Molecular gas in LIRGs

The molecular gas in Luminous Infrared Galaxies I. CO lines,
extreme physical conditions, and their drivers

Padelis P. Papadopoulos, Paul P. van der Werf, E.M. Xilouris, K.G. Isaak, Yu Gao, Email: padelis@mpifr-bonn.mpg.de    S. Mühle,
Max Planck Institute für Radioastronomie, Auf dem Hügel 69, D-53121 Bonn, Germany
Leiden Observatory, Leiden University, P.O. Box 9513, NL-2300 RA Leiden, The Netherlands
Institute of Astronomy and Astrophysics, National Observatory of Athens, I.Metaxa & Vas.Pavlou str., GR-15236, Athens, Greece
Research and Scientific Support Department, European Space Agency, Keplerlaan 1, 2200 AG, Noordwijk, The Netherlands
Purple Mountain Observatory, Chinese Academy of Sciences, Nanjing, Jiangsu 210008, China
Argelander-Institut für Astronomie, Auf dem Hügel 71, D-53121 Bonn, Germany
Accepted … Received …; in original form …
Abstract

We report results from a large molecular line survey of Luminous Infrared Galaxies (LIRGs: 10 L) in the local Universe (z0.1), conducted during the last decade with the James Clerk Maxwell Telescope (JCMT) and the IRAM 30-m telescope. This work presents the CO and CO line data for 36 galaxies, further augmented by multi-J total CO line luminosities available for other IR-bright galaxies from the literature. This yields a combined sample of N=70 galaxies with the star-formation (SF) powered fraction of their IR luminosities spanning (10-210) , and a wide range of morphologies. Simple comparisons of their available CO Spectral Line Energy Distributions (SLEDs) with local ones, as well as radiative transfer models discern a surprisingly wide range of average ISM conditions, with most of the surprises found in the high-excitation regime. These take the form of global CO SLEDs dominated by a very warm (100 K) and dense (n10 cm) gas phase, involving galaxy-sized ((few)10 M) gas mass reservoirs under conditions that are typically found only for (1-3)% of mass per typical SF molecular cloud in the Galaxy. Furthermore some of the highest excitation CO SLEDs are found in Ultra Luminous Infrared Galaxies (ULIRGs, 10 L), and surpass even those found solely in compact (star-formation)-powered ´hot´-spots in Galactic molecular clouds. Strong supersonic turbulence and high cosmic ray (CR) energy densities rather than far-UV/optical photons or SNR-induced shocks from individual SF sites can globally warm the large amounts of dense gas found in these merger-driven starbursts and easily power their extraordinary CO line excitation. This exciting possibility can now be systematically investigated with Herschel and ALMA. As expected for an IR-selected (and thus (SF rate)-selected) galaxy sample, only few “cold” CO SLEDs are found, and for fewer still a cold low/moderate-density and gravitationally bound state (i.e. Galactic-type) emerges as the most likely one. The rest remain compatible with a warm and gravitationally unbound low-density phase often found in ULIRGs. Such degeneracies, prominent when only the low-J SLED segment (J=1–0, 2–1, 3–2) is available, advise against using its CO line ratios and the so-called =M(H)/(1-0) factor as star-formation mode indicators, a practice that may have lead to misclasification of the ISM environments of IR-selected gas-rich disks in the distant Universe. Finally we expect that the wide range of ISM conditions found among LIRGs will strongly impact the factor, an issue we examine in detail in paper II (Papadopoulos et al. 2012).

keywords:
galaxies: ISM – galaxies: starburst – galaxies: AGN – galaxies: infrared – ISM: molecules – ISM: CO
pagerange: The molecular gas in Luminous Infrared Galaxies I. CO lines, extreme physical conditions, and their driversThe molecular gas in Luminous Infrared Galaxies I. CO lines, extreme physical conditions, and their driverspubyear:

1 Introduction

The population of luminous infrared galaxies (LIRGs), discovered by the Infrared Astronomical Satellite (IRAS) to have bolometric luminosities dominated by the infrared part of their Spectral Energy Distributions (SEDs) (e.g. Soifer et al. 1987), contains some of the most extreme star-forming systems in the local Universe. At these deeply dust enshrouded star-forming systems dominate the luminosity function of the local Universe, and at surpass even optically selected QSOs (Soifer & Neugebauer 1991; Sanders & Ishida 2004 and references therein). Their large reservoirs of molecular gas mass (10-10 M) discovered via CO J=1–0 observations (Tinney et al. 1990; Sanders, Scoville, & Soifer 1991; Solomon et al. 1997), along with clear evidence of strong dynamical interactions and mergers in many LIRGs (e.g. Sanders & Ishida 2004), make these systems unique local examples of dust-enshrouded galaxy formation in the distant Universe (e.g. Smail, Ivison, & Blain 1997; Hughes et al. 1998).

Interferometric imaging of the CO 1-0 line (and occasionally of J=2–1), revealed gas disks with D0.5 kpc and surface densities () , often decoupled from the stellar components of merging galaxies (e.g. Sanders et al. 1988a; Wang et al. 1991; Planesas et al. 1991; Bryant & Scoville 1996, 1999; Downes & Solomon 1998). Naturally the CO(2–1)/(1–0) ratio was the first to be systematically measured (Krügel et al. 1990; Braine & Combes 1992; Horellou et al 1995; Aalto et a. 1995; Albrecht, Krügel, & Chini 2007) even if it is insensitive to the presence of dense and warm gas typical near SF sites. The advent of submm interferometric CO(3–2) imaging revealed the distributions of such gas in LIRGs via the distribution of CO (3–2)/(1–0), (3–2)/(2–1) ratios (e.g. Sakamoto et al. 2008; Wilson et al. 2009; Iono et al. 2007, 2009), but its utility remains limited by dissimilar u-v coverage and lack of zero-spacing information (Iono et al. 2004). Thus single dish measurements of total molecular line luminosities and their ratios remain a primary tool for probing the average molecular gas properties in LIRGs (and a prerequisite for interferometric images that contain all spatial information).

Ideally a combination of low to mid-J rotational lines of heavy rotor molecules with high critical densities such as HCN ((2-40) for J=1–0, 3–2), and CO J+1J lines from J=0, 1, up to at least J2 (33 K,  cm) are necessary to probe the large range of physical properties within GMCs ((15-100) K, (few)(-) cm). Sensitivity limitations and/or lack of multi-beam receivers confined such measurements to few nearby LIRGs and mostly towards their nuclei (e.g. Devereux et al. 1994; Dumke et al. 2001; Meier et al. 2001; Zhu et al. 2003), while heavy rotor molecular lines are faint with HCN(1–0)(1/5-1/10)CO(1–0) even in ULIRGs, and as low as 1/40CO(1–0) in typical spirals (Solomon et al. 1992). Multi-J observations of the luminous CO line emission are not limited by sensitivity rather by the lack of multi-beam receivers and/or beam-matched observations in widely different frequencies. Thus, while several studies use CO(3–2)/(1–0) as a warm/dense gas tracer in substantial LIRG samples (e.g. Mauersberger et al. 1999; Yao et al. 2003; Narayanan 2005; Mao et al. 2011), these are typically confined solely within their nuclear regions. CO J=4–3 or higher-J line observations are even more sporadic, hindered by increasing atmospheric absorptions at 460 GHz, while also confined to the nuclear regions of nearby LIRGs (e.g. White et al. 1994; Güsten et al. 1996; Petitpas & Wilson 1998; Nieten et al. 1999; Mao et al. 2000; Bayet et al. 2006). This is unfortunate since, while the difficulty of measuring reliable CO line ratios increases with wider J-level separations (for the same dish: =), so does their diagnostic power. Nevertheless molecular Spectral Line Energy Distributions (SLEDs) remain the key tool for probing the average state of the molecular gas in galaxies, and for estimating total and star-forming molecular gas masses. Finally local CO SLEDs provide a necessary benchmark for the usually more sparsely sampled ones for galaxies at high redshifts (e.g. Weiss et al 2007), and have been recently used as SF mode (merger versus disk-driven star formation) indicators for gas-rich near-IR selected disks in the distant Universe (Dannerbauer et al. 2009; Daddi et al. 2010).

We used the James Clerk Maxwell Telescope (JCMT111The James Clerk Maxwell Telescope is operated by the Joint Astronomy Centre (JAC) on behalf of the Science and Technology Facilities Council (STFC) of the United Kingdom, the Netherlands Organisation for Scientific Research, and the National Research Council of Canada) on Mauna Kea in Hawaii (USA), and the IRAM 30-meter telescope at Pico Veleta (Spain) to conduct a multi-J CO, HCN line survey of such systems and study the molecular ISM in some of the most extreme star-forming systems in the local Universe. First results regarding the CO J=6–5 lines and dust emission SEDs have already been published (Papadopoulos et al. 2010a,b, hereafter P10a,b). In this paper we present the entire (JCMT)+(IRAM) CO, CO line dataset, further augmented with all reliable measurements of total CO line luminosities of LIRGs from the literature, which yields currently the largest such database assembled for local star-forming galaxies. For a few LIRGs in our sample their CO SLEDs have been extended towards higher-J levels (from J=5–4 up to J=13–12) with the SPIRE/FTS aboard the Herschel Space Observatory.

The layout of this work is as follows: in Sections 2, 3 we describe the sample and the observations, Section 4 contains the data reduction, the literature search, and line flux rectifications for any discrepant values found. In Section 5 we investigate the molecular ISM exitation in LIRGs using comparisons with local environments, and propose a new two-phase ISM model for extreme starbursts, motivated by recent views on SF-powered radiative feedback onto the ISM. In the same saction we also briefly discuss the role of AGN in driving the global ISM excitation. In section 6 one-phase radiative transfer modeling of the CO line ratios is used to systematically extract the average densities, temperatures, and dynamical states of the average ISM environments encountered in our sample, and assemble a comprehensive picture of their range. In section 7 we present the CO SLED excitation range possible from J=1–0 up to J=7–6, and discuss their power sources. There we also discuss the (CO SLED)(ISM state) degeneracies, and their impact on the classification of SF modes in galaxies, using the CO line ratios of recently discovered near-IR selected gas-rich disks at high redshifts as an example. We present our conclusions in Section 8, where we also list a few important questions regarding the expected effects of ISM excitation conditions on molecular gas mass estimates in LIRGs and ULIRGs, the subject of our second paper (Papadopoulos et al. 2012). Throughout this work we adopt a flat -dominated cosmology with =71 km s Mpc and =0.27.

2 The sample

The sample was drawn from two CO J=1–0 surveys of LIRGs by Sanders et al. 1991, and Solomon et al. 1997 (themselves drawn from the IRAS BGS flux-limited sample with 5.24 Jy; Soifer et al. 1987, 1989; Sanders et al. 2003), so that all galaxies in our sample have at least one measurement of this basic H-tracing line. We imposed two additional criteria, namely: a) (the maximum redshift for which the JCMT B-band receivers can tune to CO J=3–2), and b) compact CO-emitting regions (sizes from CO interferometric images) so that single telescope pointings or small maps can record total line fluxes. If CO images were unavailable then radio continuum (Condon et al. 1990, 1996; Crawford et al. 1996), sub-mm dust emission (Lisenfeld, Isaak, & Hills 2000; Mortier et al. 2011) or near-IR images (Zenner & Lenzen 1993; Murphy et al. 1996; Scoville et al. 2000) helped place an upper limit on the size of their CO-bright emission unaffected by extinction. This is because sub-mm emission, sensitive to both the warm dust associated with molecular gas and the cold dust concomitant with the more extended HI (Thomas et al. 2001) sets , while the radio continuum, tied to the far-IR emission through a well-known correlation, traces the star forming regions of LIRGs (Condon et al. 1990) where most of the molecular gas lies.

Year periods spectral lines (K)
1999 06/07–20/07 CO 3–2, 4–3 450-700 (B 3), 1130–2380 (W/C)
2001 10/12–28/12 CO 3–2 380-550 (B 3)
2002 24/02–25/02 CO 3–2 630-690 (B 3)
17/04–18/04 CO 3–2 760, 1100-1700 (B 3)
17/06–30/06 CO 2–1, 3–2 550 (A 3), 430-650, 900-1200 (B 3)
20/11–23/11 CO 4–3 2600-4800, 6000-9300 (W/C)
2003 06/11 CO 3–2 4000-4400 (B 3)
2004 20/01 CO 3–2 530-620 (A 3)
02/04–29/05 CO 2–1, CO 3–2 300-450 (A 3), 470-2400 (B 3)
13/07–25/08 CO 2–1, CO 3–2 320-510 (A 3), 800-1000 (B 3)
28/09–10/11 CO 2–1, CO 3–2 300-600 (A 3), 530-1350 (B 3)
18/11 CO 4–3 2450-2800 (W/C)
2005 20/02 CO 6–5 4500-5200 (W/D)
17/04-23/04 CO 3–2, 4–3, 6–5 720-1100 (B 3), 1600 (W/C), 3700-5500 (W/D)
22/08 CO 2–1 415-420 (A 3)
10/10–28/10 CO 2–1 230-420 (A 3)
15/12–31/12 CO 2–1, CO 3–2 330-490 (A 3), 750-980 (B 3)
2006 15/12–18/12 CO 2–1 250-280 (A 3/ACSIS)
2007 16/12–21/12 CO 2–1 230-280 (A 3/ACSIS)
22/02–24/02 CO 2–1 310-320 (A 3/ACSIS)
2008 09/05 CO 3–2 1630 (HARP-B/ACSIS)
2009 06/01–07/01 CO 6–5 2600-3200, 7500 (W/D, ACSIS)
22/01–25/01 CO 6–5 1700-2600, 9000 (W/D, ACSIS)
01/02 CO 6–5 2200-3500 (W/D, ACSIS)
27/01 CO 6–5 2000-2900 (W/D, ACSIS)
02/03 CO 6–5 1300-1500 (W/D, ACSIS)
13/03–15/03 CO 6–5 1900-3100 (W/D, ACSIS)
2010 11/09 CO 6–5 1400-1800 (W/D, ACSIS)

values include atmospheric absorption. High values of 600 K and 350 K) are measured in a few cases close to tuning range limits (320 GHz for B 3, 215 GHz for A 3), or when 5 GHz, i.e. close to a strong atmospheric absorption feature at .

The high values found only for VIIZw 31, a circumpolar source at the JCMT latitude, observed at elevation of

Table 1: JCMT: observing periods, CO lines, typical system temperatures
Figure 1: The redshift distribution of the combined sample (see section 2).

We have also extensively searched the literature for all local LIRGs whose total CO J+1J fluxes from J+1=1 up to J+1=3 are measured (for higher-J lines such measurements are currently non-existent in the local Universe). Unfortunately this forced the omission of many, often beam-matched, observations (e.g. Devereux et al. 1994; Mauersberger et al. 1999; Mao et al. 2011) as in many such cases CO line emission extends well beyond the nuclear region where such observations were conducted. The combined sample of 70 LIRGs (Table 3) is currently the largest for which their molecular gas can be studied using CO lines which can probe physical conditions from the quiescent to the star forming phase. Its redshift distribution is shown in Figure 1, while its IR luminosity (the SF-powered part) range is (10–210) L (computed over =(8-1000) m). The sample is sparce towards low IR luminosities since galaxies with (1-5)10 L are usually extended (e.g. Tinney et al. 1990) and thus rarely have total CO J+1J fluxes available for J+13. We tried to alleviate this bias by including the few low-L galaxies whose total CO J=1–0, 3–2 line luminosities have been recently measured (Leech et al. 2010).

Telescope 110–115 GHz 210–230 GHz 315–345 GHz 430–461 GHz 620–710 GHz
IRAM 30-m () 6.3  () 7.9, 8.7(
JCMT () 25-28() 28-38() 50-74() 49-62()
NRAO 12-m () 35  () 55()
FCRAO 14-m () 42()
Onsala 20-m () 31()
SEST () 27() 41()
HHSMT () 50  ()
NRO 45-m () 2.4  ()
CSO () 43  ()

The telescope and temperature scale type (see Kutner & Ulich 1981 for definitions).

The frequency range per receiver.

For ==: =4.95 Jy/K, used to obtain the CO 1-0 fluxes from all the IRAM 30-m spectra in the literature that are reported in the scale (unless a different is mentioned).

Measured at 110 GHz (section 2.3.2), also in the NRAO 12-m User’s Manual 1990 Edition, Figure 14.

The scale is sometimes used to report data from the FCRAO 14-m (e.g. Sanders et al. 1986), for which (for ) is adopted.

http://www.ls.eso.org/lasilla/Telescopes/SEST/html/telescopeinstruments/telescope/index.html.

The Heinrich Hertz Submillimeter Telescope (Arizona, USA) (from Narayanan et al. 2005).

For the NRO 45-m telescope in Nobeyama (Japan) at 115 GHz: and adopting , (from the NRO website).

Caltech Submillimeter Observatory (CSO): , where , and yielding (using and , .

Table 2: Point source (Jy/K) conversion factors and HPBW values used for the observations and literature data

3 The observations

We used the A 3 receiver (211-276 GHz, DSB operation) on the 15-meter JCMT on Mauna Kea in Hawaii (USA), to observe the CO J=2–1 (230.538 GHz), and CO J=2–1 (220.398 GHz) lines, and its B 3 (315-370 GHz) and W/C (430-510 GHz) receivers operating single sideband (SSB) for CO J=3–2 (345.796 GHz) and J=4–3 (461.041 GHz) line observations in our sample. The observations were conducted during several periods from 1999 up to 2010 (see Table 1 for specific periods and typical system temperatures). CO J=3–2 observations beyond 2008 utilized the new 16-beam HARP-B SSB receiver (325-375 GHz). The decomissioning of the W/C JCMT receiver before completion of the survey as well as the several CO J=4–3 lines redshifted into the deep 450 GHz atmospheric absorption band meant that such measurements could be conducted for only 10 out of the original sample of 30 LIRGs.

The Digital Autocorrelation Spectrometer (DAS) was used in all JCMT observations until 2006, while the new spectrometer ACSIS was employed aftewards. At 345 GHz we used its 920 MHz (800 km ) or 1.8 GHz (1565 km s) bandwidth mode, depending on the expected line width and the need for maximum sensitivity (i.e. when 920 MHz bandwidth was sufficient to cover the line, dual-channel operation was possible with B 3, and was used for better sensitivity). For the high frequency W/C observations the widest 1.8 GHz bandwidth was used throughout whose 1170 km s velocity coverage adequately covered the FWZI of all the CO 4–3 lines observed. For the CO, CO J=2–1 lines both bandwidth modes were used, yielding (1200–2345) km s velocity coverage. Beam-switching with frequencies of =(1-2) Hz, at throws of 120-180 (in azimuth) ensured flat baselines. The beam sizes were: =22, =14 and =11. We checked and updated the pointing model offsets every hour using continuum and spectral line observations of strong sources, with average residual pointing rms scatter =2.5.

3.1 The CO J=6–5 observations

The first measurements of the CO J=6–5 line (691.473 GHz) were conducted for the luminous ULIRG/QSO Mrk 231 and the LIRG Arp 193 in our sample using the old JCMT W/D band (620-710 GHz) receiver (operating in SSB mode) on 20 of February and 22 of April 2005 respectively, under excellent, dry conditions (0.035). The typical system temperatures were (3700–5500) K (including atmospheric absorption). The DAS spectrometer was used in its widest mode of 1.8 GHz ( at 690 GHz), and beam switching at frequencies of =2 Hz with azimuthal throws of resulted in flat baselines. The beam size at 691 GHz was =. Good pointing with such narrow beams is crucial and was checked every (45-60) mins using differential pointing with the B3 receiver (350 GHz). This allows access to many more suitable compact sources in the sky than direct pointing with the W/D receiver at 690 GHz, and was found accurate to within 2.6 (rms) during that observing period.

The other CO J=6–5 measurements were conducted during 2009, with the upgraded W/D receiver equipped with new SIS mixers (effectively the same type installed at the ALMA telescopes in this waveband) which dramatically enhanced its performance. The resulting low receiver temperatures (550 K) allowed very sensitive observations with typical (1500–3000) K (including atmospheric absorption) for 0.035-0.06. Dual channel operation (after the two polarization channels were aligned to within ) further enhanced the W/D band observing capabilities at the JCMT (see Table 1). The ACSIS spectrometer at is widest mode of 1.8 GHz was used, while in a few cases two separate tunings were used to create an effective bandwidth of 3.2 GHz (1390 km s at 690 GHz) so that it adequately covers (U)LIRG CO lines with FWZI(800–950) km s. Rapid beam switching at =4 Hz (continuum mode) and azimuthal throw of yielded very flat baselines under most circumstances. The pointing model was updated every 45-60 mins using observations of compact sources with the W/D receiver, as well as differential pointing with the A 3 receiver, yielding rms residual error radius of 2.2. The final CO J=6–5 observations were conducted in 2011 during which I Zw 1 was observed, with only one W(D) receiver channel functioning, under dry conditions (0.05) that yielded (1400–1800) K. The same beam-switching scheme was used, while two separate tunings yielded an effective bandwidth of 3.2 GHz covering the wide CO line of this ULIRG/QSO (e.g. Barvainis et al. 1989). The pointing uncertainty remained within the range of previous CO J=6–5 observations. Nevertheless we wish to note that isolated cases of large pointing offsets reducing the observed CO J=6–5 line fluxes have been found (e.g. for Arp 220, see P10a and P10b), and may have affected a few of these highly demanding CO line measurements.

3.2 The IRAM 30-m observations

Observations of CO, CO J=1–0, 2–1 with the IRAM 30-m telescope were conducted during two sessions in 2006 namely, June from 20 to 25, and November from 26 to 28. In both periods the A100/B100 (3 mm) and A230/B230 (1 mm) receivers were used, connected to the 1 MHz (A100/B100, 512 MHz) and 4 MHz (A230/B230, 1 GHz) filterbanks. During the first period the A230/B230 receivers were used to observe the CO J=2–1 line. If the latter was strong and detected in about a hour or less, the 1 mm receivers were then re-tuned to CO J=2–1. For sources with very weak CO lines (e.g. 08030+5243, 08572+3915), no attempt of observing CO J=2–1 was made. Data were acquired under New Control System (NCS) in series of four-minute scans, each comprised of eight 30-sec subscans. The typical system temperatures (including atmospheric absorption) for the CO 2–1 observations were (210-230 GHz)(220–500) K, with the lowest mostly during the CO observations (though for occasional tunings towards the edge of the band and/or bad weather conditions (700-900) K). For most sources data were acquired in two or more different days to ensure a line detection, and as a consistency check. Pointing and focus were checked frequently during the observations with residual pointing errors 3 (rms).

During the November period receivers A100/B100 were used to observe CO J=1–0 line simultaneously to the CO J=2–1 line observed with A230/B230 (tuned to the same line each time). The pointing error stayed 3 (rms), except during November 26 when it went up to 6 (corresponding data were omitted). The typical system temperatures were 110-160 K, (200-380) K, and (210–230 GHz)(330-425) K. Finally, in order to maintain very flat baselines, the wobbler switching (nutating subreflector) observing mode with a frequency of 0.5 Hz and beam throws of 180–240 was employed during both observing sessions. The beam sizes were: =22 and (210-230 GHz)=11, with corresponding beam efficiencies222http://www.iram.fr/IRAMES/index.htm of: =0.75, =0.52 and =0.57, and forward beam efficiencies of =0.95 and =0.91. We also note that in most cases we had redudant CO J=2–1 measurements with the JCMT, and then: a) adopted the average when JCMT/IRAM values agreed to within 20% (most cases), or b) adopted the JCMT measurement (as its wider beam is less prone to flux loss due to pointing offsets and/or beam-throw/flux-loss uncertainties) if a discrepancy larger than the aforementioned was found.

Figure 2: The high-J CO J+1J, J+13 spectra. In the few cases where all three CO J=3–2, 4–3 and 6–5 lines are available we omit the overlay of CO J=2–1 in order to reduce confusion (the J=2–1 lines are all shown in Figure 3). The velocities are with respect to =(LSR) (Table 3), and with typical resolutions (10–50) km s. A common color designated per transition is used in all frames.
Figure 3: continued
Figure 4: continued
Figure 5: continued
Figure 6: The CO, CO line data. The velocities are with respect to =(LSR) (Table 3), and with typical resolutions (35–90) km s. A common color designated per transition is used in all frames.
Figure 7: continued
Figure 8: continued

4 Data reduction, line intensity estimates, literature data

In both telescopes the output spectra are in the scale (see Kutner, & Ulich 1981). We inspected all individual 10min/(4-6)min JCMT/IRAM spectra for baseline ripples and to clip any intensity “spikes” in individual channels. The edited spectra were then co-added using a -weighting scheme and linear baselines were subtracted from each final co-added spectrum. These spectra are shown in Figures 2 and 3 and were used to derive the velocity-integrated molecular line flux densities from

(1)

where =15.62, =3.905 and is the aperture efficiency defined against the scale (=, where is the aperture efficiency measured against the scale, as is more typical, and is the rearward spillover and scattering efficiency, Rohlfs & Wilson (1996), Eqs 8.16, 8.17). The factor , with x= and =source diameter, accounts for the geometric coupling of the beam (its gaussian part) to a disk-like source, when a CO emission size was available, and where is the pointing error radius (see 4.1). The total point-source conversion factors adopted for the JCMT, the IRAM 30-m telescope, and all the data gleaned from the literature (for the corresponding output antenna temperature scales) are comprehensively tabulated in Table 2.

4.1 Aperture efficiencies, line intensity uncertainties, biases

Aperture efficiencies of large high-frequency sub-mm telescopes such as the JCMT can change significantly (especially for 460 GHz) depending on a variety of factors (e.g., elevation, thermal relaxation of the dish or its re-shaping after an holography session). In order to track them over an decade of observations (during which the JCMT dish has been re-adjusted quite a few times) we conducted frequent aperture efficiency measurements using planets and adopted the average obtained per observing period for deriving the line fluxes of all the sources observed during it. In many cases, as a cross-check, we distributed the measurements of very CO-luminous LIRGs over several widely separated periods, during which very different aperture efficiencies (sometimes up to a factor of 2) were often measured. In all such cases Equation 1, with the appropriate values, yielded velocity-integrated line fluxes in excellent agreement. Indicatively most aperture efficiencies measured for the JCMT lay within 0.41–0.56 (B-band, 315-350 GHz), and 0.21–0.31 (C-band, 430-461 GHz). For the three periods of the more demanding W/D band observations we derived =0.25 (2005), 0.32 (2009), 0.27 (2010) from planetary measurements. The uncertainties for the reported velocity-integrated line flux densities have been computed from

(2)

where is the stochastic error of the average line intensity (averaged over the line FWZI ), is the telescope efficiency factor (used to derive the integrated line flux from the temperature scale of the output spectrum, e.g., for )333More precisely the 3nd term in Equation 2 embodies the uncertainty of total efficiency factor needed to convert the scale (corrected only for atmospheric extinction) to a final fully corrected (apart from source-beam coupling) (or ) scale., and its uncertainty. The first term is estimated from the spectra shown in Figures 2, 3 using

(3)

where is the stochastic intensity dispersion, estimated from the line-free part of the spectrum (for a given velocity channel width ), while = and = are the number of channels within the line FWZI and the line-free baseline (with channels symmetrically around the line) respectively. The second term in Equation 2 accounts for line calibration errors (due to a host of factors such as imprecise knowledge of the calibration loads, uncertainties in the atmospheric model and the derived extinction etc). Observations of numerous strong spectral line standards and planets during each observing period yielded intensity dispersions of 15% ( and ), 20% (), and 25% (), which we adopt as the combined calibration (cal) and uncertainties per observing band at the JCMT. For the 30-m we consider these to be 15% for both 3 mm and 1 mm bands.

Finally, even with the accurate tracking and pointing achievable by enclosed telescopes such as the JCMT, the residual rms pointing errors and the narrow beams of large mm/sub-mm telescopes at high frequencies can lead to a substantial and systematic reduction of measured fluxes of compact sources. We try to account for this as described in P10a, by applying a scaling factor to the line fluxes of all point-like sources with CO emission region diameters of (where =2.5 and =3 are the pointing error radii). This factor is (see P10a):

(4)

where is the beam HPBW, is the rms pointing error per pointing coordinate. For , is replaced in Equation 1 by as the beam-source coupling correction is overtaken by the pointing error correction. At 345 GHz and 460 GHz for the JCMT we obtain =1.087 and =1.15, (with neglible correction at 230 GHz), while for the 30-m =1.20 (and negligible correction at 115 GHz). The CO J=6–5 observations with HPBW8 are those most succeptible to this bias and has been estimated from the pointing rms per observing session, yielding a range =1.17–1.37. For sources with CO (or sub-mm dust emission) sizes of the factor is used in Equation 1. Finally in the cases where large offsets were found between the presumed CO source center and the observed positions in the literature (see discussion in 4.2) we applied a ´beam-shift´ correction factor of = where is the (beam center)-source offset (see Table 4).

4.2 Incorporating data from the literature, the final dataset

A detailed literature search for all total CO and CO line fluxes available for LIRGs enlarged our sample (see Table 3), while allowing also a consistency check using the duplicate measurements per object (especially for the J=1–0 transition). In most cases we find good agreement among the various CO J=1–0 fluxes reported in the literature, and between the (much fewer) reported J=2–1, 3–2 line fluxes and our measurements, within the expected uncertainties444In the literature the uncertainties of mm/sub-mm line measurements are often underestimated, with only the thermal rms error reported (the first term in the expression in Equation 2). In all such cases we assumed a 15% of calibration and uncertainty in addition to the one reported.. Most cases of serious discrepancies among CO line fluxes were rectified after accounting for the different source positions often used. This typically occured because of the positional uncertainties inherent in the optical identification of heavily dust-obscured sources within the large IRAS position error ellipse (e.g. Solomon et al. 1997) and/or the better CO positions obtained for some LIRGs (e.g. via interferometry) and used for subsequent CO observations after the original J=1–0 detections with usually large (45-50) beams were made (e.g. Sanders et al. 1991; Young et al. 1995). Moreover, even when a particular LIRG is optically identified, the multi-component/interacting nature of many such systems can confound the choice of a pointing center for CO single dish observations (e.g. Leech et al. 2010). In the cases where we found different positions used for CO observations reported in the literature, we “shift/scale” the corresponding line fluxes to a common position, the interferometrically-derived CO emission peak of the LIRG (or its near-IR and/or radio continuum peak if the former was unavailable) assuming a source much smaller than the beams used (the case for most LIRGs). This rectified many of the discrepant CO line fluxes, with only a few inconsistent ones remaining, mostly between new data and much older CO observations (mainly those reported in the Sanders et al. 1991 CO J=1–0 survey). These are likely the result of improved calibration techniques (e.g. good sideband rejection), pointing accuracy, and faster beam-switching available for mm/sub-mm telescopes now than in the past.

4.2.1 Resolving a large discrepancy: new CO J=1–0 observations of IRAS 05189–2524

The largest remaining CO line flux discrepancy is for the LIRG IRAS 05189–2524 with = obtained for CO J=1–0 with the NRAO 12-m telescope (Sanders et al. 1991) and = with SEST (Strong et al. 2004). The larger value would place this ULIRG to the lowest end of gas excitation of the entire sample, with =0.24 and =0.32, typical for the coldest most quiescent GMCs found in the Galaxy and M 31 (Loinard et al. 1995; Allen et al. 1995; Fixsen et al. 1999). We used the 12-meter telescope555The Kitt Peak 12 m telescope is operated the Arizona Radio Observatory (ARO), Steward Observatory, University of Arizona, to re-observe the CO J=1–0 line in this ULIRG on the nights of 12, 13 and 24 March, 2008. The final co-added CO J=1–0 spectrum (Figure 4) was used to estimate a velocity-integrated line flux of =, in good agreement with the value reported by Strong et al. 2004 but less than half that reported by Sanders et al. 1991. We report the average of ours and the Strong et al. 2004 value (Table 4).

Figure 9: The CO J=1–0 spectrum of IRAS 05189-2524, centered at cz=12800 km s, obtained with the 12 m telescope (see section 4.2.1).
Name RA (J2000) Dec (J2000) () Refs
00057+4021 00 08 20.58 40 37 55.5 0.0445 (194.5) (co) 1
00322–0840 (NGC 157) 00 34 46.48 08 23 47.8 0.0055 (23.3) (co,sm,x) 2
00509+1225 (I Zw 1, PG 0050+124) 00 53 34.92 12 41 35.5 0.0611 (270.3) (co) 3,4
01053–1746 (Arp 236) 01 07 47.00 17 30 24.0 0.0200 (85.8) (sm,x) 5,6
01077–1707 01 10 08.20 16 51 11.0 0.0351 (152.3) (sm) 5,6
01418+1651 (III Zw 35) 01 44 30.50 17 06 08.0 0.0274 (118.2) (sm) 5,6
02071+3857 (NGC 828, VI Zw 177) 02 10 09.43 39 11 26.3 0.0178 (76.2) (co) 7,8
02080+3725 (NGC 834) 02 11 01.55 37 40 01.3 0.0154 (65.8) (cm) 9,10
02114+0456 (Mrk 1027) 02 14 05.60 05 10 27.7 0.0297 (128.3) (sm,x) 5,6
02321–0900 (NGC 985, Mrk 1048) 02 34 37.74 08 47 14.7 0.0430 (187.7) (co) 11
02401–0013 (NGC 1068) 02 42 40.74 00 00 47.6 0.0037 (13.3) (co,sm,x) 12
02483+4302 02 51 36.01 43 15 10.8 0.0514 (225.8) (co) 1
02512+1446 (UGC 2369) 02 54 01.80 14 58 14.0 0.0312 (135.0) (sm,cm,x) 5,6,13
03359+1523 03 38 46.90 15 32 55.0 0.0353 (153.2) (cm) 5,13
04232+1436 04 26 04.94 14 43 37.9 0.0796 (356.4) (cm,sm,x) 14,6,13
05083+7936 (VII Zw 031) 05 16 46.51 79 40 12.5 0.0543 (239.0) (co) 1
05189–2524 05 21 01.11 25 21 45.9 0.0427 (186.4) (cm,ir) 9,15,13
08030+5243 08 06 50.10 52 35 05.4 0.0835 (375.4) (ir) 14,16
08354+2555 (NGC 2623, Arp 243) 08 38 24.10 25 45 16.5 0.0185 (79.3) (co) 17
08572+3915 09 00 25.41 39 03 54.1 0.0582 (256.9) (co) 18
09126+4432 (Arp 55) 09 15 54.90 44 19 54.4 0.0399 (173.3) (co,cm,dbl) 5, 13,19
09320+6134 (UGC 05101) 09 35 51.53 61 21 11.6 0.0393 (171.1) -(ir,co) 14,15,20
09586+1600 (NGC 3094) 10 01 26.00 15 46 14.0 0.0080 (34.0) (ir,cm) 21,22,13
10039–3338 (IC 2545) 10 06 04.50 33 53 03.0 0.0341 (147.9) (sm) 5,6
10035+4852 10 06 45.83 48 37 46.1 0.0648 (287.5) (sm) 14,6
10173+0828 10 20 00.19 08 13 34.5 0.0489 (214.4) (co,cm) 23,13
10190+1322 10 21 42.60 13 06 54.4 0.0765 (342.2) (co,dbl) 24
10356+5345 (NGC 3310) 10 38 45.90 53 30 11.7 0.0033 (14.0) (co,sm,x) 2
10565+2448 10 59 18.15 24 32 34.4 0.0428 (188.2) (co) 1
11191+1200 (PG 1119+120) 11 21 47.12 11 44 18.3 0.0500 (219.4) (co) 25
11231+1456 (IC 2810, UGC 6436) 11 25 45.00 14 40 36.0 0.0341 (147.9) (cm,sm) 21,13,26
11257+5850 (Arp 299) 11 28 32.45 58 33 45.8 0.0103 (43.8) (co,sm,x) 27,6
12001+0215 (NGC 4045) 12 02 42.30 01 58 38.0 0.0066 (28.0) (cm) 21,13
12112+0305 12 13 45.77 02 48 39.3 0.0727 (324.4) (co,dbl) 9,18
12224–0624 12 25 03.90 06 40 53.0 0.0263 (113.4) (cm) 21,13
12243-0036 (NGC 4418) 12 26 54.70 00 52 39.0 0.0073 (31.0) (cm) 21,13
12540+5708 (Mrk 231) 12 56 14.21 56 52 25.1 0.0422 (184.1) (co) 1
13001–2339 13 02 52.10 23 55 19.0 0.0215 (92.3) (ir) 5,22
13102+1251 (NGC 5020) 13 12 39.90 12 35 59.0 0.0112 (47.7) (cm) 21,13
Arp 238 (UGC 08335) 13 15 30.20 62 07 45.0 0.0315 (136.3) (cm,dbl) 5, 10
13183+3423 (Arp 193) 13 20 35.32 34 08 22.2 0.0233 (100.2) (co) 1
13188+0036 (NGC 5104) 13 21 23.10 00 20 32.0 0.0186 (79.7) (cm) 21,13
13229–2934 (NGC 5135) 13 25 43.97 29 50 01.3 0.0136 (58.0) (cm) 10
13362+4831 (NGC 5256) 13 38 17.90 48 16 41.0 0.0278 (120.0) (cm, sm) 5,13,6
13428+5608 (Mrk 273) 13 44 42.12 55 53 13.5 0.0378 (164.4) (co) 1
13470+3530 (UGC 8739) 13 49 14.20 35 15 23.0 0.0168 (71.9) (cm,x) 21,13
F13500+3141 (3C 293) 13 52 17.82 31 26 46.4 0.0446 (194.9) (co) 28
F13564+3741 (NGC 5394) 13 58 33.60 37 27 13.0 0.0125 (53.3) (cm) 21,13
14003+3245 (NGC 5433) 14 02 36.00 32 30 38.0 0.0145 (61.9) (cm) 21,13
14151+2705 (Mrk 673) 14 17 21.00 26 51 28.0 0.0366 (159.0) (opt) 5
14178+4927 (Zw 247.020, Mrk 1490) 14 19 43.20 49 14 12.0 0.0256 (110.3) (cm) 21,13
14280+3126 (NGC 5653) 14 30 10.40 31 12 54.0 0.0119 (50.7) (cm) 21, 13
14348–1447 14 37 38.32 15 00 22.7 0.0825 (370.7) (co,dbl) 29,13
15107+0724 (Zw 049.057) 15 13 13.07 07 13 32.0 0.0129 (55.0) (co) 23
15163+4255 (Mrk 848, Zw 107) 15 18 06.20 42 44 42.0 0.0402 (175.1) (sm,cm) 5, 6, 13
15243+4150 (NGC 5930, Arp 090) 15 26 07.90 41 40 34.0 0.0089 (37.8) (cm) 21,13
15322+1521 (NGC 5953p, Arp 091) 15 34 32.30 15 11 38.0 0.0065 (27.6) (cm) 21,13
15327+2340 (Arp 220) 15 34 57.24 23 30 11.2 0.0182 (78.0) (co) 1
15437+0234 (NGC 5990) 15 46 16.50 02 24 56.0 0.0128 (54.6) (cm) 21,13
16104+5235 (NGC 6090, Mrk 496) 16 11 40.70 52 27 25.0 0.0292 (126.1) (cm) 5,13
16284+0411 (MCG +01-42-008) 16 30 56.50 04 04 59.0 0.0245 (105.5) (cm) 21,13
16504+0228 (NGC 6240) 16 52 59.05 02 24 05.8 0.0243 (104.6) (co) 7,30
17132+5313 17 14 20.48 53 10 31.4 0.0507 (222.6) (cm) 31,13
17208–0014 17 23 21.92 00 17 00.7 0.0428 (186.8) (co) 1
Table 3: Observational parameters of the combined LIRG sample

Observational parameters of the combined LIRG sample Name RA (J2000) Dec (J2000) () Refs 18425+6036 (NGC 6701) 18 43 12.27 60 39 10.5 0.0132 (56.3) (cm) 9,10 19458+0944 19 48 15.47 09 52 01.3 0.1000 (454.8) (ir) 14,16 20550+1656 (II Zw 96) 20 57 23.70 17 07 44.0 0.0363 (157.7) (sm,x) 5,6 22491–1808 22 51 49.86 17 52 24.4 0.0773 (346.0) (ir,cm,x) 9,15,13 23007+0836 (NGC 7469) 23 03 15.60 08 52 26.3 0.0163 (69.7) (co) 32 23365+3604 23 39 01.25 36 21 08.4 0.0644 (285.6) (co)      1 IRAS name and the most common alternative(s), the asterisk marks sources for which CO line fluxes were obtained from an extensive literature search and line flux rectification process (see sections 2, 4.2).

Source coordinates () used for CO observations.

The redshift used for receiver tuning (or the reported in the literature for LIRGs not in the original sample), and the corresponding luminosity distance in Mpc.

Offsets between expected CO source center (assumed coincident with peak cm continuum) and observed position: = (=0 when available CO or sub-mm images defined the CO source center, see 4.2). Thus marks the true CO source center whenever 0.

CO region angular size =, from interferometric maps. If these were not available then cm, near-IR, or sub-mm continuum images were used to set upper limits on . The qualifiers for the images used are: (co), (cm), (ir), (sm) for CO, cm, IR, and sub-mm images, (x)=complex source morphology ( then denotes the overall source size), (dbl)=double CO-bright nuclei. In the latter case refers to the largest one, and the second number denotes their separation.

References used for source position (=p) and morphology/size (=m) information: 1=Downes & Solomon 1998; 2=Zhu et al. 2009; 3=Schinnerer et al. 1998; 4=Eckart et al. 1994; 5=Leech et al. 2010; 6=Mortier et al. 2009; 7=Wang et al. 1991; 8=Casoli et al. 1992; 9=Sanders et al. 1991; 10=Condon et al. 1996; 11=Appleton et al. 2002; 12=Papadopoulos & Seaquist 1998a; 13=Condon et al. 1990; 14=Solomon et al. 1997; 15=Scoville et al. 2000; 16=Murphy et al. 1996; 17=Bryant & Scoville 1999; 18=Evans et al. 2002; 19=Sanders et al. 1988; 20=Wilson et al. 2008; 21=Yao et al. 2003; 22=Zenner & Lenzen 1993; 23=Planesas et al. 1991; 24=Gracia-Carpio et al. 2007; 25=Evans et al. 2001; 26=Lisenfeld et al. 2000; 27=Aalto et al. 1997; 28=Evans et al. 1999; 29=Evans et al. 2000; 30=Tacconi et al. 1999; 31=Young et al. 1995; 32=Davies et al. 2004;33=Crawford et al. 1996

Double nuclei in near-IR (Scoville et al. 2000), but only one is CO-bright (Evans et al. 2002).

Table 4: continued
Name CO J=1–0 CO J=2–1 CO J=3–2 Refs.
00057+4021 (1.087, G) x,1
00322–0840 2
00509+1225 (1.11, K) (1.11, K) (1.27, K) x,3
01053–1746 4,5
01077–1707 4,5
01418+1651 4,5
02071+3857 x,6
02080+3725 (1.052,K) 6,7
02114+0456 4,5
02321–0900 (1.387, K) (2.087, K) x,8,9
02401–0013 10
02483+4302 (1.087, G) x,1,4,11
02512+1446 4, 5
03359+1523 4,7
04232+1436 x,11
05083+7936 (1.087, G) x,1,11,12
05189–2524 (1.10, K) (1.27, K) x
08030+5243 (1.087, G) x,11
08354+2555 7,13,14,15
08572+3915 x,11,16
09126+4432 (3.08, K) x,4,7,14,17
“ “ (SW nucleus) x, 4
“ “ (NE nucleus) x, 4
09320+6134 (1.087, G) x,11
09586+1600 (1.086,K) (1.086,K) x,6,18
10039–3338 4, 5
10035+4852 (1.20, K) (1.20, K) x,11
10173+0828 (1.087, G) x,7,18
10190+1322 (1.06, K) x,4,19
10356+5345 2
10565+2448 x,1,4,7,11,18,20
11191+1200 (1.045, K) x,21
11231+1456 (1.07, K) 7,18
11257+5850 x,4,5,15,22
12001+0215 18
12112+0305 (1.45, K) (1.26, K) x,7,16
12224–0624 18
12243–0036 18
12540+5708 23
13001–2339 4,5
13102+1251 18
Arp 238