Early observations showed that some flares have an aparent spectral flattening at mm-λ. Kaufmann et al. (1986) tried to explain that flattening with an additional high frequency spectral component. Ramaty & Petrosian (1972) argued that free-free absorption of the non-thermal gyrosynchrotron emission can explain this flatness. Chertok et al. (1995) showed a few more examples of flat spectra. They adhered to the free-free model, and argued that chromospheric evaporation could be responsible for the local density increase that absorbed the gyrosynchrotron emission.

In this work we analyze a sequence of two solar flares ocurred on December 20, 2002. The events took place in the active region AR10226 (S26 W32) at 13:16 UT and 13:18 UT. Radio data at 1-405 GHz frequency range were obtained from the Solar Submillimeter Telescope (SST), Radio Solar Telescope Network (RSTN) and the patrol radiotelescopes of the Bern University, one of them working as a null interferometer at 89.4 GHz. Images from the Hα Solar Telescope for Argentina (HASTA) were used to identify the possible emission sources.

The microwave emission during the peak time of the first flare shows the classical gyrosynchrotron spectrum shape due to a population of accelerated power law electrons in a homogeneous ambient (Ramaty 1969) with a turnover frequency around 12 GHz. The second flare has a microwave spectrum with a flat region of about one decade length (between 2 GHz and 14 GHz approximately). We used a homogeneous model to fit the gyrosynchrotron emission (Ramaty et al. 1994) which gave us, for the first flare, B = 310 G, A = 9␛, n = 5 × 10⁸ cm⁻³ and δ = 2.9, for the magnetic field strength, emitting area, number density and electron index respectively. A fitting to the spectrum of the second flare yielded the following results: B = 310 G, A = 18″, n = 1.4 × 10⁹ cm⁻³ and δ = 3.8. In addition, the flux density observed at 212 GHz could indicate the presence of a high frequency second component, as has been observed in other events (see e.g., Kaufmann et al. 2004).

Hα images reveal the presence of three kernels which flare at different times. One of them peaks simultaneously with the first radio bursts, while the brightening of the other two, are coincident with the second burst.

Taking into account the results derived from the fittings, wich showed very different electron populations, we conclude that a loop-loop interaction occurred which changed significantly the medium where the second burst took place. In order to better support this hypothesis we will analyze in detail the AR magnetic field evolution and a coronal magnetic field model will be built to identify the interacting coronal structures.