arXiv:1504.01453v1 [astro-ph.GA] 7 Apr 2015 ences. ne,8 adnRa,Saga 000 China. 200030, Shanghai Road, Nandan 80 ences, pc xlrto,Nnig209,China. 210093, Nanjing Exploration, Space ehooy 20E aionaBv,Psdn,C 12,US 91125, CA Pasadena, Blvd, California E. 1200 Technology, 035 hrotsil,V 20,USA. 22904, VA Charlottesville, 400325, etBiigRa,Nnig200,China. 210008, Nanjing Road, Beijing West 2 evtr,BakodHl,EibrhE93J UK 3HJ, EH9 Edinburgh Hill, Blackford servatory, igUiest) iityo dcto,Nnig209,C 210093, Nanjing Education, of Ministry University), jing ajn 103 China. 210093, Nanjing H RM3MTLSOE RMI SUPPORTED IS IGN AND IRAM (GERMANY) MPG TELESCOPE. (FRANCE), (SPAIN). 30M INSU/CNRS BY IRAM THE al. 10- et values (Bolatto ∗ value Way showing Milky the methods, gas than larger molecular other times 100 the from with inferred comparing mass by constrained been Yug&Soil 92,wietetoms abundant most mass two gas the molecular H while renders the components, 1982), emission CO of Scoville bright tracer & gas. its common funda- (Young of most and the out the radiation, to CO form by related gas thus how cools is about (Goldsmith emission question regime gas CO mental density molecular The and cooling temperature and 2001). in low collapse role the important monoxide gas in an carbon for molec- plays species, of prerequisite (CO) molecular cooling Among the efficient is formation. The gas 2004). ular Solomon & Gao 01,dw o1%Slr(lere ta.21) The H 2013). to 2007, al. CO et al. from (Elmegreen et Leroy Solar factor conversion 15% 1998; to al. down et has metal- Taylor 2011), lower emission 1997; increasingly CO (Israel with the licity decade, in past detected molecular the been Over of temperature clouds. characteristic gas the at emission lcrncades [email protected] address: Electronic H EKCRO OOIEEISO NA XRML EA POOR METAL EXTREMELY AN IN EMISSION MONOXIDE CARBON WEAK THE AE NOSRAIN ARE U WITH OUT CARRIED OBSERVATIONS ON BASED 6 5 4 3 9 8 7 2 1 rpittpstuigL using 2018 typeset 9, Preprint August version Draft tr omoto oeua lus(enct 1998; (Kennicutt clouds molecular of out form Stars nttt o srnm,Uiest fEibrh oa Ob Royal Edinburgh, of University Astronomy, for Institute e aoaoyfrRdoAtooy hns cdm fSci- of Academy Chinese Astronomy, Radio for Laboratory Key hnhiAtooia bevtr,CieeAaeyo Sci of Academy Chinese Observatory, Astronomical Shanghai olbrtv noainCne fMdr srnm and Astronomy Modern of Center Innovation Collaborative eateto srnm,Uiest fVrii,PO Box P.O. Virginia, of University Astronomy, of Department nrrdPoesn n nlssCne,Clfri Insti California Center, Analysis and Processing Infrared upeMuti bevtr,CieeAaeyo Sciences, of Academy Chinese Observatory, Mountain Purple e aoaoyo oenAtooyadAtohsc (Nan- Astrophysics and Astronomy Modern of Laboratory Key colo srnm n pc cec,NnigUniversity, Nanjing Science, Space and Astronomy of School ogShi Yong hsidctsta Oeiso seceigywa neteeymtlp galaxies. metal headings: these extremely Subject in in weak coolant exceedingly a is as emission role CO that indicates This ptal eovdds asmp The map. mass dust resolved spatially ekC =- msini en hc sarayfiteog oplace em me to CO enough in for faint role already search is similar to which a Se seen, m , ( plays is poor 30 factor emission CO metal conversion IRAM J=1-0 extremely if CO nearby with unclear weak a observation A is in deep peak it far-IR a but brightest out galaxies carried spiral We in gas molecular abnmnxd C)i n ftepiaycoat fgsada ea an and gas of coolants primary the of one is (CO) monoxide Carbon ,,3 1,2, 2 n e antb bevddrcl in directly observed be cannot He, and uziWang Junzhi , 1. A INTRODUCTION T E tl mltajv 03/07/07 v. emulateapj style X aais wr umliee:IM–glxe:ISM galaxies: – ISM submillimeter: – dwarf galaxies: α CO rmteC uioiyt h oeua a asta sdrvdfrom derived is that mass gas molecular the to luminosity CO the from ) 4,5 2 h-uZhang Zhi-Yu , rfre as (referred rf eso uut9 2018 9, August version Draft α arn Stierwalt Sabrina CO α 6 hina. ueof tute uGao Yu , CO has ) ABSTRACT A. sa es ee ude ie h ik a value. Way Milky the times hundred seven least at is - - 7,5,3 et.W edottesetaeey2mnts while minutes, 2 every ( spectra temperature the system out typical read the ob- We quasi-stellar hours emitting 2 jects. millimeter observations. every strong the about nearby for checked with used were was focusing throwing and Pointing beam Hz 0.5 at aetefeunycanlsaigo 9 H.The a KHz. with 195 mode of which switching Receiver spacing used, wobbler Mixer channel were standard Eight frequency backend Trans- The the Fourier (FTS) gave the far-IR 1. Spectrometers and the form is dual-polarization Fig. frequency, with in this (EMIR) shown at arcsec size as 25th 22 beam Aug peak is at IRAM which to telescope region, the 23th target millimeter as The meter Aug. Spain. 30 Veleta, IRAM from Pico the done using were 2014, J2000) A -04:42:04.70, DEC: in 10:11:06.55, (RA: region tion orglxe Bltoe l 03.Tedt r pre- are data The 2013). al. metal in most et in sented the (Bolatto measurements arguably mass galaxies gas 2014), poor cold al. for et way (Shi spatially accessible the map through dust estimated is de- resolved molecu- been It whose have the galaxies masses galaxy. poor gas increases metal the lar extremely in proximity few a emission 2004; of Its CO one Pagel potential & of (Pettini 2005). tectability Solar al. 7% et Kniazev of abundance early oxygen in and locally molec- both of galaxies tracer poor Universe. effective metal and in coolant gas can emission ular gas CO efficient if an establishes be metallicities emis- lower CO at Probing sion environments. metallicity low in 2013) § pcrmadol sdtesetawith spectra the the each with used checked conducted We only was and package. software spectrum reduction GILDAS Data in CLASS band. this at eprtr fe mohn h rqec eouinto beam ( resolution main MHz in frequency about 6.2 mK the 4.1 hours, out smoothing about 5 after throwing of about temperature level was After noise time the source gave on which discussion. total the final spectra, 50% for K 280 ,rsetvl.Cnlsosaein are Conclusions respectively. 4, e Armus Lee , h CO The etn sadafirglra . p with Mpc 1.4 at irregular dwarf a is A Sextans 9 § ∼ J .Tersl n icsinaein are discussion and result The 2. 6k s km 16 10osrain oad h trforma- star the towards observations =1-0 tn ,wt %Slrmetallicity. Solar 7% with A, xtans 8 ereHelou George , 2. o aais hlegn its challenging galaxies, oor − togcntan nthe on constraint strong a OBSERVATIONS 1 ). iyacsil rcrof tracer accessible sily sinb agtn the targeting by ission AAY ETN A SEXTANS GALAXY, a ordwarfs. poor tal 8 isegGu Qiusheng , T sys § a bu 8 K 280 about was ) 5. T ± sys the 120 esthan less ,,3 1,2, § ′′ and 3 offset , ∗ 2 Shi et al.

10-1

10-2

IRAM (Jy) density flux 10-3

10-4

101 102 wavelength (µm) 30" Fig. 3.— The Infrared SED of Sextans A within the IRAM beam is fitted to derive the dust mass. Red symbols are the Spitzer and Fig. 1.— False-color, multi-wavelength images of Sextans A. Herschel photometric points with 1-σ error bars. The blue solid Red is the sum of Herschel 160 and 250 µm data, green is GALEX line indicates the best-fit by the dust model (Draine & Li 2007). far-UV, and blue is radio 21 cm data showing the HI atomic gas. The IRAM 30 m pointing at CO J=1-0 with the beam size is 104 indicated by the white circle, while the yellow circle indicates the Sandstrom+13 star-forming disk defined at a surface brightness of 25.9 ABmag Leroy+11 −2 Israel+97 ] Sextans A arcsec in the GALEX far-UV band. 3 1 10 Israel+97 fitting

) Glover & Mac Low 2011 1 WLM Narayanan+12 Wolfire+10 Feldmann+12 102 (Kkms 2 pc ⊙ 101 [M CO α

100

7.2 7.6 8.0 8.4 8.8 9.2 12+log(O/H) Fig. 2.— The submm spectrum around the CO J=1-0 frequency with source cz=324 km/s. The red Gaussian profile is the Fig. 4.— The conversion factor (αCO) from the CO lumi- fit to the signal at 350 km/s, which gives signal-to-noise ratio of 3.7 nosity to the molecular gas mass for Sextans A is compared to for the integrated flux. The two black dotted lines are the 1-σ noise those at higher metallicities. Symbols are observations in the levels. The blue curve indicates the velocity of the HI emission line literature (Israel 1997; Leroy et al. 2011; Sandstrom et al. 2013; within the IRAM beam as measured from the HI data but with Elmegreen et al. 2013) where molecular gas masses are all based the absolute intensity scaled arbitrary (Ott et al. 2012). on the spatially resolved dust maps except for WLM (see text for The Herschel data at 70, 160, 250 and 350 µm, and the the detail). Lines are models, either empirical (Israel 1997), or theoretical models (Wolfire et al. 2010; Glover & Mac Low 2011; Spitzer data at 3.6, 4.5, 5.6, 8.0 and 24 µm data are from Narayanan et al. 2012; Feldmann et al. 2012). Shi et al. (2014) and Dale et al. (2009), respectively. Be- cause the dust emission within the IRAM beam is par- probability of 0.05% for the signal to be artificial, which tially resolved instead of a simple point spread function, corresponds to a 3.4 σ significance for a Gaussian dis- to measure the photometry within the IRAM beam, we tribution of noises. The feature lies within the veloc- first convolved all images to the 350 µm resolution based ity range of atomic gas as shown in Fig. 2. The off- on the convolution Kernels (Aniano et al. 2011). Aper- set in the velocity between the two lines may be reason- ture photometry at all bands were then measured at this able as the two trace different phases of ISM. As listed spatial resolution. The corresponding aperture correc- in Table 1, the intensity ICO(J=1−0) is 145±39 mK km −1 tion factor is taken as the ratio of the 70 um photometry s in Tmb, corresponding to a point source luminosity at its native resolution to that at the 350 µm resolution (Gao & Solomon 2004) of ∼ (3.7±1.0)×103 K km s−1 by assuming that the 70 µm map at its native resolution pc2 for a beam size of 22 arcsec, much smaller than the can resolve the dust structure. We further included the previous upperlimits (Taylor et al. 1998) of a few×104 K far-ultraviolet (UV) images from the GALEX Space Tele- km s−1 pc2 observed at the position (RA=10h11m07.3s, scope archive as well as maps of atomic gas (Ott et al. dec=-4d42m22.5s, J2000) near our pointing with a beam 2012). size of 55 arcsec. The 3-σ CO J=2-1 assuming the same velocity dispersion as CO J=1-0 is 460 mK km/s. 3. RESULTS The conversion factor αCO from CO to H2 is quanti- Fig. 2 shows the observed CO J=1-0 spectrum at a fied by comparing the CO luminosity to the molecular gas resolution of 16 km/s with cz= 324 km/s. A marginal mass that is inferred from the spatially resolved dust map signal at 350 km/s was seen. A Gaussian fitting gives (Israel 1997; Leroy et al. 2011; Sandstrom et al. 2013). S/N of 3.7 for the integrated flux, a width of 21±12 km/s As shown in Fig. 3, the infrared SED is first fitted with and peak intensity of 6.3 mK while the 1-σ noise is 1.4 the model of Draine & Li (2007) as detailed in Shi et al. mK. The distribution of the noise level is slightly non- (2014), based on which the total cold gas (HI+H2) within Gaussian. Using the negative values of channels over a the IRAM beam is measured by multiplying the dust velocity range from -500 to 1500 km/s, we estimated a mass with the gas-to-dust ratio derived from the diffuse CO in Sextans A 3

TABLE 1 The Properties Of The Sextans A with the IRAM CO (1-0) Beam.

dust dust ICO(J=1−0) ICO(J=2−1) LCO(J=1−0) Mdust MH2+HI MH2 αCO(J=1−0) SFR L8−1000µm Mstar 2 7 7 2 −4 5 5 mK km/s mK km/s K km/s pc M⊙ 10 M⊙ 10 M⊙ M⊙/pc /(K km/s) 10 M⊙/yr 10 L⊙ 10 M⊙ 145±39 < 460 3670±990 771±184 1.1±0.3 1.0±0.3 2750±1110 2.2±0.4 1.7±0.4 1.7±0.4

5 regions of the galaxy. As discussed in Shi et al. (2014), sured to be 1.7×10 L⊙. So the CO luminosity in K −1 2 the derived total cold gas mass is not sensitive to the km s pc relative to the infrared luminosity in L⊙ is absolute dust mass but to the ratio of dust masses of about 1:45, which is slightly larger than spiral galaxies star-forming and diffuse regions. As a result, as long as (Genzel et al. 2010). This is partly caused by the rarity diffuse and star-forming regions have similar dust grains, of both dust and CO in extremely metal poor galaxies different dust models gives a similar estimate of cold gas while different excitation mechanisms for CO and dust masses within the IRAM beam. To derive the gas mass of must also play some roles. If comparing to the star star-forming regions, we do assume star-forming regions formation rate (SFR) as measured from the far-UV lu- have the same gas-to-dust ratio as the diffuse regions, minosity (Kennicutt 1998), the SFR per CO luminosity which is true for spiral galaxies (e.g. Sandstrom et al. is about ten times higher than star-forming non-merger 2013). However, this assumption may break down in massive galaxies (Genzel et al. 2010). If scaling by the dwarfs as their star formation history is stochastic in stellar mass as measured from Spitzer 3.6 and 4.5 µm both space and time so that both diffuse and star-forming measurements (Eskew et al. 2012), the CO luminosity regions are affected by different physical processes that per stellar mass of Sextans A within the IRAM beam impact the gas-to-dust ratio. For example, the SN shocks is on average 10 times lower than non-merger massive may continuously destroy the dust in the diffuse regions galaxies (Genzel et al. 2010). in the past yet meanwhile dust grains in dense clouds are protected from such destruction but instead grow fur- 4. DISCUSSIONS ther. Nevertheless, such a hypothesis needs systematic J investigations before any conclusion can be made. The detection of CO 1-0 emission in Sextans A, After subtracting the atomic gas mass from the derived if it is real, offers a crucial evidence for the existence total gas mass, the molecular gas mass is shown to be of molecular gas in such a low metallicity environment, 7 confirming the result inferred from the dust measure- about 10 M⊙ as listed in Table 1. The derived αCO value 3 −2 −1 −1 ment (Shi et al. 2014). Although warm H2 has been re- is about (2.8±1.1)×10 M⊙ pc (K km s ) , ∼ 700 −2 vealed by infrared spectroscopic observations in several times larger than the Milky Way value (4.3 M⊙ pc (K km s−1)−1) (Bolatto et al. 2013). In Fig. 4, we compared extremely metal poor galaxies (Hunt et al. 2010), CO our measurement to those in the literature (Israel 1997; emission, unlike warm H2 that traces shocked regions, Leroy et al. 2011; Sandstrom et al. 2013) whose molec- is known as tracers of ISM where stars form in massive ular gas masses are derived in the same way as ours metal rich galaxies. The detection of CO in Sextans A is except for the galaxy Wolf-Lundmark-Melotte (WLM, a first step toward determining whether CO can serve a Elmegreen et al. 2013). The gas-to-dust ratio used to similar role in extremely metal poor galaxies. derive the α of WLM is based on the relationship CO is one of few efficient coolants of molecular gas CO in galaxies. In extremely metal poor galaxies, due to between αCO and metallicity (R´emy-Ruyer et al. 2014), which shows a large scatter (> 10) around the metallic- the photo-dissociation, the CO can only survive in a ity of WLM. The metallicity measurement of Sextans A tiny core of molecular gas clouds, while H2 can be self- was not taken exactly at the CO pointing. But as no shielded and exist in thick envelopes surrounding the CO metallicity gradient is seen in Sextans A (Kniazev et al. molecular core. Thus cooling through CO in extremely 2005), we did not expect large offsets (> 0.1 dex) from metal poor galaxies may be not as effective as in metal the value quoted here. Our α is much larger than rich galaxies. If the CO molecular cores in metal poor CO galaxies follow the same relationship between the viral those at higher metallicities. The overall trend of αCO with metallicity is very steep, with a power law index masses and CO luminosities of giant molecular clouds between 2.5 and 3. This implies that CO emission is in Milky Way and nearby galaxies (Solomon et al. 1987; very weak in extremely metal poor galaxies. While our Bolatto et al. 2013), the viral mass of the CO molecule works offer very interesting constraints on the α at core for the observed CO luminosity is estimated to be CO × 4 extreme low metallicity, the result is solely based on one about 3 10 M⊙. This is only a tiny fraction of dust- region of ∼ 20 arcsec compared to the entire galaxy of based total molecular gas mass, as small as 0.3%. Then five arcmin. Only about 10% of the total molecular gas the measured CO within the IRAM beam is associated of Sextans A (Shi et al. 2014) is within the beam. Large with only a small fraction of the total molecular gas within the same beam, and it cannot effectively cool the scatters of αCO may be expected with future detections of CO emission over different regions of the galaxy. bulk of the molecular gas in Sextans A. As a result of the The relative strength of the CO brightness can also photo-dissociation of CO, atomic or ionized carbon may µ be compared to other quantities of Sextans A measured become abundant. Studies did show that [CII] 158 m is within the IRAM beam as listed in Table 1. By in- much brighter relative to CO in dwarfs as compared to tegrating the IR SED of the IRAM region, the 8-1000 in spirals (Israel & Maloney 2011; Brauher et al. 2008; µm infrared luminosity of dust within the beam is mea- Madden et al. 2013). With spatially resolved studies of IC 10, Madden et al. (1997) showed that the H2 column 4 Shi et al. density associated with ionized carbon can be five times than the observation. It is still difficult to judge which the observed HI density in order to interpret the cooling model is better or worse just based on one data point, as implied by the [CII] luminosity, and the associated H2 which is also because many assumptions are invoked in mass may be 100 times the mass of H2 associated with theoretical models in order to produce the trend of αCO CO. However, unlike CO, [CII] with its high excitation as a function of metallicity. A large range of model’s pre- temperature (97 K) still could not cool the gas to the dictions reflects the limited knowledge about CO forma- low temperature at which gas may contract to the high tion and destruction, as well as properties of gas clouds density for stars to form. over different spatial scales at the extreme low metallic- Fig. 4 further shows the comparison between the mea- ity. If αCO is truly large as seen in this study, CO may sured extreme large αCO of Sextans A with models’ pre- be not an efficient coolant in metal poor galaxies, and its dictions. The empirical relationship (yellow solid line, application as a tracer of the molecular gas mass may be Israel 1997) based on galaxies above 20%Z⊙ predicts inappropriate in the early Universe. 2.7 that αCO ∝ Z , giving a value several times larger than our observed one at 7%Z⊙. Based on the principle 5. CONCLUSIONS that the CO abundance is primarily regulated by photo- We reported a marginal detection of CO J=1-0 emis- dissociation and the H2 is self-shielded, theoretical mod- sion by targeting the brightest far-IR peak in an ex- els of Glover & Mac Low (2011) and Wolfire et al. (2010) tremely metal poor galaxy, Sextans A. The signal is predict that αCO is a strong function of the visual extinc- about 10 km/s offset from the HI peak within the same tion for individual clouds. If assuming the visual extinc- beam. The αCO is further derived by comparing the es- tion is proportional to the metallicity, α then increases CO timated CO flux to the H2 mass as inferred from the rapidly with the decreasing metallicity in both models, dust map. In spite of a marginal signal, the observa- predicting values larger than the observed for Sextans tion is deep enough to play a strong limit on the αCO A as shown in Fig. 4. The model of Narayanan et al. that is about seven hundred times the Milky Way value. (2012) also employed photo-dissociation region models This suggests that CO emission is exceedingly weak in but for the integrated galaxies instead of individual extremely metal poor galaxies, challenging its role as a clouds, and found on average a flat trend of αCO as a gas coolant in these galaxies. Current theoretical mod- function of metallicity, producing a value much smaller els produce a large range of αCO at the metallicity of than the observed one at 7%Z⊙. Unlike the above mod- Sextans A. els, the model of Feldmann et al. (2012) employs small scale magneto-hydrodynamic simulations combined with large-scale simulations of gas distributions to predict We thank the anonymous referee for constructive com- αCO as a function of metallicities. For physical scales ments that improve the paper significantly. Y.S. ac- of our IRAM beam size (sub-kpc), a shallow trend as in- knowledges support for this work from Natural Sci- dicated by the purple line in Fig. 4 is predicted, with ence Foundation of China under the grant of 11373021 predicted values significantly smaller than the observa- and by the Strategic Priority Research Program “The tion. Emergence of Cosmological Structures” of the Chinese As a summary, current models predict a large range of Academy of Sciences, Grant No. XDB09000000. Y.G. αCO at the metallicity of Sextans A, some significantly is supported under grants of 11390373, 11420101002 & larger than the observed one while some much smaller XDB09000000.

REFERENCES Aniano, G., Draine, B. T., Gordon, K. D., & Sandstrom, K. 2011, Leroy, A., Cannon, J., Walter, F., Bolatto, A., & Weiss, A. 2007, PASP, 123, 1218 ApJ, 663, 990 Bolatto, A. D., Wolfire, M., & Leroy, A. K. 2013, ARA&A, 51, 207 Leroy, A. K., Bolatto, A., Gordon, K., et al. 2011, ApJ, 737, 12 Brauher, J. R., Dale, D. A., & Helou, G. 2008, ApJS, 178, 280 Madden, S. C., Poglitsch, A., Geis, N., Stacey, G. J., & Townes, Dale, D. A., Cohen, S. A., Johnson, L. C., et al. 2009, ApJ, 703, C. H. 1997, ApJ, 483, 200 517 Madden, S. C., R´emy-Ruyer, A., Galametz, M., et al. 2013, PASP, Draine, B. T., & Li, A. 2007, ApJ, 657, 810 125, 600 Elmegreen, B. G., Rubio, M., Hunter, D. A., et al. 2013, Nature, Narayanan, D., Krumholz, M. R., Ostriker, E. C., & Hernquist, L. 495, 487 2012, MNRAS, 421, 3127 Eskew, M., Zaritsky, D., & Meidt, S. 2012, AJ, 143, 139 Ott, J., Stilp, A. M., Warren, S. R., et al. 2012, AJ, 144, 123 Feldmann, R., Gnedin, N. Y., & Kravtsov, A. V. 2012, ApJ, 747, Pettini, M., & Pagel, B. E. J. 2004, MNRAS, 348, L59 124 R´emy-Ruyer, A., Madden, S. C., Galliano, F., et al. 2014, A&A, Gao, Y., & Solomon, P. M. 2004, ApJ, 606, 271 563, A31 Genzel, R., Tacconi, L. J., Gracia-Carpio, J., et al. 2010, MNRAS, Sandstrom, K. M., Leroy, A. K., Walter, F., et al. 2013, ApJ, 777, 407, 2091 5 Glover, S. C. O., & Mac Low, M.-M. 2011, MNRAS, 412, 337 Shi, Y., Armus, L., Helou, G., et al. 2014, Nature, 514, 335 Goldsmith, P. F. 2001, ApJ, 557, 736 Solomon, P. M., Rivolo, A. R., Barrett, J., & Yahil, A. 1987, ApJ, Hunt, L. K., Thuan, T. X., Izotov, Y. I., & Sauvage, M. 2010, ApJ, 319, 730 712, 164 Taylor, C. L., Kobulnicky, H. A., & Skillman, E. D. 1998, AJ, 116, Israel, F. P. 1997, A&A, 328, 471 2746 Israel, F. P., & Maloney, P. R. 2011, A&A, 531, AA19 Wolfire, M. G., Hollenbach, D., & McKee, C. F. 2010, ApJ, 716, Kennicutt, R. C., Jr. 1998, ApJ, 498, 541 1191 Kennicutt, R. C., Jr. 1998, ARA&A, 36, 189 Young, J. S., & Scoville, N. 1982, ApJ, 258, 467 Kniazev, A. Y., Grebel, E. K., Pustilnik, S. A., Pramskij, A. G., & Zucker, D. B. 2005, AJ, 130, 1558