High resolution radio observations of NGC 4631: probing the central starburst

We present high resolution radio continuum observations of the edge-on galaxy NGC 4631 with the Very Large Array A, BnA & B arrays in the L-, C-, and X-band. The observed emission is limited to the inner rigidly rotating and star forming 4 kpc-diameter disk of NGC 4631. Due to missing large-scale emission of the interferometric observations we detected only 7% and 12% of the total flux densities at L- and C-band, respectively. The spectral index of the detected emission between L-and C-band is 0.23 (S-nu proportional to nu(-alpha)), implying a thermal fraction of similar to 50%. We interpret this high thermal fraction as caused by the missing large-scale emission. The non-thermal radio emission has a more extended and diffuse distribution to which our observations are not sensitive. The emission is distributed asymmetrically with respect to the major axis, the dynamical center and the center of the stellar bulge. Gravitational interaction may play a role causing these asymmetries. We observe no correlation of our radio data with H alpha data, and argue that this is likely caused by massive dust extinction towards the central region at optical wavelengths. Point sources are identified at resolutions of 1.45 " and 0.83 " x 0.61 " in the central region, and their Aux densities and spectral indices are derived. Two non-thermal point source complexes show strong morphological evidence for being external to NGC 4631; they are likely radio galaxies in the background. Alternatively, we discuss whether one of these sources may be the weakly active nucleus with jet-like features. Only 2-3 other point sources are non-thermal. They may be identified with supernova remnants; Using CAS A as a 'standard candle', we estimate a supernova rate of 0.009 yr(-1) We estimate the thermal fraction of the total C-band radio emission of NGC 4631 with various methods, and find a likely range f(th) similar to 8-15%. Assuming a standard IMF, the thermal fraction can be translated to a high-mass (M greater than or similar to, 10 M.) star formation rate of similar to 0.45-0.85 M. yr(-1), and a supernova rate of 0.03-0.056 yr(-1). This rate is too high by a factor of three to six compared to the supernova rate estimated from the nonthermal point source population. it is shown that, in contrast to NGC 4631, the supernova and star formation rates in M 82 derived with the same methods agree well. Implications of the discrepancy, including an IMF biased toward massive stars and lower average SNR luminosities in NGC 4631 are discussed.