FIRST VIEWS OF A NEARBY LIRG: STAR FORMATION AND MOLECULAR GAS IN IRAS 04296+2923

IRAS 04296 was observed using the Palomar Hale 5 m telescope⁷ on 2005 January 23 UT with the Wide-field Infrared Camera (WIRC; Wilson et al. 2003) at prime focus, in broadband J and narrowband Pa β filters. The camera has a pixel size of 025. Exposure times on source were 11 minutes (3 × 20 s frames, 11 pointings) for the continuum band and 18 minutes (3 × 60 s frames, 6 pointings) in the emission-line band, with equivalent off-source exposure times to measure the sky. The observing conditions were good with seeing at 10 FWHM. Median dark current and filtered sky images created for each band were subtracted from the frames. Finally, a coadded mosaic was produced from each set of on-source frames. We scaled the counts in each band's mosaic, via aperture photometry of several bright, uncrowded stars in the field around the galaxy, and subtracted the J-band mosaic (see Figure 1(a)) from the Pa β mosaic, to produce an approximate continuum-subtracted image (see Figure 1(b)). The images in the broadband and narrowband filters were not explicitly flux calibrated using observations of standard stars, so we are unable to derive a value for the net Pa β flux in the continuum-subtracted image.

Zoom In Zoom Out Reset image size

IRAS 04296 was observed on 2002 December 24 with the LWS in imaging mode on the Keck I Telescope, with the standard 11.7 μm/SIC filter. The standard stars were α Aur, α Aries, and β Leo, observed both before and after the source exposures. High quality images at both wavelengths were obtained (Figure 2); however, the absolute flux calibration was uncertain due to intermittent instrumental problems, and so we reobserved the galaxy on January 31, with the same parameters as confirmation. Absolute flux calibration is estimated to be good to 10%, based on derived flux counts on standard star before/after observations throughout the night. Stellar images were about 5 pixels or 04 FWHM.

Zoom In Zoom Out Reset image size

The radio data were taken with the NRAO VLA.⁸ There were two projects: AB1092 on 2003 July 7 with the A configuration for high angular resolution at λ= 20 cm, 6 cm, 3.5 cm, 2 cm, and 1.3 cm (L, C, X, U, and K bands) and AT309 on 2005 July 5 with the array in BC configuration for moderate resolution at 1.3 cm only. 3C48 was the primary calibrator for both runs. The 20, 6, and 3.6 cm data were processed with AIPS, including baseline corrections. The 2 cm and 1.3 cm data were observed in fast-switching mode and reduced according to the AIPS prescriptions for high frequency calibration. The details of the radio imaging are discussed below (Section 3.2).

Simultaneous aperture synthesis observations of the ¹²CO(1–0) transition (115.271 GHz) and the ¹³CO(1–0) transition (110.201 GHz) of IRAS 04296 were made with the Owens Valley Radio Observatory (OVRO) Millimeter Interferometer between 2003 November 1 and 2004 May 8. The interferometer consisted of six 10.4 m antennas with cryogenically cooled SIS receivers (Scoville et al. 1994). Observations in "Compact," "Low," and "High" configuration were obtained, with system temperatures (single sideband) ranging from 220 to 450 K at 115 GHz. 64× 4 MHz channels were used to cover both transitions, giving a velocity resolution of 10.5 km s⁻¹ for CO(1–0). The phase center is within 05 of the 2 μm peak listed in Table 1 and the adopted systemic velocity (LSR) was 2100 km s⁻¹. Simultaneous low spectral resolution, wideband (128 × 31.25 MHz) COBRA observations were used to generate a 3 mm continuum image (ν_o≃ 111 GHz), free from ¹²CO(1–0), CN(1–0), and ¹³CO(1–0) contamination. The data were calibrated using the MMA software package. Phase calibration was done by observing the quasar J0336+323 every 25 minutes. Absolute flux calibration was done using Uranus as the primary flux calibrator and 3C273 and 3c454.3 as secondary flux calibration, and is good to ∼10%.

Mapmaking was done in MIRIAD and subsequent data analysis and manipulation was done with the NRAO AIPS package. Maps were generated with natural weighting to maximize sensitivity, uniform weighting to maximize resolution and robust = 0 weighting to simultaneously optimize sensitivity and resolution. Unless stated the analysis results from the naturally weighted data. Integrated intensity images are straight moment 0 maps with all emission brighter than 1.3σ (1σ for ¹³CO) per channel included. The OVRO primary beam is ∼64''at 115 GHz. Corrections for the primary beam have not been applied so emission at the edge of the primary beam will be somewhat underestimated. Structures extended on scales larger than ∼52'' for CO(1–0) will be resolved out by the interferometer. No single-dish observations of this galaxy exists, so no estimate of the amount of resolved-out flux is possible for the millimeter-wave images. However, 52'' corresponds to spatial scales of 7.3 kpc and given that the emission must be extended on this scale in an individual channel to be missed, we consider it unlikely that significant flux is resolved out. If large amounts of flux are missed then the observed molecular richness is a lower limit.

The Palomar J-band continuum image and a narrowband continuum-subtracted image of Pa β in IRAS 04296 are shown in Figure 1. The near-infrared continuum at J band reveals the stellar population. IRAS 04296 is an inclined, barred spiral with outer ring-like spiral arms (SBb/c(r)). The total extent of the emission is roughly 2–25, or 17–20 kpc (8–10 kpc in radius). The stellar distribution appears bi-symmetric and undistorted. The Paβ image shows line emission to be very concentrated and bright in the galactic nucleus, with some very weak emission along the inner arms. The compact nature of the bright Pa β emission shows that the starburst is highly concentrated at the nucleus. Since the images are not photometric, and extinction is probably variable across the galaxy, we cannot precisely quantify the relative brightnesses of disk and nuclear star formation from the Pa β image.

Mid-IR continuum emission traces warm (∼ few hundred K) dust heated by young stars in the starburst. The high-resolution (03) mid-IR continuum image at 11.7 μm made with LWS on Keck of IRAS 04296 is shown in Figure 2; the field of view of this image is only 10'' × 10'', or about 1.4 kpc on a side. A similar 18.75 μm image (not shown) will be published in a subsequent paper (J. L. Turner et al. 2010, in preparation). The 11.7 μm filter contains two polycyclic aromatic hydrocarbon (PAH) features; however, based on observations of the starburst galaxies NGC 7714 and Arp 220 (Brandl et al. 2004; Armus et al. 2007) we estimate that at most 15% of the 12 μm continuum is due to PAH emission.

The mid-IR emission comprises a bright source, which cannot be resolved at the 03 diffraction limit of Keck at 11.7 μm, and a secondary source or tail extending to the east of the main source, separated from the main source by ∼15 (∼200 pc). The total flux density is $S_{11.7\,\mu \rm {m}} = 680$ mJy. These two compact, mid-infrared sources account for roughly half of the IRAS 12 μm flux of the entire galaxy.

VLA maps of IRAS 04296 across the centimeter waveband are presented in Figure 3, with fluxes and noise values given in Table 2. The radio continuum emission is a combination of thermal bremsstrahlung emission from the H ii regions and non-thermal synchrotron emission from supernovae and their remnants associated with the starburst. The brightest radio emission, like the mid-IR emission, is concentrated within the central 1''–2'' region of ∼150–250 pc extent; higher resolution maps shown below give a better comparison with the Keck mid-IR images.

Zoom In Zoom Out Reset image size

Table 2. Infrared and Radio Fluxes

λ	rms	Beam (Robust = 0)	θmax	Peak Flux	Total Mappeda,b	Cut Fluxb,c	Cut Fluxb,c
	(mJy beam−1)	(''×'', p.a)	''	(mJy beam−1)	(mJy)	$(u,v)_{\rm min} = 18\rm k\lambda$	$(u,v)_{\rm min} = 50\rm k\lambda$
						(mJy)	mJy
20 cm	0.11	1.60 × 1.35, 82.8	38	32	130 ± 10	71 ± 5	...
6 cm	0.04	0.48 × 0.39, −83.3	10	3.8	59 ± 5	28 ± 3	...
3.6 cm	0.034	0.25 × 0.21, 88.2	7	1.5	25 ± 5	14 ± 2	11 ±2
2 cm	0.085	0.18 × 0.138, −53.9	4	1.0	10 ± 1	...	10 ± 2
1.3 cm	0.068	0.1 × 0.09, −80.39	2	0.5	7 ± 1.5	...	7 ± 1.5
1.3 cm BC array	0.16	0.91 × 0.30, 76	20	3.7	23 ± 5	18 ± 3	...
2.7 mm	0.75	4.6 × 3.8, 8	52	9.9	11 ± 2	...	...
11.7 μm	5.1 counts s−1	0.4	n.a.	71 ± 10	680 ± 100	...	...

Notes. ^aBased on "uncut" maps with full native A configuration (u, v) coverage. ^bQuoted uncertainties in the fluxes are based on three times the rms times $\sqrt{N}$ times the number of beams corresponding to the source size; they do not include resolved-out flux, or the estimated 5% (10% at 2 and 1.3 cm) absolute flux scale uncertainty. Large beam fluxes do not exist for wavelengths other than 20 cm, so we cannot estimate amounts of resolved-out flux except at 20 cm, where it is negligible. Fluxes are measured within λ/B_min for each map. ^cBased on "cut" maps restricted in (u, v) coverage. The 18 kλ fluxes correspond to the maps of Figure 3, and the 50 kλ fluxes to those of Figures 5 and 6. See Section 3.2.

Download table as:  ASCIITypeset image

Separating free–free emission from the H ii regions from the synchrotron emission is possible because these sources have different spectral and spatial characteristics. However, complications arise when comparing aperture synthesis maps at different frequencies, so the spectral analysis must be done with care. The synchrotron component in galaxies is spatially extended with a spectrum that falls steeply with increasing frequency. Free–free emission from H ii regions is compact with spectra that are either flat or rising with frequency (e.g., Turner & Ho 1994). Except at 1.3 cm, the robustly weighted VLA maps of Figure 3 were done with a single array configuration, so the maps have responses to extended structure that vary with frequency; shorter wavelength maps are less sensitive to extended emission. The shortest baselines determine the maximum detectable spatial scales, θ_max ∼ λ/B_min, which are listed in Table 2. So the maps of Figure 3 cannot be compared directly to obtain spectral index maps. Since single dish or lower resolution VLA fluxes do not exist for IRAS 04296 except at 20 cm, we cannot estimate the effects of missing short baselines except at 20 cm. Our A array maps, with total flux $S_{20\,\rm cm} = 140\pm 7$ mJy, recover the entire 20 cm flux recorded by Condon et al. (1996), and so the 20 cm map is probably a good representation of the total radio emission, although there may be faint extended emission that is undersampled. From the 20 cm image we can see that the extended radio emission is aligned along the apparent bar of the galaxy (Section 3.3). The bright mid-IR sources and the peak of the radio emission mark a starburst located at the inner end of the northwestern arm traced by CO (Section 3.3).

Free–free and non-thermal synchrotron sources can be separated by determining the spectral index, α, of the emission (S ∝ ν^α) so long as the beams are matched, not only in beam size (θ_min), but also in θ_max, defined by u_min = B^EW_min/λ, v_min = B^NS_min/λ. Toward this end we have made a series of "cut" maps, which are high-pass spatially filtered images with common θ_min and θ_max, that enable us to construct images with nearly identical (u, v) coverages and allow us to compare the radio spectra for the compact sources, at the cost of losing response to extended emission. Since the star formation arises largely in compact sources ≲1'' (150 pc), we expect to detect nearly all of the free–free emission in the "cut" maps and resolve out most of the synchrotron emission, which is our goal.

The first set of "cut" maps were made to match the beams at 20, 6, 3.6, and 1.3 cm, with a consistent (u, v)_min of 18kλ (θ_max ∼ 10''; 1.4 kpc). The beam for this series of maps is 082 × 029, position angle (p.a. −89°). For the 1.3 cm emission, we used the BC configuration data. From the cut maps, we mapped the spectral index α, shown in Figure 4. The spectral index is negative over the entire range, confirming that the extended emission centimeter-wave emission is largely non-thermal, typical of large spiral galaxies and LIRGs (e.g., Condon et al. 1991; Condon 1992; Clemens et al. 2008). Fluxes from this set of maps are measured for a 45 × 3'' box (630 pc × 420 pc) and are shown in Table 2.

Zoom In Zoom Out Reset image size

A second set of "cut" maps were made from the shorter wavelength data to image the compact emission at 3.6, 2, and 1.3 cm. For these maps, $(u,v)_{\rm min} = 50\, \rm k\lambda$ (θ_max∼ 4'', or ∼ 550 pc) and they were convolved to match the 027 × 023, p.a. −90° beam (θ_min) of the 3.6 cm map. These cut maps are shown in Figure 5, and fluxes are given in Table 2. The peak flux densities are the same within observational uncertainties for all three maps, 1.1–1.5 mJy beam⁻¹. The (u, v) restriction has eliminated about two-thirds of the total 3.6 cm flux as compared to the map of Figure 3. The compact emission that remains in these maps has a nearly flat spectrum: this is emission from H ii regions. These radio images are similar to the 11.7 μm image of Figure 2, given the lower resolution (03) of the mid-IR image. We identify the brightest, elongated radio source with the mid-IR peak, and the extension visible to the southeast in the 3.6 cm image (Figure 5) with the extension in the mid-IR image.

Zoom In Zoom Out Reset image size

A third set of "cut" maps, designed to match beams at 2 and 1.3 cm, gives the highest resolution images. Like the previous set of maps, they are cut to $(u,v)_{\rm min} = 50\,\rm k\lambda$, but these maps are instead convolved to match the smaller 2 cm beam of 015 × 014, p.a. −66°. These images resolve the brightest continuum source (Figure 6) into a double source separated by 02, or 30 pc oriented east–west. The eastern source is stronger. Slight differences in the spatial structure of the double source at the two frequencies might be due to differences in azimuthal (u, v) coverage or to 1.3 cm seeing. This source is flat in radio spectral index, indicating that the compact emission is largely free–free emission from H ii regions. Comparison of the three sets of maps suggests that the total amount of flat spectrum, compact emission in the nucleus of IRAS 04296 is ∼10 mJy.

Zoom In Zoom Out Reset image size

Continuum emission at 2.7 mm was detected with OVRO (Figure 7d). The peak 2.7 mm continuum intensity is 10 ± 1 mJy beam⁻¹ (46 × 37) with a total detected flux of 11 ± 2 mJy (Table 2). The millimeter continuum emission is confined to the nuclear starburst, with position and extent consistent with the (uncut) centimeter-wave and infrared continuum. At this wavelength, potential continuum sources include synchrotron, free–free emission, and dust. Extrapolation of synchrotron emission all the way from 20 cm to 2.7 mm is dangerous since it is sensitive to the exact spectral index, but using the 20 cm emission seen in the first "cut" map and α = −0.8 implies a contribution of 2–3 mJy. Based on our estimated nuclear gas mass (Section 4.2) we expect that ∼1–2 mJy of the 3 mm continuum is from dust emission. From this, we conclude that the millimeter continuum flux associated with star formation is ≳5–7 mJy at 2.7 mm, consistent with the fraction of the flux from the compact flat-spectrum sources seen directly in centimeter-wave continuum.

Zoom In Zoom Out Reset image size

The first observations of molecular gas toward IRAS 04296 are shown in Figures 7 and 8. The CO(1–0) integrated intensity (Figure 7) is remarkably bright, extending over a 2.5 kpc (diameter) region, peaking at the nucleus. (When no superscript appears the most abundant from ¹²C¹⁶O is implied; "¹³CO" refers to the less abundant ¹³C substituted isotopologue.) The CO emission extends well beyond the Pa β emission region (Figure 1(b)) and the radio continuum (Figure 8). Beyond the nucleus the molecular gas is bar-like, extended southeast–northwest. Weak emission is also seen to the north and west of the nucleus. The CO morphology closely follows the stellar morphology seen in the Palomar J-band and Two Micron All Sky Survey (2MASS) K_s images, as shown in Figure 7.

Zoom In Zoom Out Reset image size

Toward the nucleus, antenna temperatures in the robustly weighted image (Figure 8) peak at 10 K averaged over the inner 160 pc radius. This is a remarkably high brightness temperature for a galaxy at the distance of IRAS 04296; it implies that the beam filling factor for the CO must be near unity over regions 300–400 pc in diameter. The CO line widths are broadest toward the nucleus but are fairly modest (250 km s⁻¹; FWHM) compared to other luminous starbursts (Solomon et al. 1992, 1997; Downes & Solomon 1998), and drop to ∼50 km s⁻¹ toward the outer molecular arms. The velocity field is consistent with a rotating disk, except for modest perturbations (≲50 km s⁻¹; see below). We have fit the axisymmetric velocity field with a Brandt rotation curve; the best-fit parameters are given in Table 1. The separate CO component north of the nucleus follows the same velocity field as does the bar and nucleus. This suggests that these clouds are a continuation of the galaxy's disk (the outer spiral arms).

The molecular morphology is bar-like. The best fit for the kinematic major axis from the CO velocity field is at a position angle of 252° (to the receding axis). This is perpendicular to the long axis of the CO and IR morphology. Therefore, the observed bar-like structure seen in CO is not a result of inclination, but is intrinsic. The strong bar deduced for the stellar distribution makes it impossible to separately constrain the bar strength and the galaxy's inclination angle, i, over the inner arcminute. Instead we estimate the inclination angle from the axial ratio of the outer, presumably more circular, spiral arms seen in the J-band image (Figure 1(a)). The measurement yields i = 50 ±  4° and is consistent with the same p.a. determined from the CO velocity field. Such an inclination angle implies a peak velocity of ∼190 km s⁻¹, or a dynamical mass of $1.6\times 10^{9}\, \it M_{\odot }$ over the 7'' (R < 500 pc) and $3\times 10^{10}\, \it M_{\odot }$ over the radius, R = 30'' (R < 4.3 kpc). This suggests that IRAS 04296 is similar in size to the Milky Way at this radius. H i observations at larger radii will be needed to determine a global dynamical mass for IRAS 04296.

To verify that the gas distribution and kinematics are consistent with response to a bar potential, we generate a simple analytical, weak-bar model. These models treat the gas dissipation by adding a damping term proportional to radial velocity to the standard stellar barred orbits (e.g., Wada 1994; Lindblad & Lindblad 1994). Despite the simplicity of the model, it matches full hydrodynamical simulations with surprising fidelity (e.g., Lindblad & Lindblad 1994). The current model is that of Sakamoto et al. (1999), except that we use an axisymmetric potential that generates a Brandt rotation curve. With the analytical model we can quickly search barred galaxy parameter space for configurations consistent with what is observed.

Figure 9, which shows the bar models and rotation curve, indicates a close agreement of models with the observed morphology and kinematics. From the best models we find that the molecular gas distribution of IRAS 04296 is very well reproduced by a large axial ratio bar. We assume that the northern portion of the galaxy is the near side, consistent with two fairly straight trailing arms on the leading edge of the bar. The observed velocity residuals have the expected signatures of inflow along the downstream sides of the bar arms and weaker outflow on the upstream (so-called spray region) sides of the arms. Both modeled radial velocities and peak observed velocity residuals are in the range of V_r ∼ 10 − 40 km s⁻¹.

Zoom In Zoom Out Reset image size

From the observed free–free emission (Section 3.2) we can infer the Lyman continuum rate of the starburst, its L_IR, and star formation rate (SFR). The compact centimenter-wave emission of IRAS 04296 consists of an east–west elongated source, centered at R.A. = 04^h32^m486, decl. = 29°29'57''. The brightest emission corresponds to a region 04 × 02 in size (50 pc × 30 pc) with a flux of ∼3 mJy and a flat spectrum, located about 1'' (140 pc) WSW of the dynamical center. There is a halo of emission, stronger to the south of the double source (Figure 3). The emission measure, EM, of these H ii regions is high; our high-resolution maps (Figures 5 and 6) detect only gas with EM>10⁶–$10^7 \rm cm^{-6}\, pc$, which is characteristic of dense or compact H ii regions in the Galaxy. The central double radio source has a total flux ∼7 mJy at 1.3 cm, and ∼10 mJy at 2 and 3.6 cm. The short wavelength (2–0.26 cm) continuum emission is consistent with star formation having a free–free flux of ∼5–7 mJy, which implies an ionizing flux from young stars of $N_{{\rm Lyc}} \sim 6 \, \pm \, 2 \, \times \, 10^{53} \rm \,s^{-1}$ for optically thin emission. This total Lyman continuum rate is equivalent to 1.4 ± 1 × 10⁵ O7 (Martins et al. 2005) stars. Using STARBURST99 (Leitherer et al. 1999; Vázquez & Leitherer 2005) and a Kroupa initial mass function (IMF) with mass cutoffs of 0.1 and 100 M_☉, we infer a luminosity associated with the OB stars of $L_{{\rm OB}}\sim 2 \pm 0.7 \times 10^{10}\,\it L_\odot$ for the nuclear (R < 150 pc) starburst, to within a factor of 2, given uncertainties in free–free flux and starburst age. The mid-IR continuum flux gives a consistent picture: the 12 μm/radio flux ratio of 100 is similar to the values of 50–100 observed in both Galactic and extragalactic H ii regions (e.g., Genzel et al. 1982; Ho et al. 1989). The thermal centimeter-wave fluxes imply an SFR of ${\sim} 12\pm 6 \,M_\odot \,\rm yr^{-1}$ and a mass of $2 \pm 1\times 10^7\,\it M_\odot$ for the nuclear starburst. From the morphology, we infer that these are two massive young super star clusters, "hyper star clusters," each containing 50,000 O stars and with individual masses of ∼10⁷ M_☉. The nuclear starburst contains about half of the current global SFR of IRAS 04296, based on the ratio of LWS flux for the nuclear source to the IRAS 12 μm flux for the entire galaxy.

IRAS 04296 is very bright in CO. Its total mapped CO luminosity is L$_{{\rm CO}} = 1.4 \times 10^{9}\,\rm K\, km\,s^{-1} \, \rm pc^{2}$. The nucleus is the dominant feature in the CO maps. The peak integrated CO(1–0) intensity averaged over the central kpc (R = 35) is higher than the peak CO brightness averaged over individual giant molecular cloud (GMC) scales in the nearby gas-rich spiral, IC 342 (e.g., Meier et al. 2000). Among nearby galaxies, only NGC 2146, also an LIRG, and possibly NGC 253, can rival it in the richness of its molecular emission (Jackson & Ho 1988; Young et al. 1988).

The bright CO implies a large H₂ mass. For a Galactic conversion factor of $X_{{\rm CO}}^{{\rm gal}} = 2 \times 10^{20} \rm cm^{-2}(\rm K\, km\, s^{-1})^{-1}$ (e.g., Strong et al. 1988; Hunter et al. 1997), we would derive a molecular gas total mass of $M_{{\rm H}_{2}}(R < 500\,\mbox{pc}) = 1.6 \times 10^{9} \,M_{\odot }$ for the central kpc, which is larger than the molecular mass of the entire Milky Way. This mass is comparable to the dynamical mass for the central kpc, as determined by the rotation curve from our model (Figure 9). Either molecular gas accounts for the entirety of the dynamical mass in the inner 500 pc radius or the CO-to-H₂ conversion factor is different than in the Galaxy. Evidence is accumulating that the latter explanation is far more likely in the center of a gas-rich galaxy such as IRAS 04296. Studies of a variety of galaxies indicate that X^gal_CO as determined from the molecular clouds in the disk of our Galaxy is inappropriate for the central regions of galaxies. In (U)LIRGs, it is well known that the Galactic conversion factor consistently overestimates (nuclear) molecular masses by factors of 3–4 (e.g., Downes et al. 1993; Solomon et al. 1997; Yao et al. 2003; Narayanan et al. 2005). This is often called the "starburst" conversion factor. However, in fact, this lower conversion factor also holds in gas-rich centers of normal spiral galaxies including relatively quiescent ones such as our own (e.g., Mauersberger et al. 1996; Dahmen et al. 1998; Meier & Turner 2001, 2004; Weiss et al. 2001; Paglione et al. 2001; Meier et al. 2008). Since X_CO is a dynamical measure of mass, any contribution to the cloud line widths from systematic motions driven by the stellar potential will cause an overestimate of the gas mass. This is particularly likely in galactic centers, where tidal effects on the clouds become important (e.g., Meier & Turner 2004).

Our ¹³CO map provides an independent measure of gas mass, which we can obtain from this optically thin line by "counting molecules." If we adopt abundances of [¹²CO/¹³CO] = 60, (e.g., Wilson & Rood 1994), and [¹²CO/H₂] = 8.5 × 10⁻⁵ (Frerking et al. 1982), and an excitation temperature, T_ex = 20 K, which are typical of clouds in the nuclei of nearby galaxies, then N(H₂) column densities can be estimated from ¹³CO line strengths. The peak nuclear column densities derived from ¹³CO are 3.6 times lower than the corresponding values obtained from ¹²CO using the Galactic X_CO. Our ¹³CO mass suggests that $X_{{\rm CO}}^{{\rm nuc}} \simeq 0.6 \times 10^{20}\, \rm cm^{-2} \, (\rm K\, km\, s^{-1})^{-1}$ in IRAS 04296, similar to values found for other spiral nuclei, LIRGs, and ULIRGs.

We conclude that the nuclear H₂ mass (including He) for IRAS 04296 is $M_{H_{2}}(R< 500\,\mbox{pc}) \simeq 4.3 \times 10^{8} \,M_{\odot }$ ($\Sigma _{H_{2}} \simeq 600 \,M_{\odot } \rm \,pc^{-2}$) for the inner kpc, with estimated uncertainties of ∼50% due to unknown ¹³CO excitation temperature and abundance. Caution should be used in comparing this value to masses determined for other galactic centers in the literature in which the Galactic X_CO is used to determine the mass; those masses are probably systematically overestimated.

Outside the nucleus, the CO intensity in the disk remains high in absolute terms. Normalized by distance squared, the CO intensity of IRAS 04296 averaged over its central 45'' aperture is larger than any galaxy observed in the FCRAO survey (Young et al. 1995) out to its 30 Mpc distance. Over the central arcminute of the OVRO field of view a total of M_gas(R < 30'') = 6 × 10⁹ M_☉ of molecular gas is implied (for a Galactic conversion factor including He). Even adopting the lower M82 conversion factor (Walter et al. 2002) for this rather normal appearing disk, IRAS 04296 has twice the molecular mass of M82, including all outflows and streamers. ¹³CO(1–0) is weakly detected toward the disk in IRAS 04296, so further constraints on the validity of the conversion factor are testable here as well. At 10'' radii, the values of N(H₂) derived from ¹³CO are within a factor of 1.6 of what the standard conversion factor predicts, and along the arms at distances larger than 10'' (particularly at the bar ends), the two column densities agree within the uncertainties. This suggests that X_CO in the disk of IRAS 04296 may be consistent with Galactic disk values. However, gradients in excitation temperature, ¹³CO opacity, and isotopic abundances can affect the ¹³CO mass by at least a factor of 2, so this merits further study.

The IR luminosity of IRAS 04296 is ∼9.8 × 10¹⁰ L_☉ (Sanders et al. 2003). Averaged over the central arcminute, IRAS 04296 has L_IR/L_CO = $72 \, L_{\odot } (\rm K\, km\, s^{-1}\,\rm pc^{2})^{-1}$, L_IR/$M_{{\rm H}_{2}}$ = 18 L_☉/ M_☉,  and M(H₂/M$_{{\rm H\,\mathsc {i}}}$) ≳ 5, all typical of LIRGs (e.g., Scoville et al. 1991; Sanders & Mirabel 1996). Given the rotation curve (Section 3), an inclination of 50° and the molecular gas masses derived above, molecular gas accounts for at least 30% of the dynamical mass within the central 500 pc radius (using the lower ¹³CO mass) and 15% of the dynamical mass over the entire mapped region.

Is the molecular gas disk in IRAS 04296 so dense that it will collapse into new stars without further triggering, or is the disk stable? We can assess this using the Toomre Q parameter

$$\begin{equation} Q = \frac{\alpha \kappa \sigma }{\pi G \Sigma _{\rm gas}} = \frac{\Sigma _{\rm crit}}{\Sigma _{\rm gas}}, \end{equation} \tag{ 1 }$$

where κ is the epicyclic frequency (e.g., Binney & Tremaine 1987), σ is the gas velocity dispersion, Σ_gas is the total gas surface density, and α is a constant of order unity depending on the structure of the galactic disk (here assumed to be α = 1; Safronov 1960; Toomre 1964). If Q ≲ 1 the gas disk is unstable.

Figure 10 displays the observed values of Σ_gas (calculated both from ¹²CO and ¹³CO, neglecting the H i surface density contribution, which if conservatively estimated to be distributed uniformly over the disk contributes much less than 10% everywhere within R = 175), σ, κ, and Q as a function of galactocentric radius. The data have been averaged in azimuth assuming a disk inclined at 50° to the line of sight. For ¹²CO and the Galactic conversion factor, Q ≃ 1 for the inner 5'' and rises to 3–4 by the end of the bar. This dependence is driven primarily by the rapid drop in gas surface density just outside the nucleus together with the fact the velocity dispersion of the gas remains fairly large (σ ≃ 20–40 km s⁻¹) along the entire molecular bar. At first glance, this is consistent with gravitational instabilities driving the starburst. But note that this depends on the overestimated H₂ mass from the standard conversion factor. If we use Σ_gas derived from optically thin ¹³CO emission, a rather different picture appears. In this case, derived Qs would be effectively constant with radius over much of the bar and the disk would be stable everywhere. This suggests that the starburst is not a result of gravitational instabilities in a disk alone and possibly that the large shear/non-circular motions in the strong bar act to inhibit star formation in the arms. The morphology of CO and the starburst's location at the inner terminus of the bar (Figure 8) argue for bar induced cloud–cloud collisions as inflowing molecular gas piles up in the nucleus.

Zoom In Zoom Out Reset image size