Multi-episode chromospheric evaporation observed in a solar flare
aa r X i v : . [ a s t r o - ph . S R ] J a n Multi-episode chromospheric evaporation observed in a solar flare
H. Tian , N.-H. Chen ABSTRACT
With observations of the Interface Region Imaging Spectrograph (IRIS), we study chromo-spheric heating and evaporation during an M1.6 flare SOL2015-03-12T11:50. At the flare rib-bons, the Mg ii iv iv and Mg ii lines at these occasions are accompanied byan obvious increase in the line intensities and the HXR flux, suggesting two episodes of energyinjection into the lower atmosphere in the form of nonthermal electrons. The Mg ii k/h ratiohas a small value of ∼ ∼ xxi ∼
300 km s − to nearly zero within ∼ xxi xxi emission is found around the northern ribbon in the decayphase, though no obvious response is detected in the Si iv and Mg ii emission. We have alsoexamined the Fe xxi emission at the flare loop top and identified a secondary component with a ∼
200 km s − red shift, which possibly results from the downward moving reconnection outflow.Our analysis also suggests a reference wavelength of 1354.0878 ± xxi line. Subject headings:
Sun: flares—Sun: chromosphere—Sun: transition region—line: profiles—magneticreconnection
1. Introduction
During solar flares, a large amount of energyis suddenly released through magnetic reconnec-tion, which may heat the coronal plasma and ac-celerate charged particles. The released energy isthen transported downwards to the lower atmo-sphere in the form of thermal conduction or non-thermal electrons, resulting in significant heatingof the chromosphere to temperatures up to tens ofMK. Such a process may lead to an overpressure inthe chromosphere, which drives the heated plasma School of Earth and Space Sciences, Peking University,100871 Beijing, China; [email protected] Korea Astronomy and Space Science Institute, Dae-jeon, Republic of Korea upward to fill flare loops. This so-called chro-mospheric evaporation may proceed explosivelyor gently (e.g., Fisher et al. 1985b; Gan & Mauas1994; Milligan et al. 2006a,b; Druett et al. 2017).In the case of explosive evaporation, the heatedplasma rapidly expands upward up to several hun-dred km s − and downward at a speed of few tenskm s − simultaneously. If the energy flux is lowerthan ∼ erg cm − s − (Fisher et al. 1985a),heating in the chromosphere often leads to gen-tle evaporation upflows in a wide range of tem-peratures at a speed of several tens km s − withno accompanied downflows. A signifciant fractionof the soft X-Ray (SXR) and extreme ultraviolet(EUV) emission during solar flares may be pro-duced through chromospheric evaporation.1hromospheric evaporation has been rarely im-aged from observations (Liu et al. 2006; Ning et al.2009; Nitta et al. 2012; Zhang & Ji 2013). In-stead, they are usually observed through spec-troscopic observations of SXR and EUV spectrallines. Prior to the launch of IRIS, spectroscopicobservations of coronal emission lines at flare rib-bons usually revealed blue-wing enhancement or ablueshifted component besides a nearly stationarycomponent (Doschek et al. 1980; Feldman et al.1980; Antonucci et al. 1982; Mason et al. 1986;Mariska & Doschek 1993; Czaykowska et al. 1999;Brosius 2003; Harra et al. 2005; Del Zanna et al.2006; Falchi et al. 2006; Milligan et al. 2006a,b;Teriaca et al. 2003, 2006; Raftery et al. 2009;Milligan & Dennis 2009; Watanabe et al. 2010;Del Zanna et al. 2011; Li & Ding 2011; Graham et al.2011; Young et al. 2013). Entirely blueshifted pro-files of these hot spectral lines were identified onlyin a few flares (Brosius 2013; Doschek et al. 2013).Such a result implies that normally the spatial res-olution of these spectrometers is too low to sepa-rate evaporation flows from the ambient plasma.The blue shift appears to increase with the lineformation temperature, and it can reach up to ∼
400 km s − at temperatures higher than 10 MK.In explosive evaporation, emission lines formed attypical transition region (TR) temperatures areoften found to be redshifted. The temperature atwhich the Doppler shift transits from red to blueis often found in the temperature range of 0.8–2MK (e.g., Milligan & Dennis 2009; Chen & Ding2010; Young et al. 2013), much higher than thatpredicted by most flare models (e.g., Fisher et al.1985a). However, Liu et al. (2009a) has demon-strated that the high transition temperature maybe explained by continuous energy depositionthroughout the impulsive phase. Strong red-shifts of lines formed at 1–2 MK may also beexpected when thermal conduction is suppressed(Imada et al. 2015).Launched around the solar maximum, theInterface Region Imaging Spectrograph (IRIS,De Pontieu et al. 2014) has observed hundredsof solar flares. With unprecedented high spa-tial ( ∼ ′′ ), spectral ( ∼ − ) and tem-poral (a few seconds) resolutions, IRIS obser-vations have provided significant insight intothe process of chromospheric evaporation. InIRIS observations, profiles of the Fe xxi ∼
11 MK)are usually found to be entirely blueshifted byup to ∼
350 km s − at flare ribbons (Tian et al.2014a, 2015; Young et al. 2015; Li et al. 2015;Brosius & Daw 2015; Polito et al. 2015, 2016;Sadykov et al. 2015, 2016; Graham & Cauzzi2015; Battaglia et al. 2015; Zhang et al. 2016b;Dud´ık et al. 2016; Lee et al. 2017). Such a re-sult is consistent with hydrodynamic simulationsof a single flare loop (e.g., Emslie & Alexander1987; Liu et al. 2009a) and suggests that the res-olution of IRIS is high enough to resolve theevaporation flows or flare kernels. The blueshift of the Fe xxi line peaks before the SXRpeak and correlates with the hard X-Ray (HXR)emission in several flares, providing strong sup-port to the scenario of chromospheric evapora-tion driven by nonthermal electrons (Tian et al.2015; Li et al. 2015b, 2017a). Through sit-and-stare observations, it has been found that the blueshift of Fe xxi line often smoothly decreases ex-ponentially from the maximum value to nearlyzero within a few minutes (e.g., Tian et al. 2015;Graham & Cauzzi 2015; Li et al. 2015b). How-ever, cool TR lines such as Si iv ∼ xxi line showed no quasi-periodic signatureor even no emission.Here we report results from IRIS sit-and-stareobservations of an M1.6 flare peaked around 11:50UT on 2015 March 12. Busty enhancement of theintensity and red shift can be clearly identified atthe flare ribbons for several chromospheric and TRlines. In addition, the Fe xxi line also reveals mul-tiple episodes of emission at one location on a rib-bon. In each episode the Fe xxi line experiences asmooth decrease of blue shift from 100–300 km s − to almost zero.
2. Observations and data reduction
This M1.6 flare peaked around 11:50 UT on2015 March 12 (SOL2015-03-12T11:50) in NOAAactive region (AR) 12297. IRIS performed a sit-and-stare observation of this flare, with a cadenceof ∼ ∼ ′′ , –190 ′′ ). Due to the spectral andspatial summing, the spatial pixel size is 0.33 ′′ ,and the spectral pixels are ∼ ∼ i i i i > –155 ′′ ). This assumption can be justi-fied since these cold lines usually have very smallintrinsic velocities in the quiet solar atmosphere.Images obtained in different SJI filters and spec-tral windows are coaligned by examining the posi-tions of the fiducial marks on the slit. Figure 1(A)-(C) and (E) show the IRIS SJI 1330 ˚A image andspectral images in three wavelength windows at11:42:51 UT. The time sequences of these imagesare presented in an online video. This dataset hasbeen previously analyzed by Tian et al. (2016) andBrannon (2016) to study global sausage oscilla-tions and draining downflows in the flare loops,respectively. The 335 ˚A images obtained by the AtmosphericImaging Assembly (AIA, Lemen et al. 2012) onboard the Solar Dynamics Observatory (SDO,Pesnell et al. 2012) are also used here to revealthe morphology of the flare loops (see Figure 1(D)and the online video). During flares this passbandsamples emission mainly from the Fe xvi ′ Dwyer et al. 2010). The formationtemperature of this line is about 3 MK. The ca-dence of the AIA 335 ˚A images is 12 seconds. Thecoalignment between AIA images and IRIS imageswere achieved by matching commonly observedfeatures in the AIA 1600 ˚A (mainly FUV contin-uum and C iv ) and IRIS 1330 ˚A (mainly FUV con-tinuum and C ii ) images. The AIA 335 ˚A imageswere then automatically aligned with the IRIS im-ages since AIA images in different passbands areautomatically coaligned after applying the stan-dard data reduction routine aia prep.pro, whichis available in SolarSoft (SSW). For AIA obser-vations in other passbands we refer to Tian et al.(2016).The Reuven Ramaty High Energy Solar Spec-troscopic Imager (RHESSI, Lin et al. 2002) was inorbit night during the occurrence of this flare. For-tunately, the Gamma-ray Burst Monitor (GBM)onboad the Fermi Gamma-ray Space Telescope ob-served clear enhancement of HXR in the energyrange of 26-50 keV during this flare.
3. Flare ribbons3.1. General behavior at the flare ribbons
We mainly analyze the following spectral linesat the flare ribbons: Si iv ii ii k 2796.35 ˚A and Mg ii h 2803.52 ˚A. Thestrong Si iv , Mg ii k and h lines are core lines ofIRIS, and have been demonstrated to be very sen-sitive to energy release during flares. Observationsof IRIS often reveal the presence of many emissionlines from neutrals or singly ionized ions at flareribbons, some of which can also be seen from Fig-ure 1. For the identification of these lines, we re-fer to Tian et al. (2015) and Tian (2017). Amongthem, the Mg ii triplet lines at 2791.59 ˚A, 2798.75˚A and 2798.82 ˚A are of particular interest due totheir sensitivity to heating of the lower chromo-sphere (Pereira et al. 2015). The latter two linesusually show up as one blended line in the IRISdata.3
220 -210 -200 -190Solar-X (arcsec)-180-170-160-150 S o l a r- Y ( a r cs e c ) S o l a r- Y ( a r cs e c ) (A) (B) (C)(D) (E) Fig. 1.— (A): An IRIS/SJI 1330 ˚A image taken at 11:42:51 UT on 2015 March 12. (B), (C) & (E): IRISdetector images taken at the same time in three spectral windows. (D) An SDO/AIA 335 ˚A image takenat 11:42:50 UT. In panels (A) and (D), the slit location is indicated by the white vertical line. A movie(m1.mp4) is available online. 4 ine intensity S o l a r- Y ( a r cs e c ) S o l a r- Y ( a r cs e c ) S o l a r- Y ( a r cs e c ) Integrated intensity S o l a r- Y ( a r cs e c ) Doppler shift [km/s] -30.0 0.0 30.0
Mg II k/h ratio
Fig. 2.— Temporal evolution of the line intensities (normalized to the maximum intensity in each image)and Doppler shifts (1st order moment) of the Si iv ii ii k lines, the intensityintegrated over the wavelength range of 1353.93 ˚A–1354.13 ˚A and the Mg ii k/h ratio along the slit. TheDoppler shift values of the Si iv line have been divided by three. The green and blue contours mark the flareribbons using intensity threshholds of Mg ii iv .00 1.05 1.10 1.15 1.20 1.25 1.30Mg k/h ratio3.03.23.43.63.84.04.24.4 M g II i n t en s i t y ( Log C oun t s ) M g II D opp l e r s h i ft ( k m / s ) M g II D opp l e r s h i ft ( k m / s ) M g II k i n t en s i t y ( Log C oun t s ) (A) (B)(C) (D) Fig. 3.— Scatter plots showing the relationship between different parameters within the blue contoursshown in Figure 2. 6he Si iv line often reveals strong enhancementof the red wing at flare ribbons. Although theMg ii k, h and triplet lines become single-peakemission features at flare ribbons, they are stilloptically thick (e.g., Kerr et al. 2015; Tian et al.2015; Graham & Cauzzi 2015; Lee et al. 2017). Soit might be inappropriate to apply a single Gaus-sian fit to these line profiles. Instead, we de-rive the line intensity and centroid by simply in-tegrating a spectral line profile and calculatingthe first order moment of the spectral line pro-file, respectively. The wavelength ranges usedfor this calculation are 1402.297 ˚A–1403.238 ˚A,2791.137 ˚A–2792.207 ˚A, 2795.109 ˚A–2797.655 ˚Aand 2802.238 ˚A–2804.784 ˚A for the Si iv ii ii k 2796.35 ˚A andMg ii h 2803.52 ˚A lines, respectively. The Dopplershift can then be derived by taking the differencebetween the centroid and the rest wavelength ofthe corresponding spectral line. A similar methodwas also adopted by Brosius & Daw (2015) andZhang et al. (2016a). In Figure 2 we present thetemporal evolution of the intensities and Dopplershifts of the Si iv , Mg ii ii klines, the intensity integrated in the wavelengthrange of 1353.93 ˚A–1354.13 ˚A (mainly Fe xxi ii k/h ratioalong the slit. The behavior of the Mg ii h line issimilar to that of Mg ii k, and thus is not presentedhere.Two flare ribbons clearly show up in the in-tensity images of the Si iv and Mg ii lines. Theribbons generally reveal red shifts of these lines.Both ribbons drift slightly along the slit duringthe course of the flare. The intensities and Dopplershifts of these lines appear to show quasi-periodicfluctuations, especially at the northern ribbon.The integrated intensity of 1353.93 ˚A–1354.13˚A is much stronger between the two ribbonsthan at the ribbons, indicating that the strongFe xxi emission mainly comes from the flare loopapexes and legs. At locations of the flare loops,we also see prominent emission from the Si iv andMg ii k lines, which appears to be associated withsignificant red shift. These large red shifts areassociated with the ”C”-shape spectral featurevisible in the strong chromospheric and TR lines(see the online video). They obviously result fromcooling of the heated plasma in the flare loops, asdiscussed by Brannon (2016). In the optically thin case, the Mg ii k/h ra-tio should be 2. Our observation shows that theMg ii k/h ratio is far less than 2 at the flare rib-bons. Figures 2(H) reveals a ratio of ∼ ii k/hratio at the two bright flare ribbons appear to bethe smallest, mostly in the range of 1.1–1.2.Within the ribbons, there also appears to be anegative correlation between Mg ii k/h ratio andMg ii ii ii k/h ratio. The intensities of theMg ii triplet lines and the two Mg ii resonant linesare positively correlated at the ribbons, which isevident from Figure 3(D). The southern ribbon was crossed by the IRISslit after 11:55 UT, only in the decay phase of theflare, while the northern ribbon was sampled bythe slit during all phases of the flare. Hence wefocus on the dynamics of the northern ribbon here.Figure 4 presents the temporal evolution of theline parameters and the Mg ii k/h ratio at thenorthern ribbon. The Si iv line is always red-shifted in both the impulsive and decay phases,and the average red shift is ∼
35 km s − . The redshift is enhanced quasi-periodically on a time scaleof 1–3 minutes. Before 11:58 UT each episode ofredshift enhancement is accompanied by an obvi-ous enhancement of the Si iv line intensity. Sucha correlation, also found by Zhang et al. (2016a)and Brosius & Daw (2015), vanishes after 11:58UT. The Doppler shifts of the Mg ii lines are muchsmaller compared to that of the Si iv line. De-spite that, the quasi-periodic enhancement is alsoclearly seen in the Doppler shift. The redshift en-hancement is most significant around 11:42 UTand 11:45 UT. These two episodes of redshift en-hancement are also accompanied by an obviousincrease in the intensities of the two Mg ii lines.Interestingly, the Mg ii k/h ratio decreases from ∼ ∼ -6 -5 -5 -5 GO ES f l u x ( W m - ) I n t en s i t y ( Log C oun t s ) Mg II kMg II 2791Si IV-505101520 D opp l e r s h i ft ( k m / s ) Mg II kMg II 2791Si IV /311:45 11:50 11:55 12:00 12:05 12:10 12:15Start Time (12-Mar-15 11:41:32)1.01.11.21.31.4 M g II k / h r a t i o Fig. 4.— Upper panel: GOES 1-8 ˚A flux (solid line) and its time derivative (dotted line). Note that thepeak after 12:00 UT is produced by another flare that is not studied here. Lower three panels: Temporalevolution of the intensities and Doppler shifts of Si iv ii ii k, and theMg ii k/h ratio, at the northern ribbon. The Doppler shift values of the Si iv line have been divided bythree. The vertical dotted lines indicate some instants when the red shifts of these lines are enhanced.8
200 00500100015002000 T i m e a ft e r : : ( s ) Fe XXI & C I 1354 -200 0 y+0.7" -200 0 y+1.7" Si IV 1403
Mg II 2791 -50 0 50 100
Mg II k -100 0 100 (A) (B) (C) (D) (E) (F)
Fig. 5.— Temporal evolution of the IRIS spectra in four spectral windows at the northern ribbon. (A),(D)-(F) Line profiles of Fe xxi iv ii ii k along the cyan lineshown in Figure 2(A). (B)-(C) Same as (A) but along two lines that are 0.7 ′′ and 1.7 ′′ above the cyan line,respectively. The black and red arrows indicate the times of 11:42 UT and 11:45 UT, respectively,9
50 0 50 100 150Doppler shift (km/s)200040006000800010000 C oun t s t=37 st=210 st=273 s -40 -20 0 20 40 60Doppler shift (km/s)50010001500200025003000 C oun t s -100 -50 0 50 100Doppler shift (km/s)2000400060008000 C oun t s -100 -50 0 50 100Doppler shift (km/s)2000400060008000 C oun t s (A) (B)(C) (D)Si IV 1402.77 Mg II 2791.59Mg II k Mg II h Fig. 6.— Spectral line profiles of Si iv ii ii k and h at three different times inthe northern ribbon. The times are shown in seconds after 11:41:32 UT (see Figure 5). The vertical dashedline in each panel indicates the rest wavelength of the corresponding line.10
300 -200 -100 0 100Doppler shift (km/s)0100200300400 T i m e a ft e r : : ( s ) -300 -200 -100 0 100Doppler shift (km/s)10100 C oun t s t=37 st=105 st=220 s (A) (B) Fig. 7.— (A) Temporal evolution of the IRIS spectra in the 1354 ˚A spectral window at the northern ribbon.The two white curves mark two sessions of exponential decrease of the Fe xxi blue shift. (B) The spectra atthree different times (shown in seconds after 11:41:32 UT). D opp l e r s h i ft ( k m / s ) GOES 1-8 AngstromFERMI 26-50 keVSi IV red shiftMg II red shiftFe XXI blue shift
Fig. 8.— Time history of the GOES 1-8 ˚A flux, FERMI 26–50 keV flux, Doppler shifts of Si iv ii xxi iv line profiles. The first two episodes of en-hanced red-wing asymmetry around t = 37 s and210 s, which correspond to 11:42 UT and 11:45UT, respectively, are most evident in the Si iv andMg ii ii iv and Mg ii iv line appears to be saturated. Slight red-wing enhancement is also observed in the Mg ii kand h lines. In the absence of significant heating,e.g., at t=273 s, the Si iv line is still enhanced atthe red wing. However, the three Mg ii lines allexhibit symmetric spectral profiles.The different behavior of the Si iv and threeMg ii lines suggests that we may use these lines todisentangle different physical processes in flares.The red shift of Si iv is usually believed to bea signature of chromospheric condensation dur-ing explosive chromospheric evaporation. How-ever, flare models usually predict a condensationtime scale of ∼ ii lines, which is not seen in our data. Wethink that the persistent red shift of Si iv at theflare ribbon is likely caused by both chromospheric condensation and cooling of the heated plasma(Tian et al. 2015). The Mg ii ii ii ii triplet lines in the diagnostics of heating pro-cess in flares. It is also worth noting that the red-wing asymmetry of the Si iv and three Mg ii linesis not consistent with most existing flare models,which usually predict entirely redshifted profilesof these low-temperature lines. Future modelingefforts similar to those of Ding & Fang (2001),Liu et al. (20015) and Rubio da Costa & Kleint(2017) are required to quantitatively evaluate howthe nonthermal electrons impact the Mg ii tripletline profiles.Quasi-periodic enhancement of Si iv line pa-rameters, especially the intensity and Dopplershift, has been previously detected by IRIS atthe ribbons of a few flares. The periods aremostly in the range of 1–6 minutes (Li et al.2015a; Li & Zhang 2015; Brosius & Daw 2015;Brosius et al. 2016; Brannon et al. 2015; Milligan et al.2017), though shorter periods of 32–42 s have alsobeen found by Zhang et al. (2016a) in a circular-ribbon flare. Similar time scales are found in ourobservation (Figure 4). In addition, we find thatthe Mg ii h, k and triplet lines also reveal a simi-lar quasi-periodic behavior. This quasi-periodicitylikely results from bursty reconnection, which mayrelease energy sporadically in the corona. The re-lease energy propagates downwards along flareloops quasi-periodically, most likely in the form ofaccelerated electrons (e.g., Nishizuka et al. 2009).The energy is then deposited in the chromospheresuccessively, leading to quasi-periodic enhance-ment of the Si iv and Mg ii line intensities andDoppler shifts.In previous observations of these multi-episodedynamics of cool lines in flares, the Fe xxi lineshowed no quasi-periodic signature or even no12mission, suggesting that the evaporation flows arenot heated to 11 MK or that the radiative cool-ing time of the Fe xxi emission is too long forany detectable fluctuation (Brosius et al. 2016).However, our observation demonstrates that thesecool line dynamics can be synchronized with thehot Fe xxi emission. Figure 5(A)-(C) clearly re-veals multiple episodes of chromospheric evapo-ration, exhibiting as multi-episode Fe xxi emis-sion. The first two episodes are clearly associatedwith the two brightenings crossing the slit at 11:42UT (around t = 37 s) and 11:45 UT (around t =210 s), respectively. And they are accompaniedby the strong chromospheric condensation as re-vealed by the significant red shifts of the Si iv andMg ii lines. There appears to be no Fe xxi emis-sion until the sudden enhancement of the FUVand NUV continua at about 11:42 UT. In the firstepisode, the Fe xxi line is entirely blueshifted.An examples of this type of line profiles is pre-sented in Figure 7 (t=105 s). The blue shift ofFe xxi smoothly decreases to nearly zero within ∼ ii xxi line profilesappear to reveal two components, one nearly atrest and the other blueshifted. This could be ex-plained as a superposition of the newly evaporatedhot plasma on the flare loop that is filled withhot materials through the first episode of evap-oration in the line of sight direction. However,an unambiguous decomposition of these two pos-sible components is difficult due to the blend ofmany lines such as Fe ii ii ii ∼
200 km s − , suggesting the presenceof highly blueshifted Fe xxi emission. Since theemission of Fe xxi line is mostly very weak andblended with many other lines (e.g., Tian et al.2015; Young et al. 2015), a reliable Gaussian fit-ting can not be preformed for many of the line pro-files. As demonstrated in Tian et al. (2015), thetime evolution of the Fe xxi blue shift often can be well fitted with an exponential function. Thus, wesimply overplot two exponential functions whichfit the observed blueshifted Fe xxi emission inFigure 7(A). These two exponential functions areused to approximate the Fe xxi blue shift as afunction of time in these two episodes. Figure 5reveals more episodes of Fe xxi emission, espe-cially at locations slightly offset from the north-ern ribbon. However, the Fe xxi emission in theseepisodes is generally much weaker and not accom-panied by significant enhancement of the Si iv andMg ii emission, which may indicate the occurrenceof gentle chromospheric evaporation.Figure 8 shows the time history of the GOES 1-8 ˚A flux, FERMI 26–50 keV flux, Doppler shifts ofSi iv ii xxi iv and Mg ii lines around 11:42 UT and 11:45UT. Complete evolution of chromospheric evap-oration and the simultaneous condensation werepreviously reported for a few flares (Brosius 2003;Brosius & Phillips 2004; Brosius 2003; Tian et al.2015; Graham & Cauzzi 2015; Li et al. 2015b).However, only one episode of evaporation wasobserved in most of these observations. Usingdata taken by the Extreme-ultraviolet ImagingSpectrometer (e.g., Culhane et al. 2007) on boardthe Hinode mission, Brosius (2013) detected twoepisodes of chromospheric evaporation in theFe xxiii iv ∼ iv line pair (5.83 – 2.37, corresponding toa density range of log N e /cm − =10 – 10 ) andsuggests that the electron density is likely higher13han 10 cm − . We noticed that a similar highdensity was obtained by Lee et al. (2017) for an-other flare ribbon.
4. Flare loop top
Besides flare ribbons, the Fe xxi xxi ′′ ,-162.9 ′′ ])and during the period of 12:08:00–12:15:00 UT.This period is in the decay phase of the flarewe study here. Note that the second SXR peakin the GOES light curve shown in Figure 9(A)was produced by a different flare. The averageline profile is shown in Figure 9(B). Besides theFe xxi line, the C i i xxi line. The line width(1/e width) is 52.8 km s − . After subtractingthe small instrumental broadening of ∼ − (see the discussion in Section S5 of the Supple-mentary Materials of Tian et al. 2014b), the linebroadening is essentially only thermal. In thissituation we may safely assume that the flareloops have a zero Doppler shift on average. Thus,the derived line centroid will be the rest wave-length of the Fe xxi line, which turns out to be1354.0878 ± ± i i ′′ ,-141 ′′ ]). The derived Doppler shifts forthese two cold lines are found to be redshifted by ∼ − , which likely reflects the uncertainty inthe wavelength calibration. Figure 9 also reveals an interesting feature ofFe xxi emission at the loop top: a secondarycomponent that is redshifted by ∼
200 km s − .High-speed downflows above the flare loop topwere spectroscopically observed in only a fewflares and they were often interpreted as downwardmoving reconnection outflows (Wang et al. 2007;Hara et al. 2011; Tian et al. 2014a; Sim˜oes et al.2015). The highly redshifted emission componentof Fe xxi was detected after the impulsive phase,roughly in the time period of 11:48–12:00 UT. Itis known that additional energy release by recon-nection may continue in the decay phase of a flare.Thus, the observed secondary emission componentmight still come from the reconnection downflow.The density and/or temperature effect may ex-plain the fact that this emission component wasnot observed in the impulsive phase.
5. Summary
We have analyzed the IRIS observation of anM1.6 flare SOL2015-03-12T11:50. With a ca-dence of ∼ iv and Mg ii lines in both the impulsivephase and decay phase. The red shifts show quasi-periodic enhancement likely caused repeated chro-mospheric condensation in a scenario of bursty re-connection, especially at the northern ribbon. Be-fore 11:58 UT each episode of redshift enhance-ment is also accompanied by an obvious enhance-ment of the Si iv line intensity.(2) The persistent and periodically enhancedred wing asymmetry of the Si iv ii =[−167.9",−162.9"] −200 −100 0 100 200 300Doppler shift (km/s)05001000150020002500 T i m e a ft e r : : ( s ) −200 −100 0 100 200 300110 C oun t r a t e centroid: 1354.08 Å1/e width: 52.8 km/s Fe XXI C I O I −200 −100 0 100 200 300Doppler shift (km/s)110100 C oun t r a t e (A) (B)(C) Fig. 9.— (A) Temporal evolution of the IRIS spectra in the 1354 ˚A spectral window at the loop top (averagedover solar-Y=[-167.9 ′′ ,-162.9 ′′ ]). (B) The line profile averaged in the time range of 12:08:00–12:15:00 UT(between the two yellow horizontal bars marked in (A)). The dashed lines represent a three-componentGaussian fit to the observed line profiles. The centroid position and exponential width of the Fe xxi line aremarked in this panel. (C) The line profile averaged in the time range of 11:50:32–11:57:31 UT (between thetwo white horizontal bars marked in (A)). The arrows indicate a redshifted component of the Fe xxi line.15ng fronts were observed to cross the slit in the SJI1330 ˚A images. The greatly enhanced red wings ofthe Si iv and Mg ii lines are also accompanied byan obvious increase in the intensities of these linesand the HXR flux, suggesting multiple episodes ofenergy injection into the chromosphere in the formnonthermal electrons.(4) The Mg ii k/h ratio has a small value of ∼ ii tripletline or resonant line. The ratio decreases to ∼ iv and Mg ii lines are also accom-panied by two episodes of chromospheric evapora-tion. In the first episode, the Fe xxi xxi line reveals a nearly stationary componentand a blueshifted component, which may resultfrom a superposition of the newly evaporated hotplasma on the flare loop that is filled with hot ma-terials through the first episode of evaporation inthe line of sight direction. In both episodes, theblue shift smoothly decreases from ∼
300 km s − to nearly zero within ∼ xxi emissionwas found around the northern ribbon. However,the Fe xxi emission in these episodes is generallymuch weaker and not accompanied by significantenhancement of the Si iv and Mg ii emission, pos-sibly indicating the occurrence of gentle chromo-spheric evaporation.(7) By assuming that the hot flare loops arestationary on average, we have derived a ref-erence wavelength of the Fe xxi line, which is1354.0878 ± xxi line revealsa secondary emission component that is redshiftedby ∼
200 km s − , possibly indicating the downwardmoving reconnection outflow.IRIS is a NASA Small Explorer mission devel-oped and operated by LMSAL with mission op-erations executed at NASA Ames Research cen-ter and major contributions to downlink com-munications funded by ESA and the NorwegianSpace Center. This work is supported by NSFCgrants 11790304 and 11790300. N.C. is supportedby the ”Development of a Solar Coronagraph on International Space Station” from the numberof 2017185100. H.T. acknowledges support byISSI/ISSI-BJ to the teams ”Diagnosing heatingmechanisms in solar flares through spectroscopicobservations of flare ribbons” and ”Pulsations insolar flares: matching observations and models”.We thank Dr. P. R. Young and Prof. M.-D. Dingfor helpful discussion. REFERENCES
Antonucci, E., Gabriel, A. H., Acton, L. W., et al.1982, SoPh, 78, 107Battaglia, M., Kleint, L., Krucker, S., Graham, D.2015, ApJ, 813, 113Brannon, S. R., Longcope, D. W., & Qiu, J. 2015,ApJ, 810, 4Brannon, S. R. 2016, ApJ, 833, 101Brosius, J. W. 2003, ApJ, 586, 1417Brosius, J. W., & Phillips, K. J. H. 2004, ApJ,613, 580Brosius, J. W. 2013, ApJ, 762, 133Brosius, J. W., Daw, A. N. 2015, ApJ, 810, 45Brosius, J. W., Daw, A. N., Inglis, A. R. 2016,ApJ, 830, 101Chen, F., Ding, M. D. 2010, ApJ, 724, 640Cheng, X., Ding, M. D. 2016, ApJL, 823, L4Culhane, J. L., Harra, L. K., James, A. M., et al.2007, SoPh, 243, 19Czaykowska, A., et al. 1999, ApJ, 521, L75Del Zanna, G., Berlicki, A., Schmieder, B., & Ma-son, H. E. 2006, SoPh, 234, 95Del Zanna, G., O ′ Dwyer, B., Mason, H. E. 2011,A&A, 535, A46De Pontieu, B., et al. 2014, Sol. Phys., 289, 2733Ding, M. D., Fang, C. 2001, MNRAS, 326, 943Doschek, G. A., Feldman, U., Kreplin, R. W., &Cohen, L. 1980, ApJ, 239, 725Doschek, G. A., Warren, H. P., Young, P. R. 2013,ApJ, 767, 5516ruett, M., et al. 2017, Nature Communications,8, 15905Dud´ık, J., Polito, V., Janvier, M., et al. 2016, ApJ,823, 41Emslie, G. A., Alexander, D. 1987, Sol. Phys. 110,295Falchi, A., Teriaca, L., Maltagliati, L. 2006, Sol.Phys. 239, 193Feldman, U., Doschek, G. A., Kreplin, R. W., &Mariska, J. T. 1980, ApJ, 241, 1175Fisher, G. H., Canfield, R. C., McClymont, A. N.1985a, ApJ, 289, 414Fisher, G. H., Canfield, R. C., McClymont, A. N.1985b, ApJ, 289, 425Fisher, G. H. 1989, ApJ, 346, 1019Gan, W. Q., Mauas, P. J. D. 1994, A&A, 430, 891Graham, D. R., Fletcher, L., & Hannah, I. G.2011, A&A, 532, A27Graham, D. R., Cauzzi, G. 2015, ApJL, 807, L22Hara, H., Watanabe, T., Harra, L. K. 2011, ApJ,741, 107Harra, L. K., D´emoulin, P., Mandrini, C. H., etal. 2005, A&A, 438, 1099Imada, S., Murakami, I., Watanabe, T. 2015,Physics of plasmas, 22, 101206Kerr, G. S., Sim˜oes, P. J. A., Qiu, J., Fletcher, L.2015, A&A, 582, A50Landi, E., Young, P. R., Dere, K. P., Del Zanna,G., & Mason, H. E. 2013, ApJ, 763, 86Lemen, J. R., et al. 2012, Solar Phys., 275, 17Lee, K.-S., Imada, S., Watanabe, K., Bamba, Y.,Brooks, D. H. 2017, ApJ, 836, 150Li, D., Ning, Z. J., Zhang, Q. M. 2015a, ApJ, 807,72Li, D., Ning, Z. J., Zhang, Q. M. 2015b, ApJ, 813,59Li, D., Ning, Z. J., Huang, Y., Zhang, Q. M. 2017a,ApJL, 841, L9 Li, D., Ning, Z. J., Huang, Y., et al. 2017b, ApJ,849, 113Li, Y., Ding, M. D. 2011, ApJ, 727, 98Li, Y., Ding, M. D., Qiu, J., Cheng, J. X. 2015,ApJ, 811, 7Li, Y., Kelly, M., Ding, M. D., et al. 2017, ApJ,848, 118Li, T., Zhang, J. 2015, ApJL, 804, L8Lin, R. P., et al. 2002, Sol. Phys., 210, 3Liu, W.-J., Qiu, J., Longcope, D. W., Caspi, A.2013, ApJ, 770, 111Liu, W., et al. 2006, ApJ, 649, 1124Liu, W., Petrosian, V., Mariska, J. T. 2009a, ApJ,702, 1553Liu, W.-J., Heinzel, P., Kleint, L., Kasparova, J.,2015, Sol. Phys., 290, 3525Mason, H. E., Shine, R. A., Gurman, J. B., Har-rison, R. A. 1986, ApJ, 309, 435Mariska, J. T., & Doschek, G. A. 1993, ApJ, 419,418Milligan,R.O., Gallagher, P. T., Mathioudakis,M.,& Keenan, F. P. 2006a, ApJL, 642, L169Milligan, R. O., Gallagher, P. T., Mathioudakis,M., et al. 2006b, ApJL, 638, L117Milligan, R. O., Dennis, B. R. 2009, ApJ, 699, 968Milligan, R. O., et al. 2017, ApJL, in pressNing, Z.-J., Cao, W.-D., Huang, J. 2009, ApJ, 699,15Nishizuka, N., Asai, A., Takasaki, H., et al. 2009,ApJ, 694, L74Nitta, S., Imada, S., Yamamoto, T. T. 2012, Sol.Phys., 276, 183O ′ Dwyer, B., Del Zanna, G., Mason, H. E., Weber,M. A., & Tripathi, D. 2010, A&A, 521, A21Pereira, T. M. D., Carlsson, M., De Pontieu, B.,& Hansteen, V., 2015, ApJ, 806, 1417olito, V., Reeves, K. K., Del Zanna, G., Golub,L., Mason, H. E. 2015, ApJ, 803, 84Polito, V., Reep, J. W., Reeves, K. K., et al. 2016,ApJ, 816, 89Raftery, C. L., et al. 2009, A&A, 494, 1127Pesnell, W. D., Thompson, B. J., Chamberlin, P.C. 2012, Sol. Phys., 275, 3Reep, J. W., Warren, H. P., Crump, N. A., Sim˜oes,P. J. A. 2016, ApJ, 827, 145Rubio da Costa, F., Liu, W., Petrosian, V., &Carlsson, M. 2015, ApJ, 813, 133Rubio da Costa, F., Kleint, L. 2017, ApJ, 842, 82Sadykov, V. M., Dominguez, S. V., Kosovichev,A. G., et al. 2015, ApJ, 805, 167Sadykov, V. M., et al. 2016, ApJ, 828, 4Sim˜oes, P. J. A., Graham, D. R., Fletcher, L. 2015,A&A, 577, A68Teriaca, L., Falchi, A., Cauzzi, G., et al. 2003,ApJ, 588, 596Teriaca, L., Falchi, A., Falciani, R., Cauzzi, G., &Maltagliati, L. 2006, A&A, 455, 1123Tian, H., Li, G., Reeves, K. K., et al. 2014a, ApJL,797, L14Tian, H., DeLuca, E. E., Cranmer, S. R. et al.2014b, Science, 346, 1255711Tian, H., Young, P. R., Reeves, K. K., et al. 2015,ApJ, 811, 139Tian, H., Young, P. R., Reeves, K. K., et al. 2016,ApJL, 823, L16Tian, H. 2017, RAA, 17, 110Wang, T. J., Sui, L. H. & Qiu, J. 2007, ApJ, 661,L207Warren, H. P., Reep, J. W., Crump, N. A., Sim˜oes,P. J. A. 2016, ApJ, 829, 35Watanabe, T., Hara, H., Sterling, A. C., & Harra,L. K. 2010, ApJ, 719, 213Young, P. R., Doschek, G. A., Warren, H. P.,Hara, H. 2013, ApJ, 766, 127 Young, P., Tian, H., Jaeggli, S. 2015, ApJ, 799,218Zhang, Q.-M., Ji, H.-S. 2013, A&A, 557, L5Zhang, Q.-M., Li, D., Ning, Z. J. 2016a, ApJ, 832,65Zhang, Q.-M., Li, D., Ning, Z. J., et al. 2016b,ApJ, 827, 27Zhou, G. P., Zhang, J., Wang, J. X. 2016, ApJL,823, L19