Abstract
The magnetic reconnection exhaust is a pivotal region with enormous magnetic energy being continuously released and converted. The physical processes of energy conversion involved are so complicated that an all-round understanding based on in situ measurements is still lacking. We present the evidence of plasma heating by illustrating the broadening of proton and electron velocity distributions, which are extended mainly along the magnetic field, in an exhaust of interchange reconnection between two interplanetary magnetic flux tubes of the same polarity on the Sun. The exhaust is asymmetric across an interface, with both sides being bounded by a pair of compound discontinuities consisting of rotational discontinuity and slow shock. The energized plasmas are found to be firehose unstable, and responsible for the emanation of Alfvén waves during the second step of energy conversion. It is realized that the energy conversion in the exhaust can be a two-step process involving both plasma energization and wave emission.
Export citation and abstract BibTeX RIS
1. Introduction
Magnetic reconnection is a crucial process of tremendous energy release in various magnetized plasma environments in which magnetic energy is converted into plasma bulk, thermal, and non-thermal kinetic energy. It is in the solar wind that one can measure in situ the physical processes in an extended exhaust region far away from the x-line of magnetic reconnection, and that one can learn to what extent the reconnected topology and the associated discontinuities can influence the plasma kinetics and wave activities.
Magnetic reconnection in the magnetohydrodynamic (MHD) regime is usually considered and simulated to have a pair of rotational discontinuities (RDs) and a pair of slow shocks (SSs) bounding the exhaust region (Petschek 1964; Sonnerup 1979; Lin & Lee 1995). The fast reconnection process has been successfully modeled more recently with Hall-MHD, hybrid, and full particle-in-cell simulations (Birn et al. 2001). However, unlike MHD modeling, probably due to the size limitation (usually less than hundreds of proton inertial length) of simulation box, kinetic simulation can hardly reproduce the discontinuities, e.g., RDs and SSs (Innocenti et al. 2015).
Interplanetary reconnection exhausts are commonly identified in slow solar wind observations and occasionally in the fast streams as well (Gosling et al. 2005a; Phan et al. 2006; Gosling et al. 2007; Vörös et al. 2014). Reconnection can change the topology of the interplanetary magnetic field (IMF) lines by interchanging their footpoints on the Sun (Crooker et al. 2012), by cutting their connection to the Sun (Gosling et al. 2005b), or by altering their connectivity among multiple disconnected structures. Simultaneous measurements by several spacecraft have shown that magnetic reconnection in the solar wind can present as a quasi-steady state and a large-scale process in terms of the x-line length (Phan et al. 2006). A profile with two steps of monotonic jump in the magnetic shear component, (BL), is often present across the solar wind reconnection exhaust. This profile differs from the gradual transition in the magnetic shear component that is often the signature of magnetotail reconnection (Eastwood et al. 2010). In association with the two-step jump in BL, the corresponding velocity component (VL) shows two steps of opposite jump (up–down or down–up), indicating the existence of a pair of RDs. These RDs propagate primarily along with the jet flow at about Alfvén speed, VA, as well as apart from one another into the external upstream region. Slow shocks in space plasmas have been identified in spacecraft data (Lin et al. 2008) and have been found to adjoin with an RD some times (Whang et al. 1998; Eriksson et al. 2004).
Ions and electrons are found to be hotter in the exhaust region than the average of the temperatures outside of it. The hot protons tend to be thermally anisotropic, with being larger than (Gosling et al. 2005a; Huttunen et al. 2007). It has been predicted by kinetic simulations and confirmed in statistical observations that the increment of the ion temperature is proportional to both the ion mass and the squared exhaust velocity (Drake et al. 2009). The anisotropic heating of electron cores seems to be more evident in the events with higher inflow Alfvén speed (Pulupa et al. 2014). Electron energization in the ion diffusion region manifests as a flat-top distribution with a pair of trapping-passing boundaries that separate the trapped electrons (trapped not only by the magnetic field but also by the electric potential) and the passing electrons with small pitch angles (Egedal et al. 2008, 2012).
In the ion diffusion region, whistler waves can be excited locally, with free energy coming from the non-Maxwellian phase space distribution of the electrons (Deng & Matsumoto 2001; Fujimoto & Sydora 2008), and electron vortex flow may also be formed due to the Kelvin–Helmholtz instability of the electron shear flow as drifted by the Hall electric field and guide magnetic field (Huang et al. 2015). Turbulence in the reconnection jet flow region is observed to have intermittent coherent structures (Vörös et al. 2004; Weygand et al. 2005; Huang et al. 2012), which seem to be the sites of concentrated dissipation (Osman et al. 2014). Firehose instability are found to be existent and mitigated by the developing waves/turbulence in the 2D particle-in-cell simulated exhaust region, which is bounded by coplanar SSs and non-coplanar RDs.
In the jet, downstream and quite distant from the reconnection diffusion region, the proton and electron velocity distribution functions VDF(), the key signatures of their thermal kinetics, have yet to be revealed simultaneously. What will happen to the plasma kinetics in the exhaust if the plasma states on both sides involved in the reconnection are asymmetric in number density and magnetic field strength, which is prevalent in various environments (Aschwanden et al. 1999; Øieroset et al. 2004; Liu et al. 2011)? Are the energized plasmas stable or unstable in the exhaust region? If unstable, are they responsible for the second step of energy conversion by exciting and emanating waves? These unknown elements of importance to magnetic reconnection are to be addressed in this work.
2. Observations and Analyses
2.1. Discontinuities Bounding and Separating the Reconnection Exhaust
We choose a reconnection exhaust region in the solar wind that was encountered by the ACE and the Wind spacecraft on 1997 November 23 (Gosling et al. 2005a; Phan et al. 2009; Pulupa et al. 2014) for a comprehensive study. Figure 1 shows an overview of this event based on the plasma and magnetic field measurements. The magnetic field strength, , is weakened at the trough in the time profile (Figure 1(a)). The trough is bounded by a set of jumps in the vector magnetic field components (Figure 1(b)) and in the proton bulk velocity components (Figure 1(c))—the jumps are marked with two vertical blue dashed lines and a vertical red dashed line on each side. Walén analysis of the negatively and positively correlated and jumps on each side confirms the existence of a pair of RDs (see Figure 2). The slope and correlation coefficient for Walén that test for the preceding RD are calculated to be −0.74 ± 0.02 and 0.95, respectively. The relation between local VA and V − VHT of the trailing RD is found to have a slope of +0.76 ± 0.03 and a correlation coefficient of 0.94. Under the global Minimum Variance Analysis (MVA; Sonnerup & Scheible 1998), BN is quite small, indicating that the RDs propagate upstream of the exhaust region very slowly in the N-direction. In the L and M directions, the magnetic field line kinks of the RDs may move much faster at a substantial fraction of Alfvén speed. This MVA analysis was verified to be insensitive to the choice of the time interval as long as all of the discontinuities were included appropriately in the time interval.
Download figure:
Standard image High-resolution imageFrom the time sequences of proton density (NP) and temperature (TP) across the reconnection exhaust (Figures 1(f) and (g)), one may identify four regions (from left to right): there are two distinct regions of dense, cool plasmas separated by a transition (region 1 and region 2), and two distinct sparse, warm plasmas, also separated by a transition (region 3 and region 4). Across the interface between regions 1 and 2, one can see a decrease in , an increase in NP and TP, as well as an approaching to zero of BL. This finite-width interface is inferred to be a possible SS according to the MHD Rankine–Hugonoit conservation relation analysis (Koval & Szabo 2008; Wilson et al. 2013). Similarly, the interface between regions 3 and 4 may also be classified as another SS. The upstream and downstream velocities in the shock frames on the two sides of the exhaust are (Uup,region-1 ∼ 31 km s−1, Udown,region-2 ∼ 21 km s−1) and (Uup,region-4 ∼ 95 km s−1, Udown,region-3 ∼ 49 km s−1), respectively. The angles between the normal direction and magnetic field direction are (θBN,up,region-1 ∼ 93°, θBN,down,region-2 ∼ 97°) and (θBN,up,region-4 ∼ 45°, θBN,down,region-3 ∼ 41°), respectively. This is only a rough estimate of the shock properties, which may not be highly accurate due to the significant deviation of plasma velocity distribution from local thermodynamic equilibrium, and due to the existence of compound discontinuities that are likely not in a steady state. The small changes of θBN between up- and downstreams of the shocks are due to their coexistence with RDs, for which the situation makes them look different from the traditional pristine slow shocks.
Therefore, the pair of back-to-back SSs seems to be concurrent with the pair of back-to-back RDs. The time interval between the RDs is about 10 minutes, which is equivalent to the length of the path across the exhaust region of about 50 Earth radii.
Across the interface between region 2 and region 3 (black vertical dashed line in Figure 1), there is a declining jump in NP and a rising jump in TP, maintaining roughly the same thermal pressure on either side. We do not see a dramatic jump in B and V components, although changes slightly across the interface. This interface is thus speculated to be a type of "contact interface" that keeps the two distinct materials relatively separate from one another. To see how the particles move and how the electric field self-consistently adapt around the interface, please refer to Section 2.3.
Proton thermal anisotropy, , as derived from the Wind/3DP/PESA-low measurements, is found to be smaller than 1 in the exhaust region (Figure 1(h)) where the proton parallel plasma beta () sees a dramatic increase (Figure 1(i)). This indicates a preferential parallel heating of protons in the exhaust region associated with a lesser increment in . In Figures 1(h) and (i), the proton core and beam populations are treated as a whole when calculating and .
Pitch-angle distributions of electron differential flux density (e-flux) at two energy channels (47.5 and 228.2 eV) are illustrated in Figures 1(j) and (k). It is found that, in the exhaust region and especially in the downstream of cool and dense plasmas, there is a distinct enhancement of parallel and anti-parallel e-flux at the lower energy channels less than 150 eV. For the higher energies, the electrons are mainly travelling along the magnetic field line. Therefore, it is the lower energy rather than higher-energy electrons that may be trapped and energized by the magnetic field configuration and electric potential distribution within the exhaust region. The direction of the dominant suprathermal electron flux at higher energies indicates that the magnetic reconnection is an interchange reconnection between IMF lines with the same polarity on the Sun, rather than a reconnection between IMF lines of opposite polarities causing disconnection and strahl dropout (Gosling et al. 2005b) or a reconnection between traditional non-kinked open and closed IMF lines (Crooker et al. 2004).
2.2. Scenario of Interchange and Asymmetric Reconnection in a Global Context
The interchange reconnection between two IMF flux tubes with the same polarity at the solar surface is illustrated conceptually in Figure 3(a). The kinked IMF flux tube, which may be bent by non-uniform flow speeds or formed by an earlier reconnection at a lower altitude or a growing result of kink instability due to strong magnetic twist inside flux ropes (Zaqarashvili et al. 2014), tends to be interchange reconnected with the ambient and relatively straight flux tubes. The reconnection might also be within complex interplanetary coronal mass ejections (ICMEs) with the far-end loosely connected to the Sun due to the existence of asymmetric but still counter-streaming strahl electrons. Figure 3(b) shows how the spacecraft passes through the magnetic reconnection exhaust configuration and what the jet flow velocity vectors look like. The change of state from (large , small ) outside to (small , large ) inside of the exhaust region is also illustrated. Flow velocity vectors, as denoted with yellow arrows, have two main components (VL and VM) in the exhaust region. The VL component is the velocity of the jet flow as accelerated by the slingshot effect of the reconnected magnetic field lines (MFLs). The VM component is mainly the result of the direction change of plasma momentum in the RD moving frame (one type of De Hoffmann–Teller (HT) frame; De Hoffmann & Teller 1950), in which both the upstream and downstream velocities are field-aligned and of similar magnitude. is approximately equal to VA, since the RD is propagating mainly in the L-direction at about VA with respect to the upstream plasma. In the reference frame of the downstream mean flow, VM changes from one polarity to another polarity in the exhaust region.
Download figure:
Standard image High-resolution image2.3. Plasma Kinetics around an Evolving Contact Interface
This kind of contact interface is different from the MHD contact discontinuity, since the free penetration of protons across the interface in the collisionless space plasmas is partially held back by a self-consistent electric potential barrier developed around the interface according to kinetic simulation results (see Figure 4).
Download figure:
Standard image High-resolution imageTo see if the contact interface can be maintained somehow and the protons can be reflected back at the contact interface, a 1D hybrid simulation is conducted to model the time evolution of the contact interface. Note that, in the setup of hybrid simulation, proton kinetics are described with the Vlasov equation and electrons are approximated as a fluid (Ruan et al. 2016). This 1D model provides a rough, but qualitatively useful, approximation to the evolution of spatial distribution of plasma states along the N-direction in the distant exhaust region. Here we neglect the spatial inhomogeneity of magnetized plasmas in the L-dimension, which is assumed to be relatively uniform as compared to the N-dimension in the distant exhaust region far away from the reconnection site. This 1D approximation is inspired by Lin & Lee (1995), who utilized the 1D MHD and hybrid simulations to study the evolution of the reconnection exhaust region by solving the Riemann problem there. Our 1D model differs from the preceding works in introducing counter streams in the opposite directions on either side of the interface to mimic the downstream flows in the HT reference frame, which is moving basically along with the jet flow in the L-direction.
Initially, across the contact interface, there exists a thermal pressure balance between both sides, with a cool and dense plasma and a warm and tenuous plasma on the left and right sides, respectively (see panels (a1), (a2), and (a4) in Figure 4). The plasmas on both sides are flowing in the opposite directions heading toward the contact interface. As the time evolves, a bipolar electric field (see (b5 )in Figure 4) is formed, leading to a positive potential around the contact interface. This positive potential plays a role in reflecting the major part of protons back to their original incident sides, while leaving the other protons penetrating through the interface into the opposite side (see (b1) in Figure 4). By applying the moment analysis to the proton velocity distribution, the density and temperature are found to increase on both sides but still remain as a significant transition (imbalance) across the interface.
2.4. First Step of Energy Conversion: Energization of Protons and Electrons
The proton heating signature in the exhaust region is illustrated in Figure 5. Eight snapshots of p-VDF are aligned in position to follow the MFL configuration. The p-VDFs at (t1), (t2), (t3, t4) are upstream of the cool and dense plasmas in the transition layer, and in the deep downstream of the cool and dense plasmas, respectively. From the upstream to the downstream, the p-VDF experiences a significant change: development of a beam population parallel to B0, with the core population heading toward the contact interface. On the other side of the contact interface, the p-VDFs downstream and upstream of the warm and sparse plasmas are displayed in panels (t5)–(t8). The p-VDFs at (t5, t6) have two populations with the main population drifting along the field toward the contact interface and the second population toward the compound structure of RD and SS. The relative drift between the primary and secondary populations at (t5, t6) is larger than that at (t3, t4), causing part of the second population to extend beyond the measured range.
Download figure:
Standard image High-resolution imageThe core + beam p-VDF in the exhaust region is commonly called a dumbbell distribution if the two components are well detached from and comparable to one another. It is interesting that the p-VDFs show different patterns on both sides of the contact interface. It is expected that the HT (propagating RD) frame is moving mainly along with the jet flow in the L-direction. The upstream plasma in the HT frame moves toward the discontinuity at about VA, and turns to follow the field line direction in the downstream exhaust region with a slight slow-down in speed due to the higher thermal pressure downstream. Traditionally, in the exhaust region of symmetric reconnection, the major core population may be the plasmas coming from the nearest upstream, while the minor beam population may be from the upstream on the other side. However, in our case with reconnection between asymmetric plasmas, the beam population on one side of the contact interface is distinctly different in the thermal state from the core population on the other side, yet similar to the core population on the same side. Therefore, the beam (second) population might be attributable to reflection at the contact interface by some electric potential barrier yet to be measured and confirmed.
Note that coordinate origin of the proton-VDFs in Figure 5 has been translated to the velocity position with maximum phase space density from the original velocity position of the spacecraft, which can be approximated as static when compared to the supersonic solar wind flow. To illustrate the difference of both bulk velocity and thermal state between the upstream and downstream of the exhaust region, we present the original 3D-VDFs in Figure 6. The VDFs of α particles are also recorded in the figure, showing remarkable expansion of the contour surfaces in the exhaust region, similar to that of protons. The velocity for α particles in the figure is just derived with the proton's mass-to-charge ratio without adopting its real mass-to-charge ratio.
Download figure:
Standard image High-resolution imageThe electron velocity distribution function (e-VDF) illustrated in Figure 7 also shows a clear signature of the parallel heating of its core component in the exhaust region. As a result of the preferential parallel heating, the pitch-angle distribution of the differential flux density at energies lower than 200 eV is mainly concentrated around 0° and 180° (Figure 1(j)). For the electrons above 200 eV, which do not show a difference in e-VDF between the exhaust and ambient region, they seem not to be affected and trapped by the speculated electric potential. The same dominant direction of the asymmetric, but still counter-stream, strahl electrons above 200 eV (parallel to local B0) throughout the upstream and downstream regions indicates the nature of the interchange reconnection between two IMF flux tubes (probably affiliated with a complex B-field system involving ICME) with the same (positive) polarity of their near-end footpoints on the Sun. Note that the existence of weak, field-aligned proton heat flux in the upstreams on both sides of the exhaust region, as discerned from the measurements by the 3DP and Solar Wind Experiment (SWE) instruments, also indicates the same polarity of near-end footpoints for these two flux tubes.
Download figure:
Standard image High-resolution image2.5. Second Step of Energy Conversion: Wave Excitation and Turbulence Enhancement
Besides the RDs with steps of correlated/anti-correlated jumps mainly in BL and VL components, the fluctuations in other time intervals show a good sense of Alfvénic correlations as well, which are illustrated in Figures 8(a)–(c) showing the cross-correlation spectra of (BL, VL), (BM, VM), and (BN, VN). For example, BN and VN are highly anti-correlated on the cool and dense side of the contact interface, while in positive correlation they are on the warm and sparse side. Unlike the single step of jump in BL and VL across the RD, the BN and VN components oscillate in large amplitude over several cycles near the RD. These features indicate that Alfvén waves, besides the RDs, are propagating in the opposite direction on the either side of the contact interface. There seems to be a wave source near the contact interface, emanating Alfvénic fluctuations outward. The Alfvén waves generated in the exhaust region are stronger in power spectral density (PSD) than those in the ambient undisturbed regions (see Figure 9).
Download figure:
Standard image High-resolution imageDownload figure:
Standard image High-resolution imageThe free energy to generate these Alfvénic fluctuations may come from the plasma kinetic energy through a certain instability mode, e.g., the magnetic firehose instability (Bale et al. 2009; Maruca et al. 2012; Klein 2017). The magnetic tension can be reduced in the plasma with large and/or with relative drift between multi-components/multi-species. As a consequence, the MFL is subject to oscillating deformation and propagating waves. Patchy regions of firehose unstable plasma have been measured in exhausts in the Earth's magnetotail (Hietala et al. 2015). Those studies did not, however, present evidence of excited waves. As illustrated in Figure 5, the p-VDFs with a dumbbell shape may be prone to the firehose instability. Since the 3DP/PESA-low cannot record the full p-VDF at some times, we adopt the parameters of protons and alpha particles derived from the SWE measurements (Kasper et al. 2002; Stevens & Kasper 2007). We calculate , a parameter that indicates long-wavelength firehose instability when it takes a value greater than 1 (Chen et al. 2016), based on the magnetic pressure, PB, and the parameter sets (mass density, , drift velocity, , and parallel and perpendicular thermal pressure, & ) of various species denoted as s (proton core, proton beam, and alpha particle). The time sequence of is plotted in Figure 8(f), which shows that is larger than 1 at two moments in time, with the largest one being close to the contact interface. We also scatterplot the ion (protons and α particles) plasma parameters (, ) and compare them with the instability thresholds, finding that the downstream plasma conditions (red solid diamonds and blue solid squares for warm and cool downstreams, respectively) are closer to or even beyond the threshold of firehose instability (Hellinger et al. 2006). The threshold formulas of (, ) for MHD, kinetic parallel, and kinetic oblique firehose instabilities are , , and , respectively (Hellinger et al. 2006). The effective parallel plasma beta, , is calculated with , where the subscripts "pc," "pb," and "α" represent the proton core, proton beam components, and alpha particles, respectively, and the drift velocities are with respect to the mass center of the proton core, proton beam, and alpha population. The effective pressure anisotropy is determined by .
3. Summary and Discussion
The evidence for plasma energization in the reconnection exhaust are presented here, showing the signature of preferential parallel heating as characterized with the dumbbell-shaped p-VDF and the field-aligned elongated e-VDF. Furthermore, the derived plasma parameters are found to be locally firehose unstable in the exhaust, suggesting that the instability may be responsible for the enhanced PSD level of magnetic turbulence and the opposite propagation of Alfvenic waves away from the contact interface. The plasmas are asymmetric, with a cool and dense plasma on one side and a warm and sparse plasma on the other side of the contact interface. We hypothesize that there may be an electric potential barrier (yet to be detected in future missions) formed around the contact interface, which partly prevents the free penetration of protons and energizes the trapped electrons.
The energized plasmas, which store excess free energy, are not the ultimate deposit of converted energy. Instead, they become firehose unstable and excite waves. The emanation of Alfvén waves in the second step of energy conversion might be the source of sunward propagating Alfvén waves recently identified in the solar wind (He et al. 2015; Li et al. 2016). Magnetic reconnection was traditionally thought to occur between the IMF lines with the opposite polarities on the solar surface (Gosling et al. 1995; Huang et al. 2016). It is newly suggested that curved IMF lines with the same polarity on the Sun can also be reconnected even at 1 au, interchanging their footpoints. This discovery expands the applicability of magnetic reconnection inflow tubes stemming from the areas of the same polarity.
This study highlights that the reconnection exhaust region in the interplanetary space, which may extend over tens of Earth radii (much larger than the ion diffusion region of about 100 km), is an important site where plasma heating and wave excitation are taking place in association with the bulk jet flows. These insights help to understand (a) the complex physical processes (multiple steps of energy conversion) as triggered by the reconnection, (b) the application of energy conversion in the exhaust to the plasma heating and turbulence development in various environments, and (c) the spectroscopic diagnosis of the multi-component line profiles emitting from the flare sites on astronomical bodies.
The authors are grateful to the teams of the Wind and the SOHO spacecraft for providing the data. The work is supported by the National Natural Science Foundation of China (NSFC) under the contracts 41574168, 41222032, 41231069, 41474147, 41674171, and 41421003. J.S.H. is also supported by the National Young Talent Program of China. M.S. is supported by NASA grant no. NNX14AT26G.