A systematic study of ULIRGs using near-infrared absorption bands reveals a strong UV environment in their star-forming regions

We present a systematic study of the 3.0 um H2O ice and the 3.4 um aliphatic carbon absorption features toward 48 local ultraluminous infrared galaxies (ULIRGs) using spectra obtained by the AKARI Infrared Camera to investigate the UV environment in their star-forming regions. All the ULIRGs in our sample exhibit a ratio of optical depth of H2O ice to silicate dust (tau3.0/tau9.7) that is lower than that in the Taurus dark cloud. This implies that ULIRGs cannot be described as an ensemble of low-mass star-forming regions and that a significant amount of high-mass star-forming regions contribute to star-forming clouds in local ULIRGs. The results also show that the ratios of optical depth of aliphatic carbon to silicate dust, tau3.4/tau9.7, exhibit diverse values. We investigate two effects that can affect this ratio: the geometric temperature gradient (which increases the ratio) and the intense UV environment (which decreases it). The geometric temperature gradient is typically considered as a sign of active galactic nuclei (AGN). ULIRGs with AGN signs (optical classification, NIR color, and a PAH emission strength of 3.3 um) indeed tend to exhibit a large tau3.4/tau9.7 ratio. However, we find that the presence of buried AGN is not the only cause of the geometric temperature gradient, because the enhancement of the ratio is also evident in pure starburst-like ULIRGs without these AGN signs. Regarding the intense UV environment in star-forming regions, the correlation between the aliphatic carbon ratio and the ratio of the [C II] 158 um line luminosity to the far-infrared luminosity (L[CII]/LFIR), which represents the UV environment in photodissociation regions, implies that the intense UV environment causes the decrease of the aliphatic carbon ratio. We find that an intense UV environment (G/nH > 3) in star-forming regions is needed for the aliphatic carbon ratio to be suppressed.
Comment: 32 pages, 13 figures, 4 tables, accepted for publication in PASJ