Fast Particles in Solar Flares The view from

Fast Particles in Solar Flares The view from RHESSI (and TRACE) Ramaty High Energy Solar Spectroscopic Imager Spectroscopy Imaging: X-ray and gamma-rays Coronal sources Footpoint sources Estimates of reconnection rate Conclusions Lyndsay Fletcher, University of Glasgow MRT Newton Institute Aug 18 th

R

Ge Detector High Resolution Spectrum 1 ke. V bins at < 100 ke. V Thermal bremsstrahlung Non-thermal bremsstrahlung MRT Newton Institute Aug 18 th 2. 2 Me. V line

Movie: Eduard Kontar MRT Newton Institute Aug 18 th

Electrons from photons: forward fitting Photon spectrum I( ) is related to source-averaged electron spectrum: photons e. g. F(E) modelled with Maxwellian plus two powerlaws Power-law electrons Holman et al 2003.

Electrons from photons: numerical inversion Photon spectrum I( ) is related to source-averaged electron spectrum by Write as discretised matrix equation Solve – eg as minimisation problem with smoothing Bars – inversion Full line – forward fitting Piana et al 2003

The devil in the details Are features in the source spectrum real properties of the spectrum? Or do they arise because of simplifications made in deducing them? Typically, neither forward-fitting nor inversion takes account of: - non-uniform ionisation of chromosphere - photospheric hard X-ray albedo - electron-electron bremsstrahlung Forward fitting (at present) further ignores the possibility of - multi-thermal plasmas MRT Newton Institute Aug 18 2004

Effect of Albedo Inversion with correction for reflection of photons from photosphere can smooth out some of the interesting features (Kontar, Alexander and Brown 2004, in prep. ) MRT Newton Institute Aug 18 2004

Source position as a function of energy Figure: Amir Caspi, UCB Higher energy emission from higher in the looptop – Strongly implies multi-thermal distribution MRT Newton Institute Aug 18 2004

Comments on the electron energy budget/spectrum Total energy deposited by non-thermal electrons is ~ 2. 1024 J in a large (X) flare (assuming cold target, collisionally thick) We cannot uniquely determine the low-energy cutoff or turn-over in the power-law electron component. We can in most cases obtain an upper limit to the cutoff / turnover of typically 20 - 40 ke. V. Most spectra require a double power-law fit above thermal component (but may disappear with further corrections to crosssection) The (minimum) total energy deposited by non-thermal electrons is comparable to the peak total energy in thermal plasma MRT Newton Institute Aug 18 2004

Electron number flux Max number flux = 2 -5 1036 electrons s-1 Holman et al 2003 Coronal density ~ 109 cm-3 So need to accelerate all the electrons in 1027 cm 3 every second MRT Newton Institute Aug 18 2004

July 23: electrons and ions Protons with 10 s of Me. V energy undergo spallation reactions on heavy ions, produce neutrons which are slowed down and undergo capture on H Neutron capture line at 2. 223 Me. V 2. 2 Me. V centroid (i. e. protons) displaced from 50 ke. V centroid (i. e. electrons) by ~ 20” (~5 sigma result) No H , EUV, X-ray enhancement at 2. 2 Me. V centroid location (From Hurford et al. 2003) MRT Newton Institute Aug 18 2004

NB. TRACE image from ~ 45 mins later

October 28: electrons and protons 2. 2 Me. V image (protons) is integrated over 15 minutes • • • Electrons and protons both close to ribbons 2) possible small difference of position: < 15” ( ~104 km) e and p are accelerated in loops of similar size Image: courtesy Krucker & Hurford MRT Newton Institute Aug 18 2004

October 28 Coronal Source Coronal sources can be well-fitted with thermal bremsstrahlung spectra. Temperatures up to ~ 40 MK First appear just before or ~ simultaneously with footpoints Often move during flare (limb events) Image: courtesy Krucker & Hurford MRT Newton Institute Aug 18 2004

RHESSI CLEAN images at different energies: 3 Nov 2003 Image: Astrid Veronig MRT Newton Institute Aug 18 2004

Evolution of RHESSI footpoints and looptop source Footpoints: 70 -100 ke. V Loop top: 20 -25 ke. V Time evolution: black white Image: Astrid Veronig MRT Newton Institute Aug 18 2004

Inferring coronal reconnection rate Reconnection produces a coronal electric field – may directly accelerate particles Outside reconnection region: E+v B =0 Measure of E given by rate of advection of B into reconnection region Ez 2 -D configuration The flare is clearly a 3 -D configuration. However, we still expect high fluxes of fast particles at times of high reconnection rate MRT Newton Institute Aug 18 2004

Flux, spectrum and ‘reconnection rate’ Rapidly reconfiguring magnetic fields should in principle provide a high energy input rate for acceleration of particles Movement of RHESSI source centroids (30 -50 ke. V) show chromospheric mappings of evolving coronal field High HXR flux/hard spectrum occur during intervals of rapid footpoint separation (Fletcher & Hudson 2002) MRT Newton Institute Aug 18 2004

July 23, 2002 Courtesy: Säm Krucker

October 29: HXR flux and footpoint motion. Good correlation between particle flux and ‘reconnection rate’ in later phase of flare, when footpoint motion is ~ regular Images: Säm Krucker MRT Newton Institute Aug 18 2004

July 17 2002 flare: TRACE observations MRT Newton Institute Aug 18 2004

MRT Newton Institute Aug 18 2004

time ~130 separate tracks Fletcher, Pollock & Potts 2004 MRT Newton Institute Aug 18 2004

Flare footpoints on ~ simultaneous magnetogram MRT Newton Institute Aug 18 2004

UV footpoint source intensity variations Typical examples: v BLOS I 1600 Peaks in v BLOS for individual footpoints show significant correlation in time with peaks in the UV brightness, during impulsive phase Peaks within 2 s Peaks within 8 s Observations 25 5% 45 5% Monte-Carlo simulations 8 2% 25 5% MRT Newton Institute Aug 18 2004

Typical value of v BLOS ~ several 100 V m-1 v B ~103 Vm-1 v B ~1. 5 x 103 Vm-1 Hard X-ray footpoints occur where v BLOS ~ 1 k. V m-1 MRT Newton Institute Aug 18 2004

Pairs of correlated footpoints pairs of footpoints for which UV time profiles highly correlated (lines join pairs with linear correlation coefficient > 0. 8) P 1 N P 2

Potential field extrapolation (zero free energy) P 1 N P 2 MRT Newton Institute Aug 18 2004

Conclusions • RHESSI spectroscopy gives new insights into source-averaged electron distributions • There is still more to be explored in the details: e. g. non-isothermality, • We need full imaging spectroscopy (particularly of coronal sources) to get closer to acceleration/heating mechanism • Understanding displacements between signatures of electrons and protons will require better understanding of the magnetic structure (as well as the acceleration mechanisms) • There are suggestions of a good correlation between accelerated electron flux, and a measure of the instantaneous reconnection rate MRT Newton Institute Aug 18 2004