High-resolution CO ( 1 \to 0 ) observations of five ultraluminous infrared galaxies ( ULIGs : L _ { IR } [ 8 - 1000 \mu { m } ] \gtrsim 10 ^ { 12 } L _ { \odot } ) with double nuclei are analyzed .
These sources constitute a complete subset of local ULIGs expected to be in an intermediate stage of merging and selected with projected nuclear separations of 2 \farcs 0 - 5 \farcs 4 ( 3–5 kpc ) so they could be resolved with the Owens Valley Radio Observatory Millimeter Array .
The observed pairs include two mergers with cool far-infrared colors ( 25 \micron to 60 \micron flux density ratio f _ { 25 \mu { m } } / f _ { 60 \mu { m } } < 0.2 ) from the Infrared Astronomical Satellite ( IRAS ) Bright Galaxy Sample ( IRAS 12112+0305 and IRAS 14348-1447 ) and three mergers with warm infrared dust temperatures ( f _ { 25 \mu { m } } / f _ { 60 \mu { m } } \gtrsim 0.2 ) selected from the IRAS Warm Galaxy Sample ( IRAS 08572+3915 , IRAS 13451+1232 = PKS 1345+12 , and IRAS 13536+1836 = Mrk 463 ) .
These ULIGs are further distinguished by the presence of pairs of active nuclei ; among the ten nuclei , nine have Seyfert or LINER classifications and one is unclassified .

Molecular gas is detected only on the redder , more radio-luminous nucleus of the warm objects , whereas both nuclei of the cool ULIGs are detected in CO .
The inferred molecular gas masses for the detected nuclei are 0.1 - 1.2 \times 10 ^ { 10 } M _ { \odot } , and the undetected nuclei have molecular gas masses at least 1.2–2.8 times less than that of their CO-luminous companions .
Upper limits on the extent of the CO emitting regions of each detected nucleus range from 2–4 kpc , which is about 3-6 times smaller than the average effective CO diameter of nearby spiral galaxies .
This is strong evidence that the high concentration of molecular gas is the result of tidal dissipation in ongoing mergers .
There is no correlation between the optical emission-line classification of the nuclei ( i.e. , Seyfert , LINER , or H II ) and the presence of detectable molecular gas ; however , there is a clear indication that the relative amount of molecular gas increases with the relative level of activity as measured via radio power and optical/near-infrared emission-line strength .
Star formation rates are estimated to be in the range \sim 30 - 290 M _ { \odot } year ^ { -1 } nucleus ^ { -1 } by making assumption that the radio and infrared emission arise from supernovae and dust heating by massive stars , respectively ; corresponding gas consumption timescales are 1 - 7 \times 10 ^ { 7 } years .
The nuclei detected in CO are extremely red at near-infrared wavelengths , suggestive of much dustier environments than in the companions undetected in CO .
Column density estimates are N _ { H _ { 2 } } \sim 10 ^ { 24 - 25 } cm ^ { -2 } , which correspond to more than 1000 magnitudes of extinction toward the nuclei at visual wavelengths .
Finally , the molecular gas mass densities and line-of-sight velocity dispersions show significant overlap with stellar densities and line-of-sight stellar velocity dispersions of local elliptical galaxies with M _ { V } < -19 mag , including rapidly rotating ellipticals with disky isophotes and power-law light profiles as well as slowly rotating ellipticals with boxy isophotes and cores .
This provides strong evidence that the CO-rich nuclei of these ULIGs have the phase-space density of gas necessary to form the stellar cores of elliptical galaxies .