SOLAR ELECTRON-BEAMS DETECTED IN HARD X-RAYS AND RADIO-WAVES

We present a statistical survey of electron beam signatures that are detected simultaneously at hard X-ray (HXR) and radio wavelengths during solar flares. For the identification of a simultaneous event we require a type III (normal-drifting or reverse-slope-drifting) radio burst that coincides (within +/-1 s) with a significant (greater than or equal to 3 sigma) HXR pulse of similar duration (less than or equal to 1 s). Our survey covers all HXRBS/SMM and BATSE/CGRO flares that were simultaneously observed with the 0.1-1 GHz spectrometer Ikarus or the 0.1-3 GHz spectrometer Phoenix of ETH Zurich during 1980-1993. The major results and conclusions are as follows: 1. We identified 233 HXR pulses (out of 882) to be correlated with type III-like radio bursts: 77% with normal-drifting type III bursts, 34% with reverse-slope (RS)-drifting bursts, and 13% with oppositely drifting (III+RS) burst pairs. The majority of these cases provide evidence for acceleration of bidirectional electron beams. 2. The detailed correlation with type III-like radio bursts suggests that most of the subsecond fluctuations detectable in greater than or equal to 25 keV HXR emission are related to discrete electron injections. This is also supported by the proportionality of the HXR pulse duration with the radio burst duration. The distribution of HXR pulse durations w(x) is found to have an exponential distribution, i.e., N(w(x)) proportional to exp (-w(x)/0.25 s) in the measured range of w(x) approximate to 0.5-1.5 s. 3. From oppositely drifting radio burst pairs we infer electron densities of n(e) = 10(9)-10(10) cm(-3) at the acceleration site. From the absence of a frequency gap between the simultaneous start frequencies of upward and downward drifting radio bursts, we infer an upper limit of L less than or equal to 2000 km for the extent of the acceleration site and an acceleration time of At Delta t less than or equal to 3 ms for the (greater than or equal to 25 keV) radio-emitting electrons (in the case of parallel electric fields). 4. The relative timing between HXR pulses and radio bursts is best at the start frequency (of earliest radio detection), with a coincidence of less than or equal to 0.1 s in the statistical average, while the radio bursts are delayed at all other frequencies (in the statistical average). The timing is consistent with the scenario of electron injection at a mean coronal height of h approximate to 10(4) km. The radio-emitting electrons are found to have lower energies (greater than or equal to 5 keV) than the greater than or equal to 25 keV HXR-emitting electrons. 5. The modulated HXR flux that correlates with electron beam signatures in radio amounts to 2%-6% of the total HXR count rate (for BATSE flares). The associated kinetic energy in electrons is estimated to be E = 4 x 10(22)-10(27) ergs per beam, or N-e = 4 x 10(28)-10(33) electrons per beam, considering the spread from the smallest to the largest flare detected by HXRBS. 6. The average drift rate of propagating electron beams is found here to be \dv/dt\ = 0.10v(1.4) MHz s(-1) in the frequency range of v = 200-3000 MHg which is lower than expected from the Alvarez & Haddock relation for frequencies less than or equal to 550 MHz. 7. The frequency distributions of HXR fluxes (F-x) and radio type III burst fluxes (F-R), which both can be characterized by a power law, are found to have a significantly different slope, i.e., N(F-x) proportional to F-x(-1.87) versus N(F-R) proportional to F-R(-1.28). The difference in the slope is attributed to the fundamental difference between incoherent and coherent emission processes. In summary, these findings suggest a flare scenario in which bidirectional streams of electrons are accelerated during solar flares at heights of approximate to 10(4) km above the photosphere in rather compact regions (L less than or equal to 2000 km). The acceleration site is likely to be located near the top of flare loops (defined by HXR double footpoints) or in the cusp above, where electrons have also access to open field lines or larger arches. The observed bidirectionality of electron beams favors acceleration mechanisms with oppositely directed electric fields or stochastic acceleration in an X-type reconnection geometry.