2004A&A...419..849P

Based on recent work on spectral decomposition of the emission of star-forming galaxies, we assess whether the integrated 2-10keV emission from high-mass X-ray binaries (HMXBs), L_2–10^HMXB, can be used as a reliable estimator of ongoing star formation rate (SFR). Using a sample of 46 local (z≲0.1) star-forming galaxies, and spectral modeling of ASCA, BeppoSAX, and XMM-Newton data, we demonstrate the existence of a linear SFR-L_2–10^HMXB relation which holds over ∼5 decades in X-ray luminosity and SFR. The total 2-10keV luminosity is not a precise SFR indicator because at low SFR (i.e., in normal and moderately-starbursting galaxies) it is substantially affected by the emission of low-mass X-ray binaries, which do not trace the current SFR due to their long evolution lifetimes, while at very high SFR (i.e., for very luminous FIR-selected galaxies) it is frequently affected by the presence of strongly obscured AGNs. The availability of purely SB-powered galaxies - whose 2-10keV emission is mainly due to HMXBs - allows us to properly calibrate the SFR-L_2–10^HMXB relation. The SFR-L_2–10^HMXB relation holds also for distant (z∼1) galaxies in the Hubble Deep Field North sample, for which we lack spectral information, but whose SFR can be estimated from deep radio data. If confirmed by more detailed observations, it may be possible to use the deduced relation to identify distant galaxies that are X-ray overluminous for their (independently estimated) SFR, and are therefore likely to hide strongly absorbed AGNs.