Solar flares are characterized by a sudden enhancement of the electromagnetic radiation in a broad range of wavelengths on the Sun. 1 They are believed to be one of the dominant sources of severe disturbances in the space environments of the Earth and other planets in the solar system. Flares are generally believed to result from magnetic reconnection, a physical process referring to the re-arrangement of the magnetic field topology in a plasma. Through magnetic reconnection, the free magnetic energy is released and converted into kinetic and thermal energies. 1 As a result of the re-arrangement of the magnetic field topology, a series of new loop-like magnetic structures called flare loops are often formed below the reconnection site. A hot diffuse region, consisting of plasma with a temperature of around 5-10 MK, is also observed above the loops and is called a supra-arcade fan. [2][3][4] Dark tadpole-like structures are often observed to descend through the bright supra-arcade fans. [5][6][7] Although their nature is debated, 5,6,[8][9][10] these so-called supra-arcade downflows (SADs) are generally believed to be related to the downward movement of either reconnection outflows or rapid contraction of newly formed magnetic loops during the intermittent and turbulent magnetic reconnection process in flares. SADs are possibly hotter and less dense than the surrounding plasma. 8,[11][12][13][14] They likely play an important role in shaping the plasma dynamics in supra-arcade fans; e.g., being involved in the generation of vortex shedding. 15 It is reported that SADs sometimes cause heating in the supra-arcade fans due to the compression of plasma. 16,17 However, these previously observed SADs generally do not reach the flare loops. Thus, it remains unclear what role SADs may play in heating the flare loops and the plasma in the immediate surrounding area.
In the past, SADs have been mostly observed above flare loops in the offlimb corona 18,19 when the reconnection current sheets face toward the observers. They were rarely observed on the solar disk, 20 possibly due to unfavorable geometry or reduced contrast with respect to the background emission.
On April 11, 2013, we observed an M6.5-class flare in National Oceanic and Atmospheric Administration (NOAA) active region (AR) 11,719 (Figure 1, Video S1) on the solar disk by the Atmospheric Imaging Assembly (AIA 21 ) on board the Solar Dynamics Observatory (SDO). Figure 1 shows images of the flare region at 7:50 UT as observed in different AIA filters. Most of these filters capture emission from plasma with a temperature of 0.6-3 million degrees, whereas the 131 Å and 94 Å filters are sensitive to hotter plasma with temperatures of 10 MK and 6 MK, respectively. Newly formed bright post-flare loops were observed in all of these six AIA filters. We also witnessed numerous downward-propagating SADs in the supra-arcade fan above the flare loops during the decay phase of the flare (mostly in the AIA 131 Å image sequence, some also seen in AIA 94 Å, Video S1). The favorable geometry and viewing angle of this event allowed us to observe the collision of many SADs with the flare loops and investigate the effect of collision in the local plasma heating.
To investigate possible plasma heating caused by the interactions, we focused on emission from the AIA hot channels (AIA 131 Å and AIA 94 Å) and we also performed a differential emission measure (DEM 22 ; Figure 2, Video S2) analysis using images taken in the six Fe-dominated AIA filters (discussed later).
Past observations showed that SADs mostly disappeared in supra-arcade fans before reaching flare loops. In contrast to previous cases, our on-disk observation shows that many SADs reach the apexes of flare loops and collide with them (Figure 3, Videos S1 and S3). When colliding with the loops, these SADs often reveal a sudden discontinuity in the downward propagation speed, from 250 km/s in the supra-arcade fan region to 15 km/s after the collision. Upward motion of hot plasma with a speed of 15 km/s can also be identified from some collision sites. This rebound motion may be driven by the tension force associated with the distorted magnetic field lines at the loop tops or could be associated with the backflow resulting from the impact of the reconnection downflow on the strong magnetic field region, as predicted by numerical simulations. 23 When these SADs collide with the loops, they often create propagating intensity fluctuations along the flare loops. These fluctuations reach the footpoints of the loops and lead to brightenings at the footpoint areas (Video S4). We cannot derive the propagation speed in all cases due to the limited cadence. In a few cases, we noticed a time delay of 24-60 s between the collision and footpoint brightening, which implies a propagation speed of 500-1,500 km/s.
Signatures of plasma heating have been found after the interactions, and Figure 4 shows one example (also see Video S5). We found that, when an SAD collides with the loops, the emission from the vicinity of the interaction site as well as the interacted loops suddenly decreases in all the six Fe-dominated AIA channels. A few minutes later, the intensity of AIA 131 Å starts to increase, followed by an increase of the 94 Å intensity and then the 335 Å intensity. The AIA 131 Å intensity peak is observed around 10-15 min after the collision. The intensity of AIA 94 Å peaks around 5-10 min after the peak of AIA 131 Å intensity. Similar behavior is observed both in the interacted loops (Figure 4J) and just above these loops (Figure 4I). The DEM-weighted mean temperature (Figures 2F, 4D, and4H) shows a strong increase both at the interacted loops and slightly above the apexes of the loops around 10-15 min after the collision, and the temperature increase lasts for about 10-15 min. The apexes of the loops, where most of the energy is released from the collision, appear to be hotter (15-20 MK; also see Figure 5) compared with the temperature of the heated loops (10 MK). Some earlier observations 24 also suggested the presence of such hot (10-20 MK) plasma at the loop tops during the flare decay phase, accompanied by impulsive microwave and X-ray bursts.
The Reuven Ramaty High Energy Solar Spectroscopic Imager (RHESSI 25 ) observation of our flare revealed the presence of enhanced X-ray emission at the interaction locations and along the interacted loops (Figure 5). The X-ray spectra observed by RHESSI appear to be thermal, and a spectral fitting reveals a temperature of 11-13 MK at the time of collision. Note that the integration time of RHESSI data is much longer (4 min) and the spatial resolution is 3 times lower compared with those of the AIA data. These differences may cause an underestimation in the temperature derived from the RHESSI data compared with the measurement from the highcadence and high-resolution AIA data.
The strong velocity discontinuity of SADs might indicate the presence of shocks at the loop tops (e.g., termination shocks 23,[26][27][28][29] ). In this case, sudden dissipation of the kinetic energy at the shock locations could be responsible for the strong localized heating around the loop tops. It is also possible that the significant heating results from a strong compression during the collisions. 16,17 Thermal conduction could then rapidly spread the heating to the rest of the loops. If the plasma is heated to a temperature of >20 MK, the emission from all Fe-dominated AIA channels should decrease because these channels are generally insensitive to such a high temperature. This may be the reason for the sudden decrease in the intensities of all the Fedominated AIA filters just after the collisions (Figure 4). We realize that the 193 Å filter has a response around 18 MK, which is much weaker than the filter's response around 1.5 MK. If the 18 MK plasma is much less than the 1.5 MK background coronal plasma in the line of sight, it will be very difficult to see an enhancement in the AIA 193 Å emission. Alternatively, the released energy could be immediately transported to electrons by thermal conduction (typical timescale << 1 s), which also leads to a non-equilibrium ionization condition 30 in the local plasma. During this period, the emission in all the Fe-dominated AIA filters may decrease due to the reduced efficiency of collisional excitation. Later, the ions could be heated through collisions with the slowly cooling hot electrons. Therefore, the spectral line emission from hot ions starts to increase after thermal relaxation between electrons and ions. After the AIA 131 Å intensity reaches its peak, the plasma starts to cool down, leading to intensity peaks in 94 Å and then in 335 Å. In addition, non-thermal particles may also play a role in the heating, particularly if the SADs are due to downward propagating plasmoids from the reconnection site.
We further analyzed the time variations of the AIA 131 Å intensity and DEM temperature (averaged over the whole region of Figure 6A) along with the soft X-ray fluxes measured by the Geostationary Operational Environmental Satellites (GOES) during the post-flare phase. These light curves all show clear quasi-periodic variations (Figures 6B and6C). We performed a wavelet analysis 31 for the time series of AIA 131 Å intensity and DEM temperature, and found a period of around 10 min for both of them (Figure S1). We also found that the GOES fluxes exhibit a similar behavior. Several peaks in the GOES data appear to occur at the same times as the AIA 131 Å intensity and DEM temperature peaks.
Quasi-periodic intensity variations have been frequently observed during many solar and stellar flares, [32][33][34][35][36][37] and they are often referred to as quasi- Report periodic pulsations (QPPs). These QPPs are often explained as being caused by magnetohydrodynamic (MHD) oscillations of the loops 23,38,39 or associated with repetitive/intermittent magnetic reconnection during flares. 40,41 Our detailed investigation shows that several peaks of the QPPs are clearly related to the interaction of SADs with the flare loops and the resultant heating. Three such peaks are marked by the three dashed lines (T1-T3) in the curves shown in Figures 6B and6C. The three corresponding heating signatures could be identified as local intensity enhancements in the AIA 131 Å images (Figures 6D-6F, Video S6). Hence, our observations have revealed an alternative scenario for the generation of QPPs in the decay phase of some flares. Recent numerical simulations found that plasmoid collisions with flare loops could lead to turbulent heating and QPPs above the loop tops. 42 Our observations appear to be consistent with this scenario, although it is unclear whether the observed SADs are plasmoids in the reconnection current sheet.
Mysterious dark tadpole-like downflows (SADs) are often observed at flare regions in the solar corona. Most of the previously reported SADs were observed above the solar limb. These observations showed that these Sun-ward-propagating SADs mostly disappear in the diffuse plasma region of flares before reaching the newly formed flare loops. Our observation of a flare that occurs on the solar disk reveals that many SADs not only reach the flare loops but also collide with these loops and heat the plasma to a temperature of about 10-20 MK. Since different SADs collide with the loops at different occasions, the plasma heating and resultant Extreme-UV (EUV)/Xray emission enhancement occurs quasi-periodically. Hence, our observation reveals that SADs play an important role in heating the coronal plasma in flare regions and also provides an alternative interpretation for the highly debated nature of frequently observed QPPs in solar and stellar flares.
We used data from the AIA instrument on board the SDO spacecraft, the GOES, and the RHESSI satellites.
The AIA instrument provides full-disk solar images of the Sun. In this study, we used images taken in the six Fe-dominated EUV passbands (94 Å, 131 Å, 171 Å, 193 Å, 211 Å, 335 Å) during the period of 07:40 UT to 09:20 UT on 2013 April 11. These passbands sample plasma with different temperatures in the solar atmosphere. AIA images are calibrated and co-aligned onto the same plate-scale using the aia_prep.pro routine in SolarSoftWare (SSW). The pixel size of these AIA images is 0.6 inches. Although each filter took images at a 12-s cadence, we used data with a 24-s cadence in our study since we found that many of the alternate frames are saturated at different locations.
The GOES satellite measured soft X-ray fluxes of the full Sun in two wavelength bands (0.5-4 Å and 1-8 Å) during the same time period. We also used the reconstructed RHESSI X-ray images in two energy bands (6-10 keV and 10-20 keV). The pixel size of the RHESSI images is 2.0 inches and the integration time selected for imaging is 2-4 min depending on the count rate. The GOES X-ray fluxes were obtained using the SSW software GOES Workbench. The RHESSI images were processed through SSW using subcollimators 3, 4, 5, 6, 7, 8. RHESSI spectroscopy was performed using subcollimator 3 with an integration time of 48 s. The X-ray spectral fit was performed using the SSW OSPEX package by adopting an isothermal model within an energy range of 6 keV-15 keV (above which the background becomes significant). The spectral fit of the X-ray source returns the plasma temperature T and a volume emission measure EM_V.
We applied a widely used inversion method 22 to AIA images taken in the six Fedominated passbands to compute the DEM curve at each spatial pixel and time step. DEM represents the amount of plasma as a function of temperature. The inversion method uses the temperature response functions of the six AIA channels and the measured count rates (in DN/s) in different channels. We have also applied a modified version of this method, 43 and obtained similar results. Hence, here we only present results from the former version. 22 The DEM curves in the flare region generally show a double-peak distribution, 44,45 consisting of a cold component peaking around log T/K z 6.2 and a hot component that peaks around log T/K z 7.0. The cold component is mainly from the emission of the regular coronal plasma (background emission), and the hot component arises from the heating due to magnetic energy release from the reconnection. It is proposed that the mean weighted temperature of the hot DEM component is a better representation of the temperature of the flaring plasma. 44,45 The DEM-weighted mean temperature is conventionally defined as follows, < T> h = P DEMðTÞ 3 TDT P DEMðTÞ 3 DT (Equation 1)
The integration was performed in the temperature range of 4-32 MK for investigating the dynamics of hot plasma. 44 As an example, Figure 2 shows images of the emission measure (EM) in different temperature bins as well as the calculated < T> h (also see Video S2). Here, EM refers to the integral of DEM(T) over a finite range of temperature.
The Innovation 2, 100083, February 28, 2021 www.cell.com/the-innovationThe Innovation