Introduction

A solar flare generally consists of a "thermal" and a "non-thermal" component. A "thermal" component can be explained by a Maxwellian particle distribution, i.e. one that reflects a stable, relaxed state. But energy is suddenly released when a flare happens, and we infer that this involves non-thermal effects - particle acceleration, radio emission, high-energy radiations such as hard X-rays, etc. The key distinction seems to be in the X-ray spectrum of a flare; soft X-ray photons (energies below about 10 keV) behave in a "thermal" manner, whereas hard X-rays (above about 10 keV) seem distinctly non-thermal.

Large and small flares have been found to differ quantitatively in their (thermal) soft X-ray emission in the past. Do they differ also in the (non-thermal) hard X-ray emission? RHESSI is the perfect tool for a statistical study of flares of different sizes. Due to its high spectral resolution we have very accurate spectra and can disentangle thermal and non-thermal emission.

Observations

We randomly picked 85 well observed flares from GOES class B1-M5 and fitted the spectra at the peak time of the hard X-ray flux. This gave us the spectral index of the non-thermal power law and the non-thermal flux at 35 keV.

Here is the relation between these two parameters, which characterize the non-thermal component of a flare in the simplest possible way:

figure name

Despite the large scatter one can clearly see that flares with smaller hard X-ray flux are softer. The real world is more tricky, as there is a selection effect that excludes the lower left corner. Nevertheless we have convinced ourselves that the correlation is significant.

A different picture emerges when plotting the spectral index against the background-subtracted GOES flux:

Gamma vs GOES

No clear correlation can be found. However, if the events are divided into three intervals in non-thermal flux ([10-4,10-2] for blue triangles, [10-2,1] for green squares, [1,102] for red dots), they line up along the expected correlation line (dashed) with huge scatter on both sides of the line.

Conclusions

These observations show that small flares in non-thermal flux are softer, on average, than larger flares. The correlation is much clearer when using the non-thermal flux instead of the soft X-ray flux. This is NOT due to do some kind of soft-hard-soft behaviour with the spectrum moving up and down around some fixed point as observed for individual flares, as we compare peak fluxes of different flares.

Thus we find that the smaller flares have softer hard X-ray spectra and thus are more difficult to detect than large flares with harder spectra. Small flares consequently add up to greater non-thermal energy than previously thought. This finding may have a serious effect on the total energy input into the corona by flares.

Biographical note: Marina Battaglia is working on her PhD thesis at ETH Zürich in the group of Arnold Benz.