as a function of F10.7 with the Nose et al. (2009)

911
The plasma sheet densities closer to the Earth, at ~6 -7, R E were characterized by Young et al. (1982) 912 and Kistler and Mouikis (2016). Both studies cover the "hot" population, with Young et al. (1982) 913 covering the energy range from 0.9 -17 keV using the GEOS data set (similar to the instrument on ISEE

1018
We summarize, in  The H + number density in the near-Earth plasma sheet (less than 10 R E from Earth) depends mainly on 1032 the distance to Earth and geomagnetic activity, and to a lesser degree on solar EUV. It is estimated that

Distant Plasma Sheet (> 10 R E )
In the distant tail, the total H + number density does not depend that much on geomagnetic activity

1122
The rest of the right-hand side terms of Equation 1 become non-negligible at electron scales, i.e., in the negligible. Therefore, dissipation of the magnetic field (positive J·E) inside the EDR must occur in an 1125 unconventional way, since the collisional resistivity is too weak to explain the observations. One of the 1126 primary objectives of the MMS mission is to unravel which processes generate sufficient anomalous resistivity for the magnetic field to diffuse and reconfigure inside the EDR in collisionless plasmas. Waveparticle interactions are a strong candidate for the generation of anomalous resistivity (e.g., Graham et temperature or mass, has an impact on the different characteristic time and spatial scales associated with the diffusion regions, and affects how the process converts magnetic energy into thermal and

1139
where subscript s denotes the particle species, v T is the thermal speed, q is the particle charge, m is the 1140 particle mass and B is the magnetic field magnitude. A colder plasma population will, for instance, have where c is the speed of light, 0 is the vacuum dielectric permittivity, and is the species number 1150 density. The height of the IDR and EDR are more precisely described by the particles bounce width, 1151 which involves the thermal velocity of the particles and therefore scales approximately as the Larmor 1152 radius. Figure 13 shows a 2D particle-in-cell simulation of asymmetric magnetic reconnection (Dargent et al.

1211
The normalized reconnection rate is then readily defined as the ratio between the inflow and the 1212 outflow velocity, which can be related to the aspect ratio of the diffusion region 1213 =~ ,

1214
where we have assumed an incompressible flow (∇ · = 0), and an outflow velocity equal to the Alfvén 1215 speed. The reconnection rate is directly related to the out of plane electric field in the diffusion region assume that = − × holds at the edges of the diffusion region, i.e., the ions are frozen-in to the magnetic field outside the diffusion region, leading to = . Taking advantage of ∇ · = 0, we find 1220 ~ = . (8) concluded that the Hall term was the cause of the normalized reconnection rate having a rate of 0.1.

1399
The Hall E field in the separatrices energizes the demagnetized ions that cross it, and therefore one

1403
two simulations had identical asymptotic conditions, that is, magnetic field magnitude, and total particle 1404 density and temperature. They found that the maximum Hall E field (Figures 15a and 15d) was reduced 1405 in the separatrices for the simulations with cold ions, but that at the same time the Hall E field layer was 1406 wider (Figures 15b and 15e), resulting into very similar integrated potential drops across the separatrix  The presence of multiple ion populations in magnetic reconnection also affects the topology of the IDR.

1422
Each ion population has its own characteristic spatial scales, namely the ion inertial length and the

1423
Larmor radius, which depend on the atomic mass and temperature of the population. Therefore, each 1424 ion population sets its own ion diffusion region, resulting in a multi-scale ion diffusion region (two or more layers). This behavior in magnetic reconnection has been observed both using PIC simulations and  H + in addition to hot H + and electrons. They also found a multi-scale diffusion region, where the cold ions 1435 remain magnetized down to smaller scales than the hot H + , owing to their different Larmor radius.

Kinetic effects on reconnection rate
We have already seen in section 4.1 that ions of ionospheric origin have a mass loading effect on the magnetic reconnection rate. However, based on the scaling analysis (see Equations 8 and 9) there is no 1471 direct dependence on temperature. For instance, a cold plasma population should not affect the 1472 reconnection rate as long as the total mass density remains constant. However, a cold population 1473 introduces a new length-scale and therefore should lead to a reconfiguration of the diffusion region. In 1474 particular, a cold ion population is expected to reduce the average height of the ion diffusion region ( ).
reconnection rate would then imply that the diffusion region configures itself so that the length L is 1477 reduced in order to keep constant the aspect ratio δ/L.

Effect of streaming ions (suppression of rate) 1581
An additional effect beyond the previous described slowdown scenarios is the involvement of a moving   another mechanism than can lead to the formation of such crescent-shaped distribution functions for 1619 cold ions. Cold ion distribution functions take a crescent shape along the magnetospheric separatrices, 1620 without magnetic field reversal. In this case, the driver of the signature is the Hall electric field, which 1621 accelerates and then decelerates the cold ions during one Larmor gyration.

1623
In collisionless plasmas, waves play an important role in coupling the various particle populations.

1639
However, magnetic reconnection also produces heating of the particle populations. This heating is  The results suggest that the cold ion beams are accelerated close to the X line ( Figure 18, red arrow).

1663
The study illustrates that the cold ion heating by reconnection is not homogeneous along the separatrix 1664 and suggests it may be lower close to the X line.  Quantitative and statistical analysis of the cold-ion beams are needed to assess the contribution from each acceleration mechanism.
Baker et al. (1982) proposed that O + would enhance the linear ion tearing instability in the plasma sheet, and so decrease the stability of the tail to reconnection. Multiple studies since then have attempted to closer to the reconnection region did not find that O + is enhanced prior to onset except during stormtime sawtooth events (Lennartsson et al., 1993, Kistler et al., 2006, Liao et al., 2014. Another method to test this is to determine whether there are more substorm onsets when there is more O + . Lennartsson

1761
The conditions that may suppress reconnection via diamagnetic drift can be expressed in terms of the difference in plasma  (the ratio of thermal plasma pressure to magnetic pressure) on either side of the the reconnection site: where L is the current sheet width and is the ion inertial length (Swisdak et al. 2010) not expected to be important because the  magnitude in the plasma sheet is typically smaller than in plasma sheet populations in both sides, and therefore no substantial  is expected across the current sheet.
1818 c For calculating  in the magnetosphere, 50 nT is used for the magnetic field strength. Values used 1819 for "cold", "warm", and "hot" temperatures are 1 eV, 100 eV to 1 keV, and 10 keV, respectively.

1827
contribution of each source to the plasma sheet is still subject of debate and we discuss it in Section 5.

1828
In addition to plasma sheet ions, cold and heavy ionospheric outflows, detached plasmaspheric material 1829 and the WPC populations can also be present at the reconnection regions (cf. Sections 2.1 and 2.2).

1830
Section 4 has focused on reviewing the effects of these cold and heavy ionospheric-originating 1831 populations in magnetic reconnection.

1832
Cold and heavy ionospheric ions are often present and have non-negligible contributions to the mass 1833 density in the two key reconnecting regions: the dayside magnetopause and the magnetotail. These 1834 ionospheric ions mass-load the reconnecting flux tubes, and locally reduce the reconnection efficiency.
indirect evidence, using geomagnetic indices, of global reduction of the coupling to the solar wind when In addition, cold and heavy ionospheric ions introduce multiple spatial and time scales into the reconnection process, owing to the dependence of gyroradius and gyrofrequency on temperature and particle mass. A large set of microphysical effects, including multiple-scale IDRs, wave-particle aspect ratio of the diffusion region and the energy conversion mechanisms are changed, resulting in significantly modify the normalized reconnection rate once a steady state has been reached (Divin et al.,     Table 4 summarizes these open questions, grouped in two categories. the solar wind, we cannot directly determine the origin of the H + in the magnetosphere: it all looks the dominant during southward IMF periods associated with increased geomagnetic activity. Overall, the relative contributions of the ionosphere and the solar wind are estimated to be of the same order of is the most straightforward method to discriminate the origin of H + , but these models are challenging 1888 because they need to account for many processes occurring at very different scales.        How is the plasma sheet formed?
Does the variable magnetospheric density affect the global coupling with the solar wind efficiency?
Kinetic physics of magnetic reconnection How do the microphysics introduced by multiple ion populations change reconnection at MHD scales?
Does the WPC alter the suppression of magnetic reconnection?
Which portion of the reconnection energy is taken by cold and heavy ions?
What are the effects of cold electrons in magnetic reconnection?
How ionospheric ions in the plasma sheet condition the onset of magnetic reconnection?