AGN are cooler than you think: The intrinsic far-IR emission from QSOs

Abstract

We present an intrinsic AGN spectral energy distribution (SED) extending from the optical to the submm, derived with a sample of unobscured, optically luminous (νLν,5100 > 1043.5 erg s−1) QSOs at z < 0.18 from the Palomar Green survey. The intrinsic AGN SED was computed by removing the contribution from stars using the 11.3 μm polycyclic aromatic hydrocarbon (PAH) feature in the QSOs’ mid-IR spectra; the 1σ uncertainty on the SED ranges between 12 and 45 per cent as a function of wavelength and is a combination of PAH flux measurement errors and the uncertainties related to the conversion between PAH luminosity and star-forming luminosity. Longwards of 20 μm, the shape of the intrinsic AGN SED is independent of the AGN power indicating that our template should be applicable to all systems hosting luminous AGN (νLν, 5100 or L X(2--10keV)LX(2--10keV) We present an intrinsic AGN spectral energy distribution (SED) extending from the optical to the submm, derived with a sample of unobscured, optically luminous (νLν,5100 > 1043.5 erg s−1) QSOs at z < 0.18 from the Palomar Green survey. The intrinsic AGN SED was computed by removing the contribution from stars using the 11.3 μm polycyclic aromatic hydrocarbon (PAH) feature in the QSOs’ mid-IR spectra; the 1σ uncertainty on the SED ranges between 12 and 45 per cent as a function of wavelength and is a combination of PAH flux measurement errors and the uncertainties related to the conversion between PAH luminosity and star-forming luminosity. Longwards of 20 μm, the shape of the intrinsic AGN SED is independent of the AGN power indicating that our template should be applicable to all systems hosting luminous AGN (νLν, 5100 or L X(2--10keV)LX(2--10keV) ≳ 1043.5 erg s−1). We note that for our sample of luminous QSOs, the average AGN emission is at least as high as, and mostly higher than, the total stellar powered emission at all wavelengths from the optical to the submm. This implies that in many galaxies hosting powerful AGN, there is no ‘safe’ broad-band photometric observation (at λ < 1000 μm) which can be used in calculating star formation rates without subtracting the AGN contribution. Roughly, the AGN contribution may be ignored only if the intrinsic AGN luminosity at 5100 AA is at least a factor of 4 smaller than the total infrared luminosity (LIR, 8 H u SX > u-B uL 1043.5 erg s−1). We note that for our sample of luminous QSOs, the average AGN emission is at least as high as, and mostly higher than, the total stellar powered emission at all wavelengths from the optical to the submm. This implies that in many galaxies hosting powerful AGN, there is no ‘safe’ broad-band photometric observation (at λ < 1000 μm) which can be used in calculating star formation rates without subtracting the AGN contribution. Roughly, the AGN contribution may be ignored only if the intrinsic AGN luminosity at 5100 AA is at least a factor of 4 smaller than the total infrared luminosity (LIR, 8–1000 μm) of the galaxy. Finally, we examine the implication of our work in statistical studies of star formation in AGN host galaxies.