The Evolution of Plasma Composition During a Solar Flare
Andy S.H. To, David M. Long, Deborah Baker, David H. Brooks, Lidia van Driel-Gesztelyi, J. Martin Laming, Gherardo Valori
DDraft version February 22, 2021
Typeset using L A TEX twocolumn style in AASTeX63
The Evolution of Plasma Composition During a Solar Flare
Andy S.H. To , David M. Long , Deborah Baker , David H. Brooks , Lidia van Driel-Gesztelyi ,
1, 3, 4
J. Martin Laming , and Gherardo Valori University College London, Mullard Space Science Laboratory, Holmbury St. Mary, Dorking, Surrey, RH5 6NT, UK College of Science, George Mason University, 4400 University Drive, Fairfax, VA 22030, USA LESIA, Observatoire de Paris, Université PSL, CNRS, Sorbonne Université, Univ. Paris Diderot, Sorbonne Paris Cité, 5 place JulesJanssen, 92195 Meudon, France Konkoly Observatory, Research Centre for Astronomy and Earth Sciences, Hungarian Academy of Sciences, Konkoly Thege út 15-17.,H-1121, Budapest, Hungary Space Science Division, Naval Research Laboratory, Code 7684, Washington, DC 20375, USA (Received November 30, 2020; Revised February 16, 2021; Accepted February 19, 2021)
Submitted to ApJABSTRACTWe analyse the coronal elemental abundances during a small flare using Hinode/EIS observations.Compared to the pre-flare elemental abundances, we observed a strong increase in coronal abundanceof Ca
XIV <
10 eV), as quantified bythe ratio Ca/Ar during the flare. This is in contrast to the unchanged abundance ratio observed usingSi X X Keywords:
Sun: abundances - Sun: corona - Sun: magnetic fields INTRODUCTIONComposition of plasma in the solar corona is a tracerof the flow of plasma and energy from the solar interior.Different complex processes such as the propagation andabsorption of waves, convection of hot plasma and re-connection and reconfiguration of magnetic fields canaffect the flow and composition of plasma. This pro-duces a clear and observable variation in the elementalabundances of coronal plasma across different regions ofthe solar atmosphere (e.g., Brooks et al. 2015).In order to parameterise and study the coronal ele-mental abundances, we use the first ionisation poten-tial (FIP) bias, defined as taking the ratio of an ele-ment’s coronal to photospheric abundance with respectto H. The FIP bias varies from solar structure to solarstructure, and is closely linked to the Sun’s magneticfield on all scales. Observed changes in elemental abun- dances take place on time scales of years (solar cycle),days or hours (magnetic flux emergence and decay), orminutes (flare) (Brooks et al. 2017; Baker et al. 2018;Warren 2014). In a quintessential active region, low-FIP elements (FIP < eV) such as Ca, Mg and Siexhibit enhanced abundances, while high-FIP elements(FIP ≥ eV) such as Ar, O and S retain their pho-tospheric abundances. This elemental fractionation isknown as the FIP effect. The FIP bias of the photo-sphere is typically measured to be 1, with higher FIPbias values of 3-4 found in the closed loops of a devel-oped active region (cf. Del Zanna & Mason 2014; Bakeret al. 2013, 2015, 2018). The FIP bias value of quiet-Sun (Warren 1999; Baker et al. 2013; Ko et al. 2016)and coronal hole regions (Feldman et al. 1998; Brooks& Warren 2011; Baker et al. 2013) has been exploredextensively, and has also been used to link solar wind a r X i v : . [ a s t r o - ph . S R ] F e b To et al. back to its source regions (Gloeckler & Geiss 1989; Fuet al. 2017)In addition to the small magnetic perturbations thatare common in a typical active region, rapid changes inmagnetic connectivity can also contribute to a changein coronal composition. One of the events that causesrapid coronal composition changes is a solar flare. Solarflares are spectacular results of solar magnetic recon-nection, characterised by the sudden release of energyand plasma. This rapid change of magnetic configura-tion triggers waves and energy that propagate into thechromosphere (Fletcher & Hudson 2008), followed bythe ablation of plasma into the corona (Warren 2014).This produces the variability in coronal emission intensi-ties observed by spectrometers. The standard and staticcoronal composition picture relies on a selective mecha-nism, generating a coronal composition that is differentfrom the photospheric one. On the other hand, abla-tion does not act on specific elements. Warren (2014)studied 21 M and X class flares using Solar DynamicObservatory/EUV Variability Experiment (
SDO /EVE)showed that ablation acted on all elements and turnedthe Sun-as-a-star coronal elemental abundances tem-porarily closer to photospheric, in agreement with earlierresults of Veck & Parkinson (1981); Feldman & Wid-ing (1990); McKenzie & Feldman (1992) and Del Zanna& Woods (2013). However, an explanation that onlycomprises of ablation falls short on addressing someobservations that obtained an enhanced low-FIP ele-mental abundances (e.g. Doschek et al. 1985; Sterlinget al. 1993; Bentley et al. 1997; Fludra & Schmelz 1999;Phillips et al. 2010; Phillips & Dennis 2012; Dennis et al.2015; Sylwester et al. 2015), as well as the more recentobservations that have found evidence of an inverse FIP(IFIP) effect during the decay phase of flares occurringin very complex active regions. In such cases, the IFIPsignatures were observed with timescales of around 30minutes (Doschek et al. 2015a; Doschek & Warren 2016;Baker et al. 2019, 2020).So far, the only theoretical model that is able to ex-plain both the FIP and IFIP effects is the ponderomotiveforce fractionation model, proposed by Laming (2004,2009, 2011, 2015). In this model, FIP effect fractiona-tion is caused by the reflection and refraction of Alfvénwaves in the upper chromosphere, especially in regionswith high density gradients. The change in wave di-rection exerts a back reaction on the plasma ions, theponderomotive force, and depending on the origin andnature of the waves this can vary in magnitude andsign. The direction of the ponderomotive force leadsto ions that are guided in different directions, causingthe FIP/IFIP effect. In this paper, we present Hinode/Extreme-ultravioletImaging Spectrometer (
Hinode /EIS) observations takenduring a solar flare which originated in a reconnectingX-shaped structure rooted in a complex active region.EIS catching the flare in action provides an insight intothe rapid evolution of solar coronal composition. Theobservations are presented in Section 2, with results anddiscussion in Section 3 and Section 4. Conclusions arethen discussed in Section 5. OBSERVATIONS AND DATA ANALYSISAR 11967 was an old and complex active region with arich history. It was visible on the Southern hemisphereof the Sun from 31 January 2014 to 7 February 2014.The active region was in its second rotation on disk,and major flux emergence and reconnection could be ob-served during this rotation. Figure 1 shows the magneticcomplexity of AR 11967 from the Solar Dynamics Obser-vatory/Helioseismic and Magnetic Imager (
SDO /HMI;Hoeksema et al. 2014) line-of-sight magnetic field data.Its complex and active nature led to numerous eruptions,resulting in 28 M-class flares and 83 C-class flares.AR 11967 was particularly interesting as it hosted avery stable, long lived X-shaped structure (previouslyidentified and discussed by Jiang et al. 2017; Kawabataet al. 2017; Liu et al. 2016; Xue et al. 2017; Yang et al.2017). This structure was observed from 2-5 Feb 2014,and was located between the sunspots S1 and S3 shownin the HMI magnetogram of Figure 1.2.1.
Coronal EUV and Magnetic Field Observations
The active region could be identified and studied onits evolution across the disk using the full-disk Extreme-Ultraviolet (EUV) and line of sight magnetogram imagesfrom the Atmospheric Imaging Assembly (AIA; Lemenet al. 2012) and the HMI instruments on board the
SolarDynamics Observatory (SDO; Pesnell et al. 2012) space-craft. Three different passbands from AIA were used inthis analysis; 94 Å, 193 Å and 131 Å. Images from the193 Å passband are especially useful for FIP bias anal-ysis as they capture emission from elements at aroundthe same temperatures as one of the FIP bias pairs, Si X (258.38 Å) and S X (264.23 Å), formed at ∼ Figure 1. a) HMI continuum; HMI magnetogram; AIA 193 Å; AIA 131 Å image during the flare at 14:34 UT on 2 Feb 2014.The white box indicates the EIS field of view in Figure 3. b) AIA 94 Å light curve obtained using a 34 (cid:48)(cid:48) × (cid:48)(cid:48) box surroundingthe X-shaped structure in AR 11967. Blue horizontal lines indicate EIS rastering time and duration; Red vertical lines indicatethe flare time; Orange horizontal line in the zoomed section of the light curve indicates the EIS scan time over the X-shapedstructure. The third flare happened during the third EIS raster, and EIS observed the X-shaped structure during its decayphase, roughly 1 minute after the peak of the flare. plate scales, roll angles between AIA and HMI and cor-rectly aligns the two instruments. Both 193 Å and 131 Åbroadband images were also sharpened using the Multi-scale Gaussian Normalisation technique (MGN; Morgan& Druckmüller 2014).2.2. EUV Spectroscopic Observations
The elemental abundance results discussed herewere derived using spectroscopic observations fromthe Extreme-ultraviolet Imaging Spectrometer (EIS;Culhane et al. 2007) onboard the
Hinode space-craft (Kosugi et al. 2007). Hinode/EIS observedAR 11967 on 2 Feb 2014, making three observa-tions using two active region studies, study acronym
To et al.
HPW021_VEL_240x512v1 and Atlas_30 (Study num-ber 437 and 403 respectively). Table 1 shows the keydetails of the studies that were used to track the evolu-tion of AR 11967. Study 437 has a large field of viewof 240 (cid:48)(cid:48) × (cid:48)(cid:48) . It uses the slit scanning mode withthe 1 (cid:48)(cid:48) slit and 1 (cid:48)(cid:48) scan step size, with a long exposuretime of 60 s at each step for a total rastering time of2 hours. In contrast, study 403 is a full CCD study,which has a smaller field of view of 120 (cid:48)(cid:48) × (cid:48)(cid:48) , usinga 2 (cid:48)(cid:48) slit with a 2 (cid:48)(cid:48) step. At each pointing position, EISexposed for 30 s to produce a raster with a durationof 30 minutes. Similar to the SDO data, the EIS datawere preprocessed using the standard eis_prep.pro rou-tine available in SolarSoftWare; this accounts for darkcurrent, CCD bias, cosmic rays, hot, warm, dusty pixels,the radiometric calibration, and orbital correction. Theeis_ccd_offset.pro routine was then used to ensure spa-tial consistency between different EIS spectral windows.The two pairs of low-FIP/high-FIP elements presentedhere are close in wavelength, however the Fe lines usedfor Differential Emission Measure (DEM) span both EISCCDs, so the Del Zanna (2013) Hinode /EIS calibrationwas used to calibrate the EIS data. In order to min-imise the offset between
Hinode /EIS and
SDO , the in-tensity map of Fe
XII
XII
Composition Maps
In our FIP bias examination, two pairs of emis-sion lines, Si X X XIV
XIV Si X /S X Composition Map
The Si X (258.38 Å, FIP = 8.15 eV) and S X (264.23 Å,FIP = 10.36 eV) line pair has a comparably lower for-mation temperature at around 1.25-1.5 MK. The emis-sion from the consecutive ionisation stages of Fe VIII –Fe
XVI were fitted with a single Gaussian function, withthe exception of Fe XI , Fe XII and Fe
XIII fitted withmultiple Gaussians to obtain their intensities. The ef-fects of density variation were estimated using the fittedFe
XIII X line intensity. Since the Fe lines lie on the short-wavelength detector and theSi and S lines lie on the long-wavelength detector, theemission measure was scaled to reproduce the intensityof Si X XIII den-sity calculated above. Lastly, both Si X X X X X X ¯ χ > were discarded during our analy-sis. Since EIS is formed by both a short-wavelength andlong-wavelength detectors with an offset of 18.5 pixelsin the y direction, part of our region of interest that liesat the top of the long-wavelength detector was not cap-tured by the short wavelength detector. The lack of datafor the Si X X ¯ χ value, and this resulted in the patch of missingdata shown in Figure 3. The estimated uncertainty ofthe Si X X Ca XIV /Ar
XIV
Composition Map
The Ca
XIV (193.87 Å, FIP = 6.11 eV) and Ar
XIV (194.40 Å, FIP = 15.76 eV) emission lines are formedat a comparably higher ionisation equilibrium tempera-ture of around 3.5 MK (log T = 6.55) (Feldman et al.2009). Both lines are relatively strong and present simi-lar emissivity temperature dependence. Three Gaussianfunctions were fitted to both the Ca XIV
XIV
XIV
XIV is blended with two very faint lines along its blue wing(Brown et al. 2008). In this paper, we follow the analy-sis done by Doschek et al. (2015b); Doschek & Warren(2016, 2017); Baker et al. (2019, 2020) in producing thecomposition maps. In particular, Doschek & Warren(2017) describes the assumptions and issues that comewith this composition diagnostic. We used log abun-dance values relative to log H as; coronal-Ca = 6.93(Feldman 1992), photospheric-Ca = 6.33 (Lodders et al.2009) and coronal and photospheric-Ar = 6.50 (Lodderset al. 2009). Since the spectral intensity of Ar at thephotosphere could not be directly measured, its abun-dances value (photospheric-Ar) is determined indirectlythrough spectra observed in solar wind, solar flares or so-lar energetic particles. As a result, historically, there hasbeen a relatively large fluctuation of the photospheric-Ar abundances. Figure 2 shows a plot of the ratio ofCa XIV
XIV
XIV
XIV XV XIV ∼ log T = 6 . . The range of electron tempera-tures during the flare was also narrow, and ranges be-tween log . − log . . When these two pieces ofevidence are put side by side with the significant inten-sity ratio changes observed across the Ca/Ar maps, thetemperature and density can be seen to have a mini-mal effect for our analysis. The contribution functionsthat are convolved with the DEM at the flaring regionis shown in the Appendix.2.4. Region Definition
In Figure 3, we can observe a significant increasein the Ca/Ar intensity ratio value in our third raster(panel f) corresponding to the stable structure, whichclearly describes its X shaped morphology. Since thisX-shaped structure is extremely stable throughout theperiod studied here on 2 Feb 2014, we defined the flaringregion using pixels with an intensity ratio value > Study Number 437Study Acronym HPW021_VEL_240x512v1Fe
VIII
VIII IX IX X XI XII
XII
XIII
XIII
XIV
XIV XV XVI
XVII X X XIV
XIV (cid:48)(cid:48) × (cid:48)(cid:48) Rastering 1 (cid:48)(cid:48) slit, 120 positions, 2 (cid:48)(cid:48) coarse stepsExposure Time 60 sTotal Raster Time 2 hoursReference Spectral Window Fe
XII X X XIV
XIV (cid:48)(cid:48) × (cid:48)(cid:48) Rastering 2 (cid:48)(cid:48) slit, 60 positions, 2 (cid:48)(cid:48) stepsExposure Time 30 sTotal Raster Time 30 minutesReference Spectral Window Fe
XII
Table 1.
Hinode /EIS study details used in this study. of interest were then differentially rotated to the times ofthe two previous rasters on 2 Feb 2014, at 10:25 UT and12:29 UT. The white border in each panel of Figure 3shows the extent of this region of interest differentiallyrotated to the corresponding raster time. In each casethere is good agreement, indicating the stability of theX-shaped structure and the robust nature of the methodused to define the region of interest. RESULTSAlthough the active region was observed by EIS be-tween 1 and 5 Feb 2014, and the X-shaped structurewas apparent between 2 and 4 Feb, here we focus onits evolution between 9:30 UT and 14:50 UT on 2 Feb.The bottom panel of Figure 1 shows the averaged in-tensity in the 94 Å passband in the field of view shownby the white box in Figure 1a. Multiple brighteningscan be identified, with the 3 blue horizontal lines in-dicating the periods of the EIS rasters. The first flarewas the strongest, starting at 09:26 UT, ∼ ∼ ∼ To et al.
Figure 2.
Left to right: a) Intensity ratio of Ca
XIV
XIV log N = 9 . , . , . ; b) Intensity ratio of Ca XV XIV the third EIS raster described here. The flare startedat 14:27 UT, approximately 9 minutes into the raster,peaked at 14:37 UT and decayed to background inten-sity level by ∼ Magnetic Field Evolution
AR 11967 was a highly complex AR built up by atleast three major bipoles in various stages of their evolu-tion. Figure 1 shows both S1 and S2, which were leading(positive) polarity spots produced by two flux emergenceepisodes before the AR rotated onto the disk. We cansee that the negative S5 is located tightly below S1, butin fact S1 and S2’s following (negative) polarities hadalready dispersed by this time. Despite this, S1 wasstill a large spot exceeding 100 MSH according to theDebrecen Photoheliographic Data. A major flux emer-gence occurred in the AR during the EIS observationperiod, with its principal spots being S4 (positive) andS3 (negative) in Figure 1. The parallel strands of oppo-site polarities in between S4 and S3 indicate that thisemerging bipole is strongly sheared. The magnetic po-larity pattern (magnetic tongues, cf. Luoni et al. 2011)is indicative of negative, left-handed twist. As S4 andS3 are separating in the course of flux emergence, thenegative polarity S3 is moving towards the positive po-larity S1. As the flare observed with EIS took place inbetween spots S3 and S1 (cf. Figure 1), a forcing by theirpersistent approach was creating strong field gradientsin the proximity of the X-shaped structure (Jiang et al.2017; Kawabata et al. 2017).3.2.
Enhanced Low FIP Elemental Abundances of theThird Flare
Figure 3 shows the Si X /S X composition map andthe Ca XIV /Ar
XIV intensity ratio map from the three EIS rasters made on 2 Feb 2014. EIS made two rasterobservations after the first flare, with the second flareoccurring before, and the third flare during, the thirdEIS raster. It can be seen in the first row of Figure 3that, across the three observations, in a temperaturerange ∼ X /S X composition mapdoes not show drastic changes, and the map is generallyred/dark orange in colour, comparable to a compositionmap of a quiet sun region or a small active region.However, at a higher temperature (corresponding tothe Ca XIV /Ar
XIV ratio maps), a significant enhance-ment in FIP bias can be observed during the third flare.For the first two rasters, the region of the solar flareshows up as red in the intensity ratio map, indicatingan ordinary quiet sun FIP bias value of 2, similar tothe FIP bias value calculated using the Si X /S X com-position map. However, a clear change in coronal abun-dances can be seen in the third raster, where EIS ob-serves the third flare (as a yellow region, indicating en-hanced FIP) which peaked at 14:29 UT. The shape ofthe enhanced intensity ratio patch fully traces the topol-ogy of the X-shaped flare structure.These changes in the maps are reflected in the datahistograms in Figure 4 taken around the flaring re-gion. The Si X /S X histogram shows little change be-tween the three observations. On the other hand, theCa XIV /Ar
XIV ratio value histogram shows a drasticenhancement in the mean FIP bias value, which changesfrom a quiet sun value of ∼ DISCUSSION AND INTERPRETATIONIn this study, we have analysed the instantaneouscoronal plasma evolution of a small flare in a recon-necting X-shaped structure located in the highly active
Figure 3.
Top to bottom: Hinode/EIS Si X X XIV
XIV
XIV
XIV X X X X Hinode /EIS. Offset between the twodetectors contributed to the patch of dark pixels in the composition maps; The Ca
XIV
XIV
To et al.
Figure 4.
Top to bottom: Histogram of a) Si X X XIV
XIV
AR 11967 observed by
Hinode /EIS on 2 Feb 2014. Thisopened up a possibility of obtaining composition duringa solar flare. Over the course of the three observations,a low FIP bias value of ∼ Physical Interpretation
We suggest two possible scenarios that could help toexplain the evolution of the composition, in addition tothe ablation picture we expect after flares. First, thefact that S as a high-FIP element has a relatively lowFIP value which is close to 10 eV. Secondly, the potential shift of elemental fractionation height due to the strongmagnetic fields observed here.The model to explain the observed fractionation of el-emental abundances in the solar atmosphere is a quasi-static model of long-lasting structures such as coronalloops, active regions and solar wind outflow regions (cf.Laming 2004; Laming et al. 2019). Under these staticscenarios, in closed loops, the fractionation occurs atthe top of the chromosphere. Warren (2014) studied theevolution of plasma composition in large flares which in-volves a dynamic rapid evolution of plasma parameters(highly time dependent), and suggested that in somedramatic time-dependent events like large M and X classflares, the observed elemental abundance variation wasdue to plasma ablated from deep in the chromosphere,below where fractionation occurs. However, this is notthe case here, where an enhanced abundance of Ca, orin other studies, where an enhancement of other variouslow-FIP elements (e.g., Fludra & Schmelz 1999; Phillipset al. 2010; Phillips & Dennis 2012; Dennis et al. 2015)was observed. Therefore, we have to look for other mech-anisms to find an explanation.4.1.1.
Partial Ionisation of Different Elements
Firstly, the different composition evolution in the Si/Sand Ca/Ar line pairs could be due to the actual FIPvalues of these elements. We have adopted the conven-tion of categorising Si and Ca as low FIP elements andS and Ar as high FIP elements. While S is typicallycategorised as a high FIP element, it has a relativelylow FIP value of 10.36 eV, whereas Si, typically cate-gorised as a low FIP element, has a comparable FIPvalue of 8.15 eV. The difference between their FIP val-ues is therefore merely 2.21 eV. In contrast, as Ar is anoble gas, it has a much higher FIP of 15.76 eV, withthe result that the difference in FIP between it and Ca(FIP = 6.11 eV) is 9.65 eV.During a flare, magnetic reconnection produces Alfvénwaves that travel from the corona to the chromosphere(e.g. Fletcher & Hudson 2008). The refraction of suchwaves at the top of the chromosphere can then create anupward ponderomotive force that acts only on ions (cf.Laming 2009). However, the same reconnection processalso accelerates and heats particles, which at their im-pact onto the chromosphere lead to heating and conse-quent ablation of plasma. For fractionation to occur, theAlfvén speed must be greater than the electron thermalspeed, so that following coronal energy release, Alfvénwaves get to the chromosphere first to cause fractiona-tion before the heat conduction flux arrives to cause theevaporation. If this is violated, unfractionated plasmawill be evaporated into the flare loops. However, if thetwo speeds were about equal, fractionation with extraionisation of S would result. For the flare studied here,which was very small, the heating produced could havebeen sufficient to ionise S with its relatively low FIP,with the ponderomotive force bringing up both Si and Sin tandem, thus maintaining the quiet-sun FIP bias val-ues that can be seen across the Si/S composition mapsand their respective data histograms. On the contrary,due to the relatively high FIP of Ar, the small heatingassociated with the observed small flare was insufficientto ionise Ar. Instead, the more abundant low FIP Caions were brought up by the ponderomotive force, thusgenerating the fractionation observed in the third Ca/Arintensity ratio map.Within the above interpretation, this observationcould open up a unique way to probe into the fraction-ation height in the solar atmosphere by different solarphenomena. With the partial ionisation of S but notAr, one can estimate the upper chromospheric tempera-ture at the time of the flare. Using Laming et al. (2019),the chromospheric model by Avrett & Loeser (2008) andcollisional ionisation equalibrium calculation from Maz-zotta et al. (1998), the chromospheric temperature needsto be . × K < T < . × K . (1) The true temperature is likely to be lower, because cal-culations from Mazzotta et al. (1998) neglect photo-ionisation and density effects. Nonetheless, this pos-tulate provides an additional insight into the work doneby Warren (2014) and Dennis et al. (2015), who stud-ied the composition evolution in large flares. For largerflares, more energy is released, which should produce acorrespondingly large increase in temperature, thus ion-ising Ar. This would then produce sufficient ionised Arto experience the ponderomotive force, resulting in thephotospheric-like coronal composition more commonlyobserved during flares.4.1.2. Fractionation in the Low Chromosphere
Another plausible scenario comes from the fact thatthe flare studied here took place between the strongmagnetic fields of two large sunspots. This has the effectof lowering the plasma β = 1 height and consequentlythe fractionation height of the different elements. Asidentified in Fletcher & Hudson (2008), a solar flarewhich triggers reconfiguration of the magnetic field gen-erates large-scale Alfvén waves which propagate fromthe flaring site to the lower chromosphere, where thebackground gas is neutral H. In this region, high FIPS behaves like a low FIP element (Laming et al. 2019);behaviour consistent with the statistically insignificantchanges of Si/S FIP bias observed here. This requires arelaxing of the quasi-static nature of FIP models, to al-low fractionation to occur throughout the chromosphereand not just at the top, which follows if the waves are inresonance with the loop. As with the scenario involvingpartial ionisation described above, this would result infractionation of Ca and both Si and S, producing theconstant Si/S and an increase in Ca/Ar observed here.These observations are consistent with the ponderomo-tive force interpretation for variation in FIP bias.4.2. AIA Wavelet Analysis
Although the two physical interpretations both seemplausible, they both rely on flare-driven downward trav-elling Alfvén waves which induce the ponderomotiveforce acting upwards. To try and identify this essentialsignature of Alfvén waves propagating to the chromo-sphere, the AIA data in different passbands were anal-ysed using a wavelet technique similar to that previouslyused by e.g., Milligan et al. (2017); Hayes et al. (2016).The intensity in the HMI continuum, 1600 Å, 1700 Å,304 Å, 131 Å and 94 Å passbands was averaged withinthe region indicated by the white box shown in Figure 1.The top left of Figure 5 shows the normalised waveletanalysis of the wavelengths, with the top right panelshowing the detrended data produced by subtracting0
To et al.
Figure 5.
Wavelet analysis of AIA 94 Å, 131 Å, 304 Å, 1600 Å, 1700 Å and HMI continuum during the third flare at 14:37 UT. a smoothed version of the data. This was then anal-ysed using the wavelet technique of Torrence & Compo(1998). It can be seen in panels c–h of Figure 5 thateach of the different passbands (with the exception ofthe HMI continuum) exhibits a variation in the signal as-sociated with the small-scale brightening discussed here.However, the signals are in the form of a single peak in-stead of an oscillation. Moreover, in terms of the signalstrength, apart from 304 Å, no other passbands show asignificant intensity. It is also notable that most of theobserved signal for each passband is below the white tri-angular lines, indicating that they are not statisticallysignificant and outside the zone of influence.This suggests that the Alfvén waves produced by theflare in the corona proposed by Fletcher & Hudson(2008) could not be observed in the lower solar atmo-sphere using the data available here. The lack of a sta- tistically sound signal that could support wave propa-gation is disappointing, but this result could still informseveral investigation directions for future work. Firstly,the instrument, AIA, used in the analysis has a rela-tively longer cadence of 12 s, which places a limit onthe frequency of the observable signal. Using an instru-ment with a higher cadence, such as the
Geostation-ary Operational Environmental Satellite’s
X-ray sensor(GOES/XRS) could show the finer scale oscillations, al-beit without the spatial resolution required here. Aback of the envelope calculation gives insight into thewavelet period we can look into in the future, usinga simple calculation from the coronal loop length and v A = B / √ µ ρ , where v A is the Alfvén speed. Inour case, taking from the observations and assuming asemi-circle coronal loop, the loop length can be esti-mated to be ∼ (cid:48)(cid:48) ; and the x-shaped structure is rooted1in an intense magnetic field with a line of sight fieldstrength of ∼ ∼ × − kgm − (Kohutova & Verwichte 2017),this gives the line of sight Alfvén speed, v Az , to be ∼ kms − , and a resonant wave period of 140 seconds.This value can then be compared with future waveletanalysis and help pinpoint the Alfvén waves.Secondly, AIA is a broadband imager, which capturesspectral lines across a wide range of ions. Although astrong signal can be found in the 304 Å wavelet analysis,which seems to suggest variations indeed exist in thechromosphere, the contributing spectral lines could notbe precisely identified. The use of co-temporal narrowband spectroscopic data could provide more insight intothe wave propagation depth, and perhaps open up thepossibility of correlating the oscillation strength with thethe observed FIP bias. CONCLUSIONSIn this paper we have presented observations of thehighly complex active region, AR 11967, which haveprovided a unique opportunity to study the evolution ofcomposition in a small flare that occurred between twolarge sunspots. On 2 Feb 2014, EIS observed a smallflare within an X-shaped structure in the active region.The results show very different composition evolutionat two different line pairs of different formation tem-peratures, with an unchanged composition obtained inthe lower temperature Si X /S X composition map in theflaring region, while a significant FIP bias value increasewas observed in the higher temperature Ca XIV /Ar
XIV composition ratio.We propose two possible physical interpretationswhich could explain or contribute to the strange com-position evolution. Firstly, in the case of partial ionisa-tion, due to the relatively low first ionisation potentialof S, both Si and S were ionised by the small third flarewith relatively low chromospheric heating, leaving theSi X /S X FIP bias value unchanged. While the lowerFIP of S meant that it could be ionised by the smallflare observed here, the much higher FIP value of Armeant that it could not be easily ionised. The resultingponderomotive force produced by the waves originat-ing from magnetic reconnection leading to the flare thusonly brought up the ionised Ca, Si and S in tandem.A similar yet non-mutually exclusive mechanism couldbe present in our second interpretation, fractionationat the lower chromosphere. Our flare occurred betweentwo sunspots, rooted in a region of strong magnetic field.This has the effect of lowering the regions of fractiona-tion of the different elements. Alfvén waves generatedby the flare travelled to the lower chromosphere, where the background gas is neutral H. Under this condition, Sbehaves like a low-FIP element. Therefore both Si andS fractionate in this flare and the Si/S ratio does notchange, while the high-FIP Ar does not change its be-haviour. This creates a significant discrepancy that canbe observed between the two sets of composition maps.Both interpretations of the evolution of composition dur-ing a small flare that we present here are consistent withthe ponderomotive force interpretation for variation inFIP bias.Since these two interpretations involve Alfvén wavesthat travel to the chromosphere, a wavelet analysis ap-proach applied to the co-temporal AIA data was usedto search for wave signatures. However, waves in thelower chromosphere during the reconnection were notdetected with this method. Nonetheless, this providesthe further investigation direction at correlating elemen-tal fractionation with wave oscillations. Firstly, the rel-atively long cadence of AIA (12 s) limits the observationof very high frequency signals. Secondly, AIA/
SDO isa broad band imager. Signals obtained from differentpassbands merely give a very arbitrary idea of of the so-lar altitude. The use of co-temporal narrow band spec-trometer will give more insight of the wave propagationdepth, and perhaps open up the possibility of correlat-ing the oscillation strength with the observed FIP bias.Some of the work to combine observations between dif-ferent layers of the solar atmosphere has been done inBaker et al. (2021), using the Interferometric BIdimen-sional Spectrometer (IBIS), EIS and magnetic field mod-elling. Ground-based instruments like IBIS and DanielK. Inouye Solar Telescope (DKIST) will be extremelyvaluable in the future, by directly observing where thewave refraction and reflections are proposed to happen.The upcoming Solar-C EUVST and its wide range oftemperature coverage can also contribute massively byobserving different layers of our Sun’s atmosphere si-multaneously. If the existence of Alfvén waves can beconfirmed, this observation could be the another stepto understand the physical mechanism behind composi-tion evolution during flares, and current theories needto factor the time evolution of composition into theirmodelling. APPENDIXIn this section, we address the concern of tempera-ture effect on the Ca
XIV /Ar
XIV intensity ratio value.In Figure 2, the electron temperature at the flaring re-gion ranges from log . − log . . This puts thetheoretical intensity ratio at the uphill region in panela of Figure 2, suggesting that a change in temperaturecould produce the observed change in the intensity ratio.2 To et al.
Figure 6.
Log normalised contribution functions, G(T) ofCa
XIV
XIV
A differential emission measure (DEM) was performedaround the region of interest. Over the temperaturerange, two curves coincide very well, and the peak tem-perature does not change significantly, suggesting thedramatic increase in intensity ratio value is real. ACKNOWLEDGMENTSA.S.H.T. thanks the STFC for support via fundinggiven in his PHD studentship. D.M.L. is grateful to theScience Technology and Facilities Council for the awardof an Ernest Rutherford Fellowship (ST/R003246/1).The work of D.H.B. was performed under contract tothe Naval Research Laboratory and was funded by theNASA Hinode program. D.B. is funded under STFCconsolidated grant number ST/S000240/1 and L.v.D.G.is partially funded under the same grant. L.v.D.G. ac-knowledges the Hungarian National Research, Devel-opment and Innovation Office grant OTKA K-113117.J.M.L. was supported by the NASA Heliophysics GuestInvestigator (80HQTR19T0029) and Supporting Re-search (80HQTR20T0076) programs, and by Basic Re-search Funds of the Office of Naval Research. G.V.acknowledges the support from the European Union’sHorizon 2020 research and innovation programme un-der grant agreement No 824135 and of the STFC grantnumber ST/T000317/1. Hinode is a Japanese missiondeveloped and launched by ISAS/JAXA, with NAOJ asdomestic partner and NASA and STFC (UK) as inter-national partners. It is operated by these agencies in co-operation with ESA and NSC (Norway). AIA data cour-tesy of NASA/SDO and the AIA, EVE, and HMI scienceteams. CHIANTI is a collaborative project involvingGeorge Mason University, the University of Michigan(USA) and the University of Cambridge (UK).REFERENCES