Evidence of chromospheric molecular hydrogen emission in a solar flare observed by the IRIS satellite
MMNRAS , 1–10 (2020) Preprint 5 February 2021 Compiled using MNRAS L A TEX style fi le v3.0 Evidence of chromospheric molecular hydrogen emission in a solar fl areobserved by the IRIS satellite Sargam M. Mulay, ★ Lyndsay Fletcher , † School of Physics & Astronomy, University of Glasgow, G12 8QQ, Glasgow, UK Rosseland Centre for Solar Physics, University of Oslo, P.O.Box 1029 Blindern, NO-0315 Oslo, Norway
Accepted XXX. Received YYY; in original form ZZZ
ABSTRACT
We have carried out the fi rst comprehensive investigation of enhanced line emission from molecular hydrogen, H at 1333.79 Å,observed at fl are ribbons in SOL2014-04-18T13:03. The cool H emission is known to be fl uorescently excited by Si �� emission was observed when theSi �� fl are impulsive phase and gradual decay phase, but it dimmed during the GOESpeak. H line broadening showed non-thermal speeds in the range 7-18 km s − , possibly corresponding to turbulent plasma fl ows. Small red (blue) shifts, up to 1.8 (4.9) km s − were measured. The intensity ratio of Si �� �� fi rmed that plasma was optically thin to Si �� (where the ratio = 2) during the impulsive phase of the fl are in locations wherestrong H emission was observed. In contrast, the ratio di ff ers from optically thin value of 2 in parts of ribbons, indicating a rolefor opacity e ff ects. A strong spatial and temporal correlation between H and Si �� emission was evident supporting the notionthat fl uorescent excitation is responsible. Key words:
Sun: atmosphere – Sun: activity – Sun: chromosphere – Sun: fl ares – Sun: transition region – Sun: UV radiation Solar fl ares can have an impact on all layers of the solar atmosphere.There are strong signatures from the mid-upper chromosphere, tran-sition region and corona, but enhanced emission corresponding toexcitation of the chromospheric temperature minimum region (TMR)and possibly the photosphere is also detected. The mechanisms ofexcitation so deep in the atmosphere are not clear, as it seems highlyunlikely that fl are-accelerated electrons can penetrate there. How-ever, depending on optical conditions, high-energy photons can. Inthis paper, we present observations made during a fl are of enhancedline emission from molecular hydrogen, H which has a formationtemperature of 4200 K (Innes 2008) corresponding to the TMR. H line emission is thought to be formed by photo-excitation ( fl uores-cence) by ultraviolet (UV) radiation from the transition region, anda recent theoretical study by Jaeggli et al. (2018) put the location ina narrow range around 650 km above the photosphere for a range oftemperature strati fi cations and radiation conditions, including corre-sponding to a fl are. H emission thus gives a new view of conditionsin the TMR during a fl are.The formation of molecular spectra is more complex than atomicspectra. Every electronic state has multiple vibrational and rotational(sub-)states of di ff erent energies, and so excitation or de-excitationbetween electronic states can be between any of these vibrational orrotational states allowed by quantum-mechanical selection rules. Theelectronic excitation from the ground state to the fi rst (Lyman band) ★ E-mail: [email protected] † E-mail: [email protected] or second (Werner band) electronic excited state of H molecule oc-curs due to absorption of far-UV photons. Many excited vibrationallevels of the upper electronic state may become populated duringthe electron excitation process, since there are no selection rules onthe vibrational transitions. The de-excitation to the electronic groundstate (with a time scale of 10 − sec) occurs by emitting the far-UV emission lines ( fl uorescence) in Lyman or Werner bands of H .There are many levels in the ground state with a signi fi cant popula-tion, which gives many more options for fl uorescence. More detailedinformation about the H lines and their UV exciter wavelengths aregiven in Table 1 and in Appendix.H emission (in P and R branches that corresponds to rotationalquantum number, Δ J = -1 and Δ J = +1 respectively) in solar UVspectra in the range 1175-1714 Å was fi rst reported by Jordan et al.(1977, 1978) in observations of a sunspot umbra from the fi rst rocket fl ight of the Naval Research Laboratory’s High Resolution Telescopeand Spectrograph (HRTS). Most of the lines observed belonged tothe Lyman band of H (Herzberg & Howe 1959; Abgrall et al. 1993a)which are fl uoresced by H Lyman 𝛼 red wing photons, and by strongtransition region lines, C �� , Si �� and O �� . The authors identi fi edtwo groups of lines within the Lyman band of H corresponding totwo groups of transitions; the fi rst with vibrational quantum number, v � = 1 (upper level) and v �� = n; 2 < n < v � = 0 and v �� = 4, 5 and 6.They identi fi ed excitation by photons in the red wing of the broadand intense Lyman 𝛼 transition region line emitted in the region ofa spot or pore as responsible for the high fl uorescent intensity ofthe H lines. H fl uorescence in the Lyman band due to transitionregion lines O �� , O � , C �� , C ��� , C �� , and Si �� was also observed © 2020 The Authors Mulay, S. M. et al.
Table 1.
Details of H emission lines observed by IRIS in C �� and Si �� windowsColumn 1 Column 2 Column 3 Column 4 Column 5 Column 6 Column 7 Column 8H Transition Branch Exciting line Observed solar Instruments FWHM References 𝜆 (Å) ( v � - v �� ) ( Δ J = ± 𝜆 (Å) regions (Å)1333.475 0-4 R0 Si �� �� �� �� �� �� �� �� � �� Notes -
More details about IRIS H lines can be found at https://pyoung.org/iris/ Figure 1.
Left panel: (a) GOES X-ray M7.3 class fl are observed on April 18, 2014, and (b) derivative of X-ray fl uxes. The orange and black curves show X-ray fl uxes in the 0.5 - 4.0 Å and 1.0 - 8.0 Å channels of the GOES-15 respectively. The blue dashed lines indicate the time slot (12:42 to 12:52 UT) during the risephase of the fl are where the physical parameters for H line were measured. Panel (c): AIA 1600 Å image of the active region (AR) 12036. The northern andsouthern ribbons are shown by black arrows. The black boxed region indicate the IRIS SJI fi eld-of-view (FOV). Panel (d): The SJI image in the Si �� fl are. The vertical black line shows the IRIS spectrograph slit position. The slit captured emission from the southernribbon and di ff erent parts of it are named as Ribbon 1 (‘R1’) and Ribbon 2 (‘R2’). in a sunspot light bridge region during the second fl ight of HRTS(Bartoe et al. 1979).H emission in the Werner band (H lines in Q branch that corre-sponds to Δ J = 0, Abgrall et al. 1993b) was found for the fi rst timein the solar atmosphere using the fi rst fl ight HRTS data (Bartoe et al.1979). This H emission in a sunspot (transitions in the v � - v �� = 1-5and 1-6) was fl uoresced by the O �� resonance line. The authors pro-vided a list of wavelengths in transitions (in the v � - v �� = 1-0 to 1-7bands) which were later found to be in good agreement with the lines observed in a sunspot (Schüehle et al. 1999) recorded by the SolarUltraviolet Measurements of Emitted Radiation (SUMER; Wilhelmet al. 1995) instrument. H emission has also been observed in thequiet sun (Sandlin et al. 1986) and by Innes (2008) in active regionplage associated with the footpoints of X-ray micro fl ares, near thefootpoint of a brightening X-ray loop and at the location of strongtransition region out fl ow.Previously unidenti fi ed lines in Skylab fl are observations by Cohenet al. (1978) were identi fi ed by Bartoe et al. (1979) as H UV lines
MNRAS , 1–10 (2020) vidence of molecular hydrogen emission in a fl are fl uoresced by O �� , C �� , C �� and Si �� . Bartoe et al. (1979) reportedthat these H lines decreased rapidly in intensity with time, presum-ably as the line intensity and width of the exciting transition regionline decreased. The spectra were recorded at the beginning of the fl are gradual phase, and the spectrograph slit reportedly did not crossthe fl are ribbon (Bartoe et al. 1979). In contrast, the observations wereport here cover impulsive and gradual phases, and the fl are ribbonsare in the fi eld of view.With data from the Interface Region Imaging Spectrograph (IRIS;De Pontieu et al. 2014), we can examine carefully the spatial and tem-poral evolution of H emission, in much more detail. IRIS observesa number of molecular H lines which mostly come from the Lymanband (Herzberg & Howe 1959; Sandlin et al. 1986). H emissionlines in IRIS spectra from the fl aring chromosphere were identi fi edby Young et al. (2015) and Li et al. (2016); H lines are also reportedin absorption, as features in Si �� spectral lines (Schmit et al. 2014).These are interpreted as due to pockets of cool (photospheric tem-perature) plasma, in which molecules can form, in the upper solaratmosphere above a source of Si �� emission.Table 1 displays H lines in the IRIS range, in the C �� and bothSi �� spectral windows. Column 1 indicates the wavelength of theH lines, Column 2 speci fi es the transitions from lower vibrationalstates ( v �� ) to higher vibrational states ( v � ), whereas Column 3 showscorresponding transitions in the R ( Δ J = +1) and P ( Δ J = -1) branches.Column 4 indicates the transition region lines which excite the H emission as identi fi ed in the references given in Column 8. A possiblealternative excitation route for the upper level of H at 1333.797 Åwhich is analysed in this paper, involving Si �� and C �� lines, isdiscussed in Appendix A. In Columns 5-8, we list the solar regions,instruments, and full-width-half-maximum (FWHM) of the line fromthe literature, if the H line has previously been reported. A moreextensive list of H emission lines in the UV part of the spectrumbetween ∼ �� fl are. The IRIS fl are observation started at12:33:38 UT, about 2.5 minutes after the start time (12:31 UT) of theGOES X-ray fl are. The IRIS slit was well-positioned over the fl areribbon. The event has also been studied by Brannon et al. (2015),Brosius & Daw (2015), Cheng et al. (2015) and Brosius et al. (2016).Brannon et al. (2015) and Brosius et al. (2016) focused on coherentquasi-periodic pulsations in IRIS and EIS data, seen in both fl areribbons during the impulsive rise, fi nding intensity pulsations (inIRIS and EIS) and velocity pulsations (in IRIS), with di ff erent peri-ods at di ff erent locations in the ribbon. In the IRIS observations (thesame raster study we analyse here), a sawtooth pattern of Dopplershifts with an average period of ∼
140 s and oscillation speed of ∼
20 km s − was measured using Si �� , with similar behaviour andamplitudes seen in O �� and C �� lines (Brannon et al. 2015). Theselines have di ff erent formation temperatures, so the similar velocitypatterns led the authors to suggest that the transition region and up-per chromosphere is compressed during the fl are, resulting in theselines all originating from a very narrow range of heights undergoingessentially the same behaviours. Brosius et al. (2016) reported anoscillation period of 75.6 ± �� , Mg �� ,Mg ��� , Si ��� , Fe ��� , and Fe ��� ). All these lines were red-shifted,whereas a couple of components of Fe ����� line pro fi le were highlyblue-shifted, indicating explosive chromospheric evaporation.In this paper, we focus on the H fl are ribbon emission, producedby much cooler plasma, along with its exciting Si �� lines. The paperis organised as follows. In Section 2, we provide details about the observational study and data analysis. We discuss and summarise theresults in Section 3. The Geostationary Operational Environment Satellite (GOES-15)recorded X-ray fl uxes for the M7.3 fl are in two channels, 1-8 Å and0.5-4 Å. The GOES fl are started at 12:31 UT, peaked at 13:03 UTand ended at 13:20 UT. The X-ray fl uxes and their derivatives (whichare - via the Neupert e ff ect (Neupert 1968; Hudson 1991; Dennis &Zarro 1993) - a proxy for the hard X-ray fl ux) are shown in panels(a) and (b) of Fig. 1 respectively. The fl are originated from NOAAactive region 12036 (S15 W42 ).The UV signatures of the fl are were observed as two bright fl areribbons located between two sunspots of con fi guration 𝛽𝛾 . We iden-tify the ribbons as northern ribbon and southern ribbon. These areseen in the UV images in the 1600 Å channel (see panel (c) of Fig. 1)from the Atmospheric Imaging Assembly instrument (AIA; Lemenet al. 2012) on board the Solar Dynamic Observatory (SDO; Pesnellet al. 2012), at a resolution of 0.6 �� per pixel and 12 sec cadence. TheAIA data was obtained from the Virtual Solar Observatory (VSO)and prepared using the standard AIA package aia_prep.pro avail-able in the Solarsoft libraries (SSW; Freeland & Handy 1998). Theblack box in panel (c) of Fig. 1 shows the IRIS slit-jaw imager (SJI) fi eld-of-view (FOV) overlaid on an AIA 1600 Å image. The entiresouthern ribbon and a small part of the northern ribbon were capturedby the SJI.A joint IRIS-Hinode Operation Plan (IHOP241) observational se-quence was run on April 18, 2014, between 12:33 and 17:18 UT. TheIRIS slit was well-positioned to cross southern ribbon and emissionspectra in Si �� , O �� , C �� , Mg �� and H lines were taken by the IRISspectrograph (SG) in sit-and-stare mode, remaining stationary withrespect to the solar surface at the slit position shown by the blackvertical line in panel (d) of Fig. 1. Spectra were captured from thesouthern ribbon at around 200 slit positions. We identify di ff erentparts of this ribbon as Ribbon 1 (‘R1’) and Ribbon 2 (‘R2’). Theevolution of this ribbon was captured by the SJI (in C �� �� �� .Using the iris_orbitvar_corr_l2s.pro routine, the data werecorrected for orbital variation (both the thermal component and thespacecraft velocity component). The dust particles on the SJI CCDproduce black dots/patches in the images. They were removed us-ing the iris_dustbuster.pro routine and the cosmic rays werealso removed using the despik.pro routine. We used the strongphotospheric O � , and both Si �� lines are shown in Fig. 2. The panels (a-c)show detector images for di ff erent IRIS spectral windows. A numberof lines observed in these windows are indicated. The white dashedlines indicate the pixels along the slit which were used to obtainthe example averaged spectra, shown in panels (d-f) of Fig. 2. H ftp://ftp.swpc.noaa.gov/pub/warehouse/2014/ https://sdac.virtualsolar.org/cgi/search https://iris.lmsal.com/data.html MNRAS , 1–10 (2020) Mulay, S. M. et al.
Figure 2.
The detector images (panels (a-c)) and spectra (panels (d-f)) were obtained for slit position number 118 that shows emission at 12:51:57 UT. Thespectra were obtained by averaging pixels between 650 and 660 along the slit. The white dashed lines indicate emission in these pixels. The blue dashed linesindicate Gaussian components used for fi tting the lines and the entire fi t is shown by solid red line. The horizontal cyan lines indicate a fi t for backgroundemission. Figure 3.
The IRIS spectral images for H (panels (a) and (d)), Si �� �� fl ares ribbons (same regions that are shown in panel (d) of Fig. 1). Panel (d-f): spectral images at single wavelength values of H at 1333.79 Å, Si �� at1393.76 Å and Si �� at 1402.77 Å. The dark vertical lines at 12:43, 13:09 and 13:17 UT indicate that the IRIS spectra is missing at those slit locations/timings. at 1333.47 Å was very weak throughout the fl are evolution exceptfor slit position number 118 where the line was strong. We fi ttedboth H lines with a single Gaussian. The Gaussian components andthe entire fi tted lines are shown by blue dashed and solid red linesrespectively.Based on sunspot, coronal hole, and quiet sun spectra obtainedfrom the SUMER spectrograph, Curdt, W. et al. (2001) observedthat the S � line at 1333.80 Å is very close to H at 1333.797 Å. In addition, Li et al. (2016) also mentioned that there is a possibleblend based on IRIS fl are observation. In order to identify a blend,the unblended S � line at 1401.51 Å could be taken as a reference. Thebehaviour (in intensity, velocity, and width) of this S � and the possiblyblended H line could be tested for evidence of correlation that wouldindicate an important contribution of S � to the line pro fi le. A detailedanalysis has been carried out, as discussed in Appendix B, and weobserved that the behaviour of the S � ff erent than MNRAS , 1–10 (2020) vidence of molecular hydrogen emission in a fl are Figure 4.
Panels (a-b): Light curves for R1 and R2. The total intensities (DNs) are obtained for the pixels where H (between 1333.76 and 1333.87 Å) and Si �� (1402.5 and 1403.6 Å) emission is presented. The background emission (total intensity at slit 21, 12:36:46 UT, X-pix = 21 and Y-pix = 635-650) was subtractedfrom the total intensities before plotting. The negative intensities are removed from the data which result in discontinuities in the light curves. Panels (c-d):Scatter plots for R1 and R2 emission seen in H (between 1333.76 and 1333.87 Å) and Si �� (1402.5 and 1403.6 Å) lines. The intensities are displayed with starsymbols, and solid blue lines indicate the linear fi t to the data. The equations for the fi tted lines along with fi t parameters are given and the Pearson correlationcoe ffi cients are displayed as ‘R’. Table 2.
IRIS observation details of a fl areIRIS Spectrograph (SG) Slit-Jaw-Imager (SJI)Observation date 18-April-2014Observation ID 3820259153IHOP ∗ �� (slit width) 0.33 �� Field-of-view 0.166 �� × �� �� × �� Exposure time (sec) 7.9 1.98Cadence (sec) 9.4 28 (for C �� , Si �� , Mg �� ) ∗ that of H . Hence, we conclude that the S � line at 1333.80 Å is notblended with H line at 1333.79 Å.During the evolution of the fl are, the southern ribbon was observedto move southwards in SJI images, and this displacement was nicely observed in the spectral images. The emission from R1 as it movessouth is shown in the time stackplots of spectral images in Fig. 3.R1 and R2 are indicated by white arrows. R2 has a less well-de fi nedmotion. The images shown in panels (a)-(c) of Fig. 3 are created bysumming DNs over the wavelength ranges 1333.76-1333.87 Å forH , and 1393.5-1394.6 Å and 1402.5-1403.6 Å for the Si �� lines. �� By selecting particular wavelengths, we can examine in some detailthe ribbon evolution and correlations between the exciter wavelengthand the fl uorescent emission. The fi rst thing to notice, in panels (a-c)of Fig. 3, is that the H emission becomes visible when the Si �� fl uorescentexcitation is responsible. The alternative explanation for enhancedH emission, that the number of H molecules has increased, isunlikely in a fl are whose main outcome is chromospheric heating andthus molecular dissociation (the H dissociation energy is 4.55 eV).In panels (d-f) of Fig. 3, we show spectral image stackplots at singlewavelength values of 1333.79 Å for H , 1393.76 Å for Si �� and1402.77 Å for Si �� lines. MNRAS000
IRIS observation details of a fl areIRIS Spectrograph (SG) Slit-Jaw-Imager (SJI)Observation date 18-April-2014Observation ID 3820259153IHOP ∗ �� (slit width) 0.33 �� Field-of-view 0.166 �� × �� �� × �� Exposure time (sec) 7.9 1.98Cadence (sec) 9.4 28 (for C �� , Si �� , Mg �� ) ∗ that of H . Hence, we conclude that the S � line at 1333.80 Å is notblended with H line at 1333.79 Å.During the evolution of the fl are, the southern ribbon was observedto move southwards in SJI images, and this displacement was nicely observed in the spectral images. The emission from R1 as it movessouth is shown in the time stackplots of spectral images in Fig. 3.R1 and R2 are indicated by white arrows. R2 has a less well-de fi nedmotion. The images shown in panels (a)-(c) of Fig. 3 are created bysumming DNs over the wavelength ranges 1333.76-1333.87 Å forH , and 1393.5-1394.6 Å and 1402.5-1403.6 Å for the Si �� lines. �� By selecting particular wavelengths, we can examine in some detailthe ribbon evolution and correlations between the exciter wavelengthand the fl uorescent emission. The fi rst thing to notice, in panels (a-c)of Fig. 3, is that the H emission becomes visible when the Si �� fl uorescentexcitation is responsible. The alternative explanation for enhancedH emission, that the number of H molecules has increased, isunlikely in a fl are whose main outcome is chromospheric heating andthus molecular dissociation (the H dissociation energy is 4.55 eV).In panels (d-f) of Fig. 3, we show spectral image stackplots at singlewavelength values of 1333.79 Å for H , 1393.76 Å for Si �� and1402.77 Å for Si �� lines. MNRAS000 , 1–10 (2020)
Mulay, S. M. et al.
These Si �� single wavelength values were chosen as they areamong those responsible for exciting H lines observed by IRIS(Table 1).We note that if the fl uorescing atom was moving at speed, sothat the frequency absorbed was Doppler shifted, then the relevantexciting wavelength would have to be corrected from the values givenin Table 1. However, the Doppler speeds measured for the H linesare very small (see Section 2.3), so that the correction is much smallerthan the width of one wavelength pixel, and can be ignored.During the GOES rise phase of the fl are from 12:41 to 12:55 UT,coincident with the beginning of the impulsive phase as indicatedby the GOES derivative, strong emission in the H line (summedover the wavelength range 1333.76-1333.87 Å) originated from fl areribbon R1 and very weak emission from ribbon R2 (see panel (a)of Fig. 3). H emission from R1 reduces considerably at the GOESpeak between 12:54 and 13:03 UT and then brightens again brie fl ybetween 13:04 and 13:10 UT, during the gradual phase. This is moreclearly visible in panel (d) of Fig. 3, corresponding to the singlewavelength pixel at 1333.79 Å. At the time of this later brightening,the GOES derivative indicates that the fl are impulsive phase is over.The Si �� emission for both lines was observed throughout the fl areevolution, (see panels (b) and (c) of Fig. 3) and shows many of the fi nespatial and temporal details seen in H . However, there was very littleemission seen at the H fl uorescent exciting frequency of 1402.77 Åbetween 12:55 and 13:04 UT in R1 (see panel (f) of Fig. 3) similar tothe H emission at 1333.79 Å. This is also true of Si �� at 1393.76 Å,which excites H at 1333.475 Å (see panel (e) of Fig. 3).Also, evident is that R2 is very bright in both the total Si �� intensityplot and the 1402.77 Å plot, but very faint in H . We examine thisfurther by plotting light curves for ribbon R1 (between 12:42 and13:10 UT) and R2 (between 12:47 and 12:52 UT) (see top panels ofFig. 4). At each time, the total intensities were obtained for all pixelsin panel (a) of Fig. 3 where emission from H (between 1333.76 Å and1333.87 Å) and Si �� (between 1402.5 and 1403.6 Å) was observed.For R1, despite being almost three orders of magnitude di ff erent inDN, the overall pattern of H intensity variation is very similar to theintensity variation in Si �� , showing many of the same small-scalefeatures. For R2, there is little small-scale intensity variation in Si �� and H . Scatter plots for R1 and R2 (see panels (c) and (d) of Fig. 4)show a positive correlation between H and Si �� line intensities.As remarked above, ribbon R2 is very bright at the exciting wave-length (i.e. at Si �� at thesame time and location is faint. This is seen also in the shallowergradient for the R2 scatter plot in panel (d) of Fig. 4. To excite theH line, the emission at 1402.77 Å must be able to penetrate to thelocation where molecular hydrogen is present, so it may be that theopacity of the chromosphere down to this level at the R2 location ishigher than at the R1 location. It is perhaps notable that R1 crosses aplage region, whereas; R2 crosses a spot penumbra, which would beexpected to have di ff erent temperature, density and hence, opacitystructures.In order to investigate this, we used a spectroscopic diagnostic tool– the intensity ratio of the resonance lines of Si �� (1393.76/1402.77)– to study the optical thickness of the plasma at the fl are location. Theplasma is considered to be optically thin if the Si �� intensity ratio is2 (Mathioudakis et al. 1999). This has been used by Yan et al. (2015),who found a ratio of less than 2, with Si �� self-absorption features intransition region brightenings further indicating that opacity e ff ectsplayed an important role. Tripathi et al. (2020) found the ratio tobe smaller than 2 at the periphery of an emerging fl ux region, butlarger than 2 in its core. As noted, our Si �� lines exhibit complex,multi-component pro fi les (shown in panels (e) and (f) of Fig. 2), however if each of the components is optically thin then the ratiofrom the intensities integrated across the line should be equal to 2.The ratio plot is displayed in panel (a) of Fig. 5. The over-plottedblack contours are Si �� emission at R1 and R2. Locations where oneor both Si �� lines are saturated and the ratio cannot be evaluated areshown in the darkest red shades.For R1, during the rising phase (12:42-12:51 UT) of the GOES fl are, the intensity ratio is 2, consistent with optically thin conditions(Brannon et al. 2015), whereas; at the GOES peak between 12:52 and13:06 UT, the intensity ratio is between 1.8 and 2.0. It is less than 2during intervals 12:52-12:54 UT and 13:03-13:06 UT. Between 13:09and 13:12 UT, the ratio was larger than 2 in R1, and at a substantialnumber of pixel locations in R2 it is larger than 2.1. However, inR2 there are also some patches between Y-pixels 628 and 650 forthe interval 12:48-12:51 UT where the ratio is smaller than 1.9. Thepanel (b) of Fig. 5 show histograms of the Si �� intensity ratios for thetwo ribbons, showing a greater tendency for the ratio to be less than2 in R1 and greater than 2 in R2. This will be discussed in Section 3,particularly in the context of fl are-speci fi c simulations by Kerr et al.(2019). We examined the spectra of H and both Si �� lines. The H fl are evolution ex-cept for slit position number 118 and it was di ffi cult to fi t the linewith a Gaussian pro fi le. Hence, we focused on the unblended H line at 1333.79 Å. The counts were also low for the H fl are evolution, so a better signal-to-noise wasobtained by averaging the spectra over a number of pixels along theY-axis (i.e. along the slit, pixel numbers are given in Column 2 ofTable 3). The line was fi tted with a single Gaussian pro fi le. The IRISinstrumental width (FWHM) is about 26 mÅ (3.51 km s − , De Pon-tieu et al. 2014) for the FUV channel. The widths of the H and Si �� line pro fi les were observed to be broader than both the instrumen-tal width and the thermal width for each line, assuming that the H emission originates from plasma at ∼ �� .Small Doppler shifts in H were detected. Both Si �� lines were verybroad (much broader than their thermal widths), and showed non-Gaussian line pro fi les at each slit position (see in panels (e) and (f) ofFig. 2). Multiple Gaussian components with di ff erent red- and blue-shifts are needed to fi t the Si �� lines. We do not explore this furtherhere, but details can be found in Brannon et al. (2015) and Chenget al. (2015). A single Gaussian component was used by Cheng et al.(2015) for simple Si �� line pro fi les and they measured a red shiftof 12 km s − , and a FWHM of 26 km s − . Brannon et al. (2015)identi fi ed pixel positions in which Si �� lines were best fi tted withtwo Gaussian components. They found periodically-varying Dopplervelocities: the bluer component had an approximately 140 s periodoscillation with a sawtooth character, varying over ±
40 km s − , andthe redder component had fl uctuations around an average redshift of50 km s − .The physical parameters determined for H line. The widths rangedfrom 0.037-0.084 Å, whereas; nonthermal velocities ranged from7.1-17.8 km s − and the Doppler velocities were measured for H line. Very small red and blue shifts were obtained at ribbon R1 during MNRAS , 1–10 (2020) vidence of molecular hydrogen emission in a fl are Figure 5.
Panel (a): The intensity ratio of Si �� �� �� line ratios for pixels in ribbons R1 and R2. The red vertical lines at 12:43, 13:09 and13:17 UT indicate that the IRIS spectra is missing at those slit locations/timings. Table 3.
Parameters derived from the H Column 1 Column 2 Column 3 Column 4 Column 5 Column 6 Column 7Slit Position Y-pixels Time Centroid FWHM V 𝑛𝑡ℎ V 𝐷𝑜𝑝𝑝𝑙𝑒𝑟
Number along the slit (UT) (Å) (Å) (km s − ) (km s − )54 672-681 12:41:56 1333.8015 0.056 ± ± ± ± ± ± ± ± ± ± ± ±
371 668-673 12:44:36 1333.8073 0.037 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± the evolution of the fl are, though almost all were consistent with zerowithin the errors. In this paper, we carried out the fi rst comprehensive investigationof a molecular H line observed by IRIS in a fl are, revealing someproperties of the cool emitting plasma. The temporal and spatial MNRAS000
371 668-673 12:44:36 1333.8073 0.037 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± the evolution of the fl are, though almost all were consistent with zerowithin the errors. In this paper, we carried out the fi rst comprehensive investigationof a molecular H line observed by IRIS in a fl are, revealing someproperties of the cool emitting plasma. The temporal and spatial MNRAS000 , 1–10 (2020)
Mulay, S. M. et al. evolution of the H emission from fl are ribbons was studied and thefollowing properties derived for this event: • The emission in the H line at 1333.79 Å and at its fl uorescentexciting wavelength of 1402.77 Å (the Si �� line), at both ribbonlocations, are strongly correlated in space and in time; • The correlation coe ffi cient for the H - Si �� intensity correla-tions for ribbons R1 and R2 are of same order of magnitude, but thegradient is di ff er by ∼ • The H line is strongest during the fl are impulsive phase, dimsduring the GOES peak, and brightens again during the gradual phase; • The H line is broadened, corresponding to non-thermal speedsin the range 7-18 km s − ; • The H line also shows small red (blue) shifts, up to 1.8(4.9) km s − • The intensity ratio of Si �� �� ff ersfrom its optically-thin value of 2 in parts of the ribbons, indicating arole for opacity e ff ects.Our analysis provides clear evidence that the H line is fl uores-cently excited by the Si �� line at 1402.77 Å as shown by the spatialand temporal correlations observed. The Si �� line, with a formationtemperature of 80,000 K, is emitted in the transition region, andsome of the downwards-going radiation arrives at much deeper lay-ers where H is formed at a temperature of ∼ emission re- fl ect the properties of this cool plasma. Measured H Doppler shiftsare consistent with zero within the errors, indicating negligible bulk fl ows along the line-of-sight. This is in contrast to the fi nding ofsystematic periodic blue- and red-shifts of ∼
20 km s − mean am-plitude in Si �� , O �� and C �� emission lines in the same event byBrannon et al. (2015). They suggested that oscillatory motion oftransition region and upper chromospheric plasma on swaying fi eldlines anchored in the deep atmosphere would give rise to the patternof Doppler shifts. If that is the cause then one would expect the ampli-tude of the oscillation to decrease with increasing depth, and indeedthe amplitude of this swaying motion has reduced to unobservablelevels in the TMR.There is clear evidence of non-thermal broadening of H , possiblycorresponding to turbulent plasma fl ows, of around 10 km s − . Sincethe H line only appears at the time of the fl are we cannot say whetherthe fl are is responsible for these non-thermal pro fi les, or whether theywere pre-existing in the TMR and only revealed by the fl uorescentemission.The correlations coe ffi cient between emission at 1402.77 Å and H are of same order of magnitude in ribbons R1 and R2, but the gradientis di ff er by ∼
50% as shown in Figure 4. This is most straightforwardlyexplained as being due to a lower fl ux of the exciter radiation arrivingat the depth in the atmosphere where the H molecule is present inR2. This could be as a result of di ff ering optical paths between theTMR where H is formed and the transition region where Si �� isemitted, caused by di ff erent chromospheric temperature and densitystructures at the two locations, through which the exciting radiationhas to pass.We cannot measure the optical depth for the downwards-going radiation, but can investigate the optical properties of plasma tothe outward-going radiation, using the ratio of the two Si �� lineintensities. During the impulsive phase, this ratio measured at ribbonR1 corresponds to an optically thin plasma (ratio of 2). A slightdecrease in the ratio to between 1.8 and 2.0 during the peak of GOESindicates an increase in the opacity.In the case of R2, there are many pixel locations where the ratio islarger than 2.1. It has been argued by Mathioudakis et al. (1999) that opacity e ff ects lead to ratios below 2, with a ratio of 1.8 correspond-ing to an optical depth 𝜏 ∼ .
25, and by Gontikakis & Vial (2018) thatvalues greater than 2 indicate a contribution from resonant scatteringof Si �� radiation. However, detailed fl are radiation hydrodynamicssimulations by Kerr et al. (2019) demonstrate that opacity e ff ects canlead to a range of ratios from 1.8 to 2.3, corresponding closely towhat we fi nd, in cases where the fl are energy carried by non-thermalelectrons exceeds 5 × erg cm − s − and/or the electron spec-trum is soft. In the simulations, the opacity e ff ects vary on timescalesof seconds. The authors argue that the intensity ratio represents theratio of line source functions, which depends strongly on the com-plex structure of the upper chromosphere and transition region. Thisrequires further theoretical study.We believe that the study of H line and derived plasma parametersprovided here will be useful for constraining further models of thechromosphere and TMR during fl ares. ORCID ID’S
Sargam M. Mulay https://orcid.org/0000-0002-9242-2643Lyndsay Fletcher https://orcid.org/0000-0001-9315-7899
ACKNOWLEDGEMENTS
SMM and LF acknowledge support from the UK Research and Inno-vation’s Science and Technology Facilities Council under grant awardnumbers ST/P000533/1 and ST/T000422/1. The authors would liketo thank Dr. Peter Young (NASA Goddard Space Flight Center, USA),Dr. Giulio Del Zanna, Dr. Helen Mason and Mr. Roger Dufresne(University of Cambridge, UK) and Prof. Durgesh Tripathi (Inter-University Centre of Astronomy and Astrophysics, India) for thediscussion and valuable comments. IRIS is a NASA small explorermission developed and operated by LMSAL with mission operationsexecuted at NASA Ames Research center and major contributions todownlink communications funded by ESA and the Norwegian SpaceCentre. AIA data are courtesy of SDO (NASA) and the AIA con-sortium. NOAA Solar Region Summary data supplied courtesy ofSolarMonitor.org. The GOES 15 X-ray data are produced in real timeby the NOAA Space Weather Prediction Center (SWPC) and are dis-tributed by the NOAA National Geophysical Data Center (NGDC).
DATA AVAILABILITY
In this paper, we used the Interactive Data Language (IDL) and Solar-SoftWare (SSW; Freeland & Handy 1998) packages to analyse AIAand IRIS data. All of the fi gures within this paper were produced us-ing IDL colour-blind-friendly colour tables (see Wright 2017). IRIShas an open data policy. The IRIS data is available at https://iris.lmsal.com/data.html and the data analysis was performed usingthe routines available at https://iris.lmsal.com/tutorials.html . The data and calculation of physical parameters for H lines areavailable at https://github.com/SargamMulay . The AIA datais available at http://jsoc.stanford.edu/ and the data wereanalysed using routines available at MNRAS , 1–10 (2020) vidence of molecular hydrogen emission in a fl are sdodocs/doc/dcur/SDOD0060.zip/zip/entry/ . The solar fl aredetails are obtained from the archive ftp://ftp.swpc.noaa.gov/pub/warehouse/ .The GOES data analysis was performed by fol-lowing IDL routines available at https://hesperia.gsfc.nasa.gov/rhessidatacenter/complementary_data/goes.html REFERENCES
Abgrall H., Roue ff E., Launay F., Roncin J. Y., Subtil J. L., 1993a, A&AS,101, 273Abgrall H., Roue ff E., Launay F., Roncin J. Y., Subtil J. L., 1993b, A&AS,101, 323Bartoe J.-D. F., Brueckner G. E., Nicolas K. R., Sandlin G. D., VanHoosierM. E., Jordan C., 1979, Monthly Notices of the Royal AstronomicalSociety, 187, 463Brannon S. R., Longcope D. W., Qiu J., 2015, The Astrophysical Journal,810, 4Brosius J. W., Daw A. N., 2015, The Astrophysical Journal, 810, 45Brosius J. W., Daw A. N., Inglis A. R., 2016, ApJ, 830, 101Cheng X., Ding M. D., Fang C., 2015, ApJ, 804, 82Cohen L., Feldman U., Doschek G. A., 1978, ApJS, 37, 393Curdt, W. Brekke, P. Feldman, U. Wilhelm, K. Dwivedi, B. N. Schühle, U.Lemaire, P. 2001, A&A, 375, 591De Pontieu B., et al., 2014, Sol. Phys., 289, 2733Dennis B. R., Zarro D. M., 1993, Sol. Phys., 146, 177Freeland S. L., Handy B. N., 1998, Sol. Phys., 182, 497Gontikakis C., Vial J. C., 2018, A&A, 619, A64Herzberg G., Howe L. L., 1959, Canadian Journal of Physics, 37, 636Hudson H. S., 1991, in Bulletin of the American Astronomical Society.p. 1064Innes D. E., 2008, A&A, 481, L41Jaeggli S. A., Judge P. G., Daw A. N., 2018, ApJ, 855, 134Jordan C., Brueckner G. E., Bartoe J. D. F., Sand lin G. D., van Hoosier M. E.,1977, Nature, 270, 326Jordan C., Brueckner G. E., Bartoe J. D. F., Sand lin G. D., Vanhoosier M. E.,1978, ApJ, 226, 687Kerr G. S., Carlsson M., Allred J. C., Young P. R., Daw A. N., 2019, ApJ,871, 23Lemen J. R., et al., 2012, Sol. Phys., 275, 17Li D., Innes D. E., Ning Z. J., 2016, A&A, 587, A11Mathioudakis M., McKenny J., Keenan F. P., Williams D. R., Phillips K. J. H.,1999, A&A, 351, L23Neupert W. M., 1968, ApJ, 153, L59Pesnell W. D., Thompson B. J., Chamberlin P. C., 2012, Sol. Phys., 275, 3Sandlin G. D., Bartoe J. D. F., Brueckner G. E., Tousey R., Vanhoosier M. E.,1986, ApJS, 61, 801Schmit D. J., Innes D., Ayres T., Peter H., Curdt W., Jaeggli S., 2014, A&A,569, L7Schüehle U., Brown C. M., Curdt W., Feldman U., 1999, in Vial J. C.,Kaldeich-Schü B., eds, ESA Special Publication Vol. 446, 8th SOHOWorkshop: Plasma Dynamics and Diagnostics in the Solar TransitionRegion and Corona. p. 617Tripathi D., Nived V. N., Isobe H., Doyle G. G., 2020, ApJ, 894, 128Wilhelm K., et al., 1995, Sol. Phys., 162, 189Wright P., 2017, ColourBlind: A Collection of Colour-blind-friendly ColourTables, doi:10.5281/zenodo.840393Yan L., et al., 2015, ApJ, 811, 48Young P. R., Tian H., Jaeggli S., 2015, ApJ, 799, 218
APPENDIX A: FORMATION OF H LINES
The absorption of far-UV photons gives rise to electronic excitationin H . There are a number of vibrational levels in each electronicstate, so de-excitation to the ground electronic state leads to theformation of H lines at a range of wavelengths. Excitation of the upper state requires photons of speci fi c wavelength, resulting fromemission in far-UV atomic lines, or continuum, or indeed other H molecular lines. Table A1 provides the exciting UV atomic emissionlines (Column 1) for the fl uorescent channels and wavelengths ofinterest for the IRIS H windows we study (Column 2); the pos-sible decay options available from the upper level (Columns 3,4);and their wavelengths/energy levels (Columns 5,6). The fl uorescencelines along with their energy levels are obtained from Abgrall et al.(1993a).If we look for example at the the H line at 1333.797 Å, it isformed by de-excitation of a level with rotational quantum numberJ=1 and vibrational quantum number v � = 0 in the fi rst electronic stateof H . Photons at 1402.648 Å excite this upper level in an upwardstransition between the ground and fi rst electronic states. The upperexcited state can then decay to any rotational and vibrational statebelow it that is allowed by quantum mechanical selection rules. Thisincludes vibrational states in the electronic ground state with energies below that from which the upwards transition was originally excited.Therefore, the H line wavelength can be signi fi cantly smaller thanthe wavelength of the photon that excited it.In the case of the fl are we examine, the UV line emission can beseen from the spectral pro fi les to dominate over the continuum at the fl uorescent channel wavelengths. The upper levels of the H lines aretherefore excited primarily by absorption of Si �� and C �� photons(and are thus a possible de-excitation option for Si �� and C �� ). Inparticular, the H line at 1333.797 Å is excited by photons in thewing of the 1402.77 Å Si �� line.In a theoretical study Jaeggli et al. (2018) identi fi ed a new pos-sible wavelength at 1393.961 Å along with the previously known1402.648 Å as a pumping source for the upper-level population ofH at 1333.797 Å. However their non-LTE modeling suggested thatin fl ares the 1402.648 Å would be dominant. APPENDIX B: UNDERSTANDING A POSSIBLE BLEND OFS � (1333.80 Å) WITH THE H (1333.797 Å) In order to identify a possible blend of S � at 1333.80 Å with the H at 1333.797 Å, we have taken S � line.We obtained 24 spectral pro fi les of S � fi t only 11 spectral pro fi les with a single Gaussiancomponent. We derived the intensities, Doppler velocities and widthsof the line and compared with H parameters. The remaining spectrawere slightly narrow in the core and broader in the wings, and twoGaussian components were needed to fi t the line. Hence, we did notuse these line pro fi les for further analysis. Figure B1 shows spectralpro fi les of two H lines (panel a) and the S � line (panel b) obtainedat slit position number 118. The Gaussian fi ts for individual lines areshown.We compared the parameters obtained from S � fi les with those obtained from H at 1333.79 Å. Figure B2 showsscatter plots for the intensity of lines (panel a), Doppler velocities(panel b) and widths of the lines (panel c). The Pearson correlationcoe ffi cients show weak positive correlation for the intensity and weaknegative correlation for the Doppler velocities. The widths of thelines show moderate correlation. The above results con fi rmed thatthe behaviour of the S � ff erent than H line.Hence, we conclude that there is an absence of any signi fi cant S � line contribution in H line at 1333.79 Å. MNRAS000
The absorption of far-UV photons gives rise to electronic excitationin H . There are a number of vibrational levels in each electronicstate, so de-excitation to the ground electronic state leads to theformation of H lines at a range of wavelengths. Excitation of the upper state requires photons of speci fi c wavelength, resulting fromemission in far-UV atomic lines, or continuum, or indeed other H molecular lines. Table A1 provides the exciting UV atomic emissionlines (Column 1) for the fl uorescent channels and wavelengths ofinterest for the IRIS H windows we study (Column 2); the pos-sible decay options available from the upper level (Columns 3,4);and their wavelengths/energy levels (Columns 5,6). The fl uorescencelines along with their energy levels are obtained from Abgrall et al.(1993a).If we look for example at the the H line at 1333.797 Å, it isformed by de-excitation of a level with rotational quantum numberJ=1 and vibrational quantum number v � = 0 in the fi rst electronic stateof H . Photons at 1402.648 Å excite this upper level in an upwardstransition between the ground and fi rst electronic states. The upperexcited state can then decay to any rotational and vibrational statebelow it that is allowed by quantum mechanical selection rules. Thisincludes vibrational states in the electronic ground state with energies below that from which the upwards transition was originally excited.Therefore, the H line wavelength can be signi fi cantly smaller thanthe wavelength of the photon that excited it.In the case of the fl are we examine, the UV line emission can beseen from the spectral pro fi les to dominate over the continuum at the fl uorescent channel wavelengths. The upper levels of the H lines aretherefore excited primarily by absorption of Si �� and C �� photons(and are thus a possible de-excitation option for Si �� and C �� ). Inparticular, the H line at 1333.797 Å is excited by photons in thewing of the 1402.77 Å Si �� line.In a theoretical study Jaeggli et al. (2018) identi fi ed a new pos-sible wavelength at 1393.961 Å along with the previously known1402.648 Å as a pumping source for the upper-level population ofH at 1333.797 Å. However their non-LTE modeling suggested thatin fl ares the 1402.648 Å would be dominant. APPENDIX B: UNDERSTANDING A POSSIBLE BLEND OFS � (1333.80 Å) WITH THE H (1333.797 Å) In order to identify a possible blend of S � at 1333.80 Å with the H at 1333.797 Å, we have taken S � line.We obtained 24 spectral pro fi les of S � fi t only 11 spectral pro fi les with a single Gaussiancomponent. We derived the intensities, Doppler velocities and widthsof the line and compared with H parameters. The remaining spectrawere slightly narrow in the core and broader in the wings, and twoGaussian components were needed to fi t the line. Hence, we did notuse these line pro fi les for further analysis. Figure B1 shows spectralpro fi les of two H lines (panel a) and the S � line (panel b) obtainedat slit position number 118. The Gaussian fi ts for individual lines areshown.We compared the parameters obtained from S � fi les with those obtained from H at 1333.79 Å. Figure B2 showsscatter plots for the intensity of lines (panel a), Doppler velocities(panel b) and widths of the lines (panel c). The Pearson correlationcoe ffi cients show weak positive correlation for the intensity and weaknegative correlation for the Doppler velocities. The widths of thelines show moderate correlation. The above results con fi rmed thatthe behaviour of the S � ff erent than H line.Hence, we conclude that there is an absence of any signi fi cant S � line contribution in H line at 1333.79 Å. MNRAS000 , 1–10 (2020) Mulay, S. M. et al.
Table A1.
Various de-excitation options for Si �� and C �� linesColumn 1 Column 2 Column 3 Column 4 Column 5 Column 6Exciting line Fluorescent channel Transition Branch H Wavenumber 𝜆 (Å) ( v � - v �� ) ( v � - v �� ) ( Δ J = ± 𝜆 (Å) (cm − )Si �� �� �� �� Notes -
The details in the Columns 1-5 are obtained from the report on molecular hydrogen by Prof.Peter Young and it is available at https://pyoung.org/iris/Vibrational quantum number, v � (upper level) and v �� (lower level),Rotational quantum number, Δ J = -1 ( P branch) and Δ J = +1 ( R branch).Column 6 provides a list of energy levels from Abgrall et al. (1993) Based on non-LTE models, Jaeggli et al. (2018) studied thestrength of H line at 1342.256 Å in the quiet-Sun spectra. Since theline at 1333.79 Å and H line at 1342.256 Å originate from the sameupper level (v � = 0), they should exhibit similar strength and proper-ties. A discrepancy between H (1342.256 Å) and S � (1396.113 Å)line intensities also led these authors to rule out the presence of ablend of the S � line (1333.80 Å) with H line (1333.79 Å). This paper has been typeset from a TEX/L A TEX fi le prepared by the author.MNRAS , 1–10 (2020) vidence of molecular hydrogen emission in a fl are Figure B1.
The spectral pro fi les of (a) H at 1333.47 Å and 1333.79 Å and (b) S � at 1401.515 Å which are obtained for slit number 118 at 12:51:57 UT. Thespectrum was obtained by averaging pixels between 650 and 660 along the slit. The blue and orange dashed lines indicate Gaussian components used for fi ttingthe lines and the entire fi t is shown by solid red line. The horizontal cyan lines indicate a fi t for background emission. Figure B2.
The scatter plots for the measured parameters (a) intensity of lines obtained from the single Gaussian fi t, (b) Doppler velocities, and (c) width of theH at 1333.79 Å and S � at 1401.515 Å lines. The data are displayed with star symbols, and solid blue lines indicate the linear fi t to the data. The equations forthe fi tted lines along with fi t parameters are given and the Pearson correlation coe ffi cients are displayed as ‘R’. MNRAS000