We performed systematic observations of the Hi Brα line (4.05 μm) in 51 nearby (z<0.3) ultraluminous infrared galaxies (ULIRGs), using AKARI near-infrared spectroscopy. The Brα line is predicted to be the brightest among the Hi recombination lines in ULIRGs with visual extinction higher than 15 mag. We detected the Brα line in 33 ULIRGs. In these galaxies, the relative contribution of starburst to the total infrared luminosity (LIR) is estimated on the basis of the ratio of the Brα line luminosity (LBrα) to LIR. The mean LBrα/LIR ratio in LINERs or Seyferts is significantly lower (~50%) than that in Hii galaxies. This result indicates that active galactic nuclei contribute signicantly (~50%) to LIR in LINERs, as well as Seyferts. We also estimate the absolute contribution of starburst to LIR using the ratio of star formation rates (SFRs) derived from LBrα (SFRBrα) and those needed to explain LIR (SFRIR). The mean SFRBrα/SFRIR ratio is only 0.33 even in Hii galaxies, where starburst is supposed to dominate the luminosity. We attribute this apparently low SFRBrα/SFRIR ratio to the absorption of ionizing photons by dust within Hii regions.

We present an analysis of the relation between star-formation (SF) and accretion luminosities of local type-2 active galactic nuclei (AGNs) at 0:01 ≤ z < 0:22. We match type-2 AGNs found in the Sloan Digital Sky Survey to current far-infrared (FIR) survey catalogues based on AKARI and Herschel. Estimating AGN luminosities from [O iii]5007 and [O i]6300 emission lines, we find a positive linear trend between FIR and AGN luminosities over a wide dynamical range. This result appears to be inconsistent with recent reports that low-luminosity AGNs show no correlation between FIR and X-ray luminosities; this contradiction is likely due to Malmquist and sample selection biases. Moreover, we also nd that pure- AGN candidates, for which the FIR radiation is thought to be AGN-dominated, show significant low-SF activities. These AGNs hosted by low-SF galaxies are rare in our sample. However, it is possible that the low fraction of low-SF AGN is caused by observational limitations, as recent FIR surveys are not suffcient to examine the population of high-luminosity AGNs hosted by low-SF galaxies.