Dense and warm neutral gas in BR1202-0725 at z = 4.7 as traced by the [O I] 145 um line
Minju M. Lee, Tohru Nagao, Carlos De Breuck, Stefano Carniani, Giovanni Cresci, Bunyo Hatsukade, Ryohei Kawabe, Kotaro Kohno, Roberto Maiolino, Filippo Mannucci, Alessandro Marconi, Kouichiro Nakanishi, Paulina Troncoso, Hideki Umehata
DD RAFT VERSION F EBRUARY
19, 2021Typeset using L A TEX twocolumn style in AASTeX63
Dense and warm neutral gas in BR1202-0725 at z = 4.7 as traced by the [O I] 145 µ m line M INJU
M. L EE , T OHRU N AGAO , C ARLOS D E B REUCK , S TEFANO C ARNIANI , G IOVANNI C RESCI , B UNYO H ATSUKADE , R YOHEI K AWABE ,
7, 8 K OTARO K OHNO ,
6, 9 R OBERTO M AIOLINO ,
10, 11 F ILLIPO M ANNUCCI , A LESSANDRO M ARCONI ,
5, 12 K OUICHIRO N AKANISHI ,
7, 8 P AULINA T RONCOSO , H IDEKI U MEHATA ,
14, 61
Max-Planck-Institut f¨ur Extraterrestrische Physik (MPE), Giessenbachstr. 1, D-85748 Garching, Germnay Graduate School of Science and Engineering, Ehime University, 2-5 Bunkyo-cho, Matsuyama 790-8577, Japan European Southern Observatory, Karl Schwarzschild Straße 2, 85748 Garching, Germany Scuola Normale Superiore, Piazza dei Cavalieri 7, I-56126 Pisa, Italy INAF – Osservatorio Astrofisico di Arcetri, Largo E. Fermi 5, I-20125 Firenze, Italy Institute of Astronomy, Graduate School of Science, The University of Tokyo, 2-21-1 Osawa, Mitaka, Tokyo 181-0015, Japan National Observatory of Japan, 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan SOKENDAI (The Graduate University for Advanced Studies), 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan Research Center for the Early Universe, The University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113-0033, Japan Cavendish Laboratory, University of Cambridge, 19 J. J. Thomson Avenue, Cambridge CB3 0HE, UK Kavli Institute for Cosmology, University of Cambridge, Madingley Road, Cambridge CB3 0HA, UK Dipartimento di Fisica e Astronomia, Universit´a degli Studi di Firenze, Via G. Sansone 1, I-50019 Sesto F.no, Firenze, Italy Escuela de Ingenier´ıa, Universidad Central de Chile, Avenida Francisco de Aguirre 0405, 171-0614, La Serena, Coquimbo, Chile RIKEN Cluster for Pioneering Research, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan (Received; Revised; Accepted)
Submitted toABSTRACTWe report the detection of [O I ] 145.5 µ m in the BR 1202-0725 system, a compact group at z = 4 . consistingof a quasar (QSO), a submillimeter-bright galaxy (SMG), and three faint Ly α emitters. By taking into accountthe previous detections and upper limits, the [O I ] /[C II ] line ratios of the now five known high- z galaxiesare higher than or on the high-end of the observed values in local galaxies ([O I ] /[C II ] (cid:38) . ). The high[O I ] /[C II ] ratios and the joint analysis with previous detection of [N II ] lines for both of the QSO and theSMG suggest the presence of warm and dense neutral gas in these highly star-forming galaxies. This is furthersupported by new CO (12–11) line detections and a comparison with cosmological simulations. There is apossible positive correlation between the [N II ] 122/205 line ratio and the [O I ] /[C II ] ratio when all local andhigh-z sources are taken into account, indicating that the denser the ionized gas, the denser and warmer theneutral gas (or vice versa). The detection of the [O I ] line in the BR1202-0725 system with a relatively shortamount of integration with ALMA demonstrates the great potential of this line as a dense gas tracer for high- z galaxies. Keywords: galaxies: evolution – galaxies: high-redshift – galaxies: ISM – galaxies: starburst – quasars: general– submillimeter: galaxies INTRODUCTIONOne of the key questions in modern astrophysics is to un-derstand the physical processes that govern the star formation
Corresponding author: Minju M. [email protected] and galaxy assembly in the early universe. Compared to typi-cal star-forming galaxies on the main-sequence (e.g., Noeskeet al. 2007; Elbaz et al. 2007; Speagle et al. 2014), QSOsand dusty star-forming galaxies – we will call the latter pop- a r X i v : . [ a s t r o - ph . GA ] F e b M INJU
M. L
EE ET AL .ulation as submillimeter-bright galaxies, hereafter SMGs ,in the generally accepted view – release enormous amountof energies coming from active black-hole accretion and/orstar formation. In this paper, we refer to main-sequence(MS) galaxies as those within ± . dex from the definitionof Speagle et al. (2014), and starburst galaxies as galaxiesat least 3 σ above the main-sequence (i.e., log ∆MS > . ).While QSOs and SMGs were discovered by different meth-ods, both populations often represent star-bursting galaxies.Of particular interest is understanding how their star forma-tion activities are regulated, their comparison with normalpopulations (i.e., main-sequence galaxies), and how they im-pact surroundings.Oxygen is the third most abundant element in the universe.The neutral oxygen has a ionization potential of 13.62 eV,which is close to that of hydrogen (13.59 eV), so the [O I ]emission line arises mostly from neutral regions. Fine struc-ture lines of oxygen serve as one of the main coolants of theinterstellar medium (ISM) at far-infrared (FIR) (e.g, Rosen-berg et al. 2015). With a critical density of ∼ cm − , [O I ]line traces much denser ISM than the [C II ] emission line.By constraining the physical properties (mainly the strengthof the radiation field and density), one can infer the dense gasdistributions where star-forming activity would take in place(e.g., Malhotra et al. 2001). This will essentially lead us tounderstand galaxy formation and evolution.The first detection of [O I ] ( P − P ) at 145.5 µ m (hereafter, [O I ] ) was reported in 1983 by Staceyet al. (1983), but the number of galaxies detected in thislower level transition was limited largely due to its fainternature relative to the higher transition of [O I ] ( P − P ) at63.2 µ m (hereafter, [O I ] ). The situation was greatly im-proved with the advent of the Herschel space telescope whichallowed studies of detailed ISM conditions of galaxies rang-ing from ultra-luminous infra-red galaxies (ULIRGs) to lowmetallicity dwarf galaxies together with other fine-structurelines (e.g., Farrah et al. 2013; Spinoglio et al. 2015; Cormieret al. 2015; Fern´andez-Ontiveros et al. 2016; Herrera-Camuset al. 2018). For high redshift galaxies ( z > . ), the linefalls into transmission windows that are observable fromground-based telescopes. However, less than a handful ofgalaxies have been observed and detected in [O I ] withhelp of galaxy lensing (Yang et al. 2019; De Breuck et al.2019) or for luminous QSO-host galaxies (Novak et al. 2019:non-detection ; Li et al. 2020).The BR1202-0725 system is a compact group at redshift z = 4 . consisting of a QSO, a SMG and (at least) three Ly-man alpha emitters (LAEs; LAE 1, LAE 2, and LAE 3) (Hu A recently used term is dusty star-forming galaxies (DSFGs) but we de-cided to use the conventional term SMG, here, because the name has beenused for BR1202-0725 for a long time. et al. 1996; Williams et al. 2014; Drake et al. 2020) . TheBR1202-QSO and BR1202-SMG are highly star-burstinggalaxies with L FIR of ∼ L (cid:12) (corresponding to SFR of ∼ M (cid:12) yr − ; e.g., Omont et al. 1996; Yun et al. 2000;Iono et al. 2006). In our previous paper (Lee et al. 2019), wereported the first detection of [N II ] 122 µ m and discussed theionized gas density for the first time at this redshift, based onthe [N II ] 122 µ m/[N II ] 205 µ m line ratio (hereafter, [N II ] / ). In this following paper, we report [O I ] line de-tections from both the QSO and the SMG. By adding thesetwo detections, the total number of the [O I ] detection at z > has now reached five.This work is organized as follows. In Section 2, we explainthe observations including ancillary data sets and data analy-ses. In Section 3, we describe the line detection and discussthe [O I ] /[C II ] line ratio in comparison with other galaxies.In Section 4, we discuss the ISM conditions inferred fromthe line ratio gathering other available information and sum-marize our findings. We adopt a standard Λ CDM cosmologywith H = 70 km s − Mpc − and Ω m = 0 . and Chabrierinitial mass function (IMF; Chabrier 2003). OBSERVATIONS AND DATA ANALYSIS2.1.
Band 7 : [O I ] 145.5 µ m observations The [O I ] 145 line observations were part of our ALMACycle 2 program (ID : L baseline between 15–558 m (C34-2/1, C34-3/(4)) on2014 December 13 and 2015 May 14 (total on-source time ofT integ = 39 minutes).The set-up for the correlator was four spectral windows(SPW), two of which were set to each sideband, each of theSPWs with a 1.875 GHz bandwidth. The spectral resolu-tion for the upper sideband was set to 3.906 MHz ( ∼ − ) to detect the [O I ] line and 7.812 MHz for the lowersideband ( ∼ − ). A strong quasar J1058+0133 andJ1256-0547 were chosen for bandpass calibration. J1256-0547 was also the phase calibrator for the Band 7 observa-tions. Ganymede and Titan were chosen for the flux calibra-tor in Band 7.We used Common Astronomy Software Applications( CASA ) (McMullin et al. 2007) version 4.2.2 for calibrationusing the pipeline script provided by the ALMA RegionalCenter staffs. We then used CASA 5.6.1 version for imag-ing and analyzing. Images were produced by
CASA task, Recent MUSE/VLT observations suggested that Ly α emission from thecompanion dubbed LAE 2 may be part of the QSO’s extended halo, thoughthe presence of a QSO companion, close to the LAE 2, is confirmed bythe detection of dust continuum, [C II ] and [N II ] (the [N II ] line ismarginal detection with S/N ∼
3) line emissions (Wagg et al. 2012; Carilliet al. 2013; Decarli et al. 2014).
ENSE , WARM NEUTRAL GAS IN
BR1202-0725
SYSTEM tclean , and deconvolved down to 1 σ noise level. The syn-thesized beam size with natural weighting is (cid:48)(cid:48) . × (cid:48)(cid:48) . .Tapered images were also created with uvtaper paramtersof (cid:48)(cid:48) . , (cid:48)(cid:48) . , (cid:48)(cid:48) . (the corresponding synthesized beams are (cid:48)(cid:48) . × (cid:48)(cid:48) . , (cid:48)(cid:48) . × (cid:48)(cid:48) . , and (cid:48)(cid:48) . × (cid:48)(cid:48) . , respec-tively) to check the existence of any extended emissions, es-pecially for QSO, that could resemble the Ly α halo (Drakeet al. 2020). We first CLEANed images from uv visibili-ties without continuum subtraction where the CLEAN maskswere made based on the position of each source (i.e., QSO,SMG, LAE 1 and LAE 2) with a spectral binning of 100 kms − . Continuum subtraction was applied using imcontsub after the full field-of-view (FoV) image was created. It wasintended to obtain improved results of continuum subtrac-tion for galaxies off from the phase center. The continuumwas subtracted with a linear fit by choosing line-free chan-nels. The 1 σ noise level after continuum subtraction at 100km s − is 0.40 mJy beam − for natural weighting. For ta-pered images, the noise levels are 0.46, 0.51 and 0.57 mJybeam − for uvtaper parameters of (cid:48)(cid:48) . , (cid:48)(cid:48) . and (cid:48)(cid:48) . , re-spectively, at 100 km s − . Primary beam correction was alsoapplied to get final images.Line intensity maps were created by choosing a chan-nel range which gives the highest peak signal-to-noise ratio(S/N). We measured line flux using the integrated map foreach source and investigated the curve of growth as a func-tion of aperture sizes using imfit . Typically, the peak S/Nswere the highest when the aperture size was set to a doublethe beam size. We measured the underlying Gaussian areain the 1D spectrum as well, which gives consistent valueswithin the errors compared to the fitted values in the 2D im-ages. The line widths are measured using the same aperturesize ( (cid:48)(cid:48) . ) and fitted the spectrum with a single Gaussiancomponent. We further investigated the reliability image-based continuum subtraction by performing continuum sub-traction in a 1D spectrum (for each source) adopting the sameaperture size, which also gave consistent values of fluxes(i.e., 1D Gaussian area) and line widths within the fitting er-rors. As for the final measurements for fluxes and line widths,we used the (cid:48)(cid:48) . -tapered image cube and took the apertureof (cid:48)(cid:48) . .2.2. Ancillary data sets of CO (12–11), HCN (6–5), andHCO + (6–5) line observations To demonstrate a supportive argument for the [O I ] detec-tions, we take two more data sets that our team were awardedusing ALMA as PI programs. One is from the same ALMACycle 2 program (ID: I ] line,as described above, and two [N II ] fine-structure lines at 122 µ m and 205 µ m. The Band 6 observations targetedthe [N II ] 205 µ m and CO (12–11) line at the same timeat the upper and lower side band each. See the details of theobservational summary presented in Lee et al. (2019). TheCO (12–11) line is detected (see Section 4.2) from the SMGand the QSO and we present the line profiles and maps inAppendix A.The other one is the Band 3 observations (ID: ≈ mins) out of 8.4 hrs requested. In thisprogram, we aimed at detecting two dense gas tracers ofHCN (6–5) and HCO + (6–5) to constrain the dense gas frac-tion and the lines are not detected for this partial execution(see Appendix B). The critical densities of the high- J HCNand HCO + transitions are two orders of magnitude higherthat of [O I ] . Still, the non-detection of the lines alterna-tively demonstrate the strength of the [O I ] line as a densegas ( (cid:38) cm − ) tracer that was detected with less than ahalf of the time executed for the Band 3 observations. RESULTS3.1.
Detection of [O I ] and line properties The bottom panels of Figure 1 show the line flux versusthe velocity of the QSO (left) and SMG (right). We detectthe [O I ] line in the QSO and SMG with a signal to noiseratio of 7 and 10, respectively. The [O I ] line from theSMG is detected at a higher significance level than the QSO.The [O I ] is not detected in LAE 1 and LAE 2 at the red-shifts and the positions reported in Carilli et al. (2013). Thepositions in Carilli et al. (2013) are from the submillimetercontinuum (SMG, QSO, and LAE 2) and [C II ] (LAE 1)emissions. The line search for every target was based onchecking the peak S/N in a fixed aperture by varying the in-tegrating range within the full velocity coverage of the uppersideband. The searching area is fixed to the beam size forall but LAE 1. As the optical position (of the redshifted Ly α emission) for LAE 1 is offset from the [C II ] position by ≈ (cid:48)(cid:48) . (Drake et al. 2020), we search for a line detectionaround a (cid:48)(cid:48) . -radius circular region from the [CII] peak tocheck if there is any emission associated to the Ly α emis-sion instead of the [C II ] . There is no significant detectionsignature for the LAE 1 above 3 σ . We summarize the lineproperties in Table 1 for the [O I ] line.The line widths (full-width half-maximum, FWHM) are ± and ± km s − for the SMG and the QSO,respectively. They are consistent with those of [C II ] ( ∼ − (SMG), ∼
300 km s − (QSO); Wagg et al. 2012; Car-illi et al. 2013; Carniani et al. 2013) within uncertainties, butthere is a hint of a broader [O I ] line width than the [C II ] for the SMG. The [N II ] fine-structure lines in our previ-ous work (Lee et al. 2019) also showed such a signature (i.e.,a broader line width) for the SMG with FWHM ∼ km M INJU
M. L
EE ET AL . Table 1.
Observational parameters of [O I ] and CO (12–11) in BR1202-0725[O I ] QSO SMG LAE1 LAE2 a LAE3 F line [Jy km s − ] . ± .
17 2 . ± . < . b < . b < . b FWHM [km s − ] ±
139 916 ± c c ... L line [ L (cid:12) ] . ± .
12 1 . ± . ... ... ...CO (12–11) QSO SMG LAE1 LAE2 a LAE3 F line [Jy km s − ] . ± .
05 2 . ± . < . d < . d < . d FWHM [km s − ] ±
88 1058 ± c c ... L line [ L (cid:12) ] . ± .
02 0 . ± . ... ... ... a Drake et al. (2020) concluded that “LAE2” is not the powering source of the Ly α emission seen the HST 775W map. Instead, the HSTemission is a stellar component passing through the QSO halo and is outshone by the halo. However, a QSO companion close to LAE 2 isconfirmed by dust, [C II ] , and (marginally) [N II ] emissions (Wagg et al. 2012; Carilli et al. 2013; Decarli et al. 2014), so we use the name“LAE2” to indicate this companion. b σ upper limit for an aperture of 1”.4 in the tapered image with uvtaper= (cid:48)(cid:48) . . The 3 σ limit is corresponding to a Gaussian area assumingthe FWHM of the lines and using the average channel noise. We adopt the FWHM values to be the same as the [C II ] from Carilli et al.(2013) for LAE 1 and LAE 2. For LAE 3, we assumed FWHM = 200 km s − , considering the reported FWHM in other literature forlow-mass galaxies (e.g., Pavesi et al. 2018; B´ethermin et al. 2020). All assumed FWHM values for LAEs are narrower than those of theLy α emissions reported in Drake et al. (2020). The noise is calculated based on the tapered cube (uvtaper = (cid:48)(cid:48) . ) with a spectral resolutionof 100 km s − . c From [C II ] observations in Carilli et al. (2013). d σ upper limit for an aperture of 2”.0. We assumed the same FWHM values that were assumed in the [O I ] line esimates. The noise iscalculated based on a cube with a spectral resolution of 50 km s − . s − . As for the QSO, the [O I ] line widths are consis-tent with those of CO, [C II ] and [N II ] , but [N II ] reported in the literature (Salom´e et al. 2012; Wagg et al.2012; Lee et al. 2019). Recent study of star-forming galaxiesat z ∼ have shown that different emission lines trace differ-ent components of galaxies (e.g., Carniani et al. 2017, 2020).However, the spatial resolution of current observations is notsufficient to spatially resolve the emission of the FIR linesin our galaxies. Future high-resolution observations will beable to reveal the origin of the discrepancy in line width andwhether the FIR lines are tracing distinct regions of the galax-ies. Having the resolution limit, we regard the lines as orig-inating from the same regions, at least globally. The derived[O I ] line luminosities are (1 . ± . × L (cid:12) and (0 . ± . × L (cid:12) for the SMG and the QSO, respec-tively. 3.2. [O I]/[C II] Line ratio The derived line luminosity ratios are L [OI]145 / L [CII]158 = . ± . and . ± . for the SMG and the QSO,respectively. The upper limits on the line ratio for LAE 1and LAE 2 are < . and < . , respectively, using thelinewidth constraints in the literature (see Table 1 footnote).In the right panel of Figure 2, we plot a stacked histogram ofthe [O I ] /[C II ] line ratios of local galaxies. It includesvarious types of galaxies from dwarf galaxies to ULIRGs andSeyfert. They are obtained from Brauher et al. (2008), Farrahet al. (2013), Rosenberg et al. (2015), Cormier et al. (2015),Spinoglio et al. (2015), Fern´andez-Ontiveros et al. (2016), and Herrera-Camus et al. (2018). When there are differentflux measurements on the same galaxies from the literaturewe plot the most recent ones. Out of 187 local galaxies con-sidered, the median (mean) value of L [OI]145 / L [CII]158 is 0.07(0.08) with a standard deviation of 0.06: ≈ of galaxieshave the [O I ] -to-[C II ] luminosity ratio below 0.13, which isthe lowest value observed in the z > galaxies. Among thelocal galaxies, ULIRGs (30 galaxies from Farrah et al. 2013;Rosenberg et al. 2015; Herrera-Camus et al. 2018) exhibita slightly higher median (mean) value of 0.10 (0.11) with astandard deviation 0.05, which is comparable (at 1 σ ) to high-z galaxies. We plot these median values on the left panel ofFigure 2. Considering these, we conclude that high- z SMGsand QSOs have higher [O I ] -to-[C II ] ratios compared totypical local galaxies at ∼ σ that are consistent with thehigh-end values observed in local ULIRGs.In the left panel of Figure 2, we show the [O I ] /[C II ]line ratio as a function of FIR luminosity for galaxies withFIR values reported. We used the 340 GHz flux (Iono et al.2006; Wagg et al. 2012; Carilli et al. 2013) to convert it into L FIR . For LAE 1 and LAE 2, we adopted the value reportedin Carilli et al. (2013) and the FIR luminosities are L FIR < . × L (cid:12) and . × L (cid:12) for LAE 1 and LAE 2,respectively.There is a hint of the [O I ] /[C II ] ratio increase as a func-tion of FIR luminosity ( L FIR ) for log L FIR /L (cid:12) (cid:38) in logscale. High- z sources detected in both lines align well withthis trend that for galaxies with higher L FIR tend to exhibit
ENSE , WARM NEUTRAL GAS IN
BR1202-0725
SYSTEM QSO SMGSMGQSO SMG
Figure 1.
Upper panel: [O I ] line maps for QSO (left) andSMG (right). The contours are drawn starting from 4 σ in steps of1 σ up to 12 σ , i.e., [4,5,6, ..., 12] σ . Negative component contoursat -4 σ is also drawn as dashed lines which do not exist nearby thesources. On the bottom left on each panel, we show the synthesizedbeam of the final image which is (cid:48)(cid:48) . × (cid:48)(cid:48) . (tapered imagewith uvtaper= (cid:48)(cid:48) . ). The panel size is (cid:48)(cid:48) in width. Bottom panel:the [O I ] line spectrum using the same image in the upper panelfor the QSO (left) and SMG (right) by taking an aperture of (cid:48)(cid:48) . .Shaded area is the velocity range to create the line intensity map.The spectral resolution is set to 100 km s − . The red curve in eachpanel shows the best-fit Gaussian fit. higher [O I ] /[C II ] values within the observed scatter. If weonly consider the local galaxy studies, which will be com-plete and not limited by sensitivity in most cases, the posi-tive correlation is observed for log L FIR (cid:38) with severalexceptional data points from the SHINING survey (Herrera-Camus et al. 2018). For a lower L FIR , e.g., in dwarf galaxies,the [O I ] /[C II ] ratio seems to saturate into a roughly con-stant value of ∼ I ] and uncertain L FIR in the LAEs do not put stricterconstraints on the correlation between the [O I ] -to-[C II ] lineratio and L FIR .The higher [O I ] -to-[C II ] line ratios observed in thehigh- z galaxies may be due to their L FIR compared to lo-cal samples. Meanwhile, the fact that high- z galaxies exhibita higher value than the average of ULIRG might suggest thatISM properties at fixed L FIR evolves as a function of red-shift. Future observations with larger number of galaxies willverify this and it will allow us to understand whether high- z starburst galaxies are different from local populations. DISCUSSION[O I ] line originates solely from neutral regions witha critical density of ∼ cm − , while the [C II ] line cancome both from ionized and neutral regions. Local galaxystudies have shown that the [O I ] /[C II ] ratio is a goodtracer of the photo-dissociation region (PDR, or neutral gas)density, once the [C II ] emission from ionized regions is sub-tracted, while the [O I ] /[O I ] is a tracer of the neutralgas temperature for a range between 100 and 400 K (e.g.,Malhotra et al. 2001; Fern´andez-Ontiveros et al. 2016) withthe caveats of optical depth effect and self-absorption in [O I ] . The line ratio between [O I ] and [C II ] ([O I ] /[C II ]) can be therefore, used as a tracer of the gas pressure in theneutral region. In the following, we first constrain the neu-tral gas fraction of the [C II ] emissions and then discuss thephysical meanings of the observed [O I ] -to-[C II ] line ra-tio based on the dust temperature constraint, the detection ofCO (12–11), and comparison with a cosmological model.4.1. Neutral fraction of the [CII] line
In order to infer a neutral gas fraction of the [C II] emissionfirst, we calculate the fraction following local galaxy studies(e.g., Croxall et al. 2017; D´ıaz-Santos et al. 2017; Herrera-Camus et al. 2018) using the [N II ] /[C II ] line ratio: f neutral[CII] = 1 − R ion (cid:18) [NII] [CII] (cid:19) (1)where R ion is the [C II ] /[N II ] luminosity ratio ifthe [C II ] line is originated only from ionized regions.Croxall et al. (2017) showed that R ion is almost constantranging between 2.5 and 3 for an electron density range of n e = [10 − cm − using the collision rates of Tayal(2008, 2011) and Galactic gas phase abundances of nitrogen(Meyer et al. 1997) and carbon (Sofia et al. 2004) (see alsoMalhotra et al. 2001; Oberst et al. 2006). The n e values con-strained in Lee et al. (2019) are n ion = +12 − and +50 − cm − for the SMG and QSO, respectively. Therefore, it isreasonable to assume a value between 2.5 and 3 as R ion . Weuse the [C II ] and [N II ] flux values reported in Wagg et al.(2012) and Lee et al. (2019), respectively. We note that Car-niani et al. (2013) reported lower values for the [C II ] fluxdue to a smaller beam size from their differently uv -weightedimages. Decarli et al. (2014) reported the [N II ] observa-tions of this system from IRAM Plateau de Bure Interfer-ometer (PdBI) observations. But the published flux values(or upper limits) are higher than what we obtained from ourALMA observations. This discrepancy can be attributed totheir low S/Ns and different ways of flux measurements. Asdiscussed in Lee et al. (2019), our [N II ] 205 flux is consistentwith the measurement by Pavesi et al. (2016), which is basedon our data set, and further checked with other independent(ALMA) data reported in Lu et al. (2017). The systematic M INJU
M. L
EE ET AL . Figure 2.
Left: Luminosity line ratios of [O I ] /[C II ] as a function of FIR luminosity. Four galaxies associated in the BR1202-0725system (SMG, QSO, LAE 1, and LAE 2) are plotted with the labels for the detection or the 3 σ upper limit constraints. For local studies, weplot Farrah et al. (2013), Cormier et al. (2015), Rosenberg et al. (2015), Spinoglio et al. (2015) and Herrera-Camus et al. (2018) for which L FIR values are available, ranging from local dwarfs to ULIRGs and Seyfert galaxies. We used the latest measurement when the same galaxies arelisted in different literature. As no detection data points are available for z = 1 − SMGs (and normal galaxies), we plot upper and lower limitsfor SMGs from Zhang et al. (2018). Higher- z ( z > ) sources (SMGs and QSOs with high L FIR ) are plotted as purple diamonds (De Breucket al. 2019; Yang et al. 2019; Li et al. 2020) and as an arrow for an upper limit from Novak et al. (2019). In addition to observational data points,three additional data points (thin diamonds) are also plotted from the hydrodynamical simulations reported in Lupi et al. (2020) using differentmodel assumptions (see text for details). We show the median values of local galaxies and ULIRGs as dotted and dashed lines, respectively.Right: Stacked histogram of luminosity line ratios of [O I ] /[C II ] for local galaxies. The bins for the histogram are set to have linearsteps in log space. For the histogram, we include local studies from Brauher et al. (2008), Farrah et al. (2013) ((U)LIRGs), Rosenberg et al.(2015) ((U)LIRGs), Cormier et al. (2015) (dwarf), Spinoglio et al. (2015) (Seyfert), Fern´andez-Ontiveros et al. (2016), and Herrera-Camuset al. (2018). The latest measurements are plotted when the same galaxies are listed in different literature. errors from the flux measurement method could change theflux value by a factor of 3, however. As it is difficult to pindown the origin of the difference in different observations,we stick to our best measurement listed in Lee et al. (2019),where the flux measurements are done similarly to the workpresented here.The inferred neutral fraction, f neutral[CII] , is 79-83% (SMG)and 83-86% (QSO) for the observed [N II ] /[C II ] lu-minosity ratios of . ± . (SMG) and . ± . (QSO). Forcomparison, local studies have found that the contribution ofthe [C II ] emission originated from H II regions to the total[C II ] is not dominant (less than 50%) across a wide range ofSFR density and metallicity (Croxall et al. 2017; D´ıaz-Santoset al. 2017; Herrera-Camus et al. 2018). Similarly, for theQSO and the SMG at z = 4 . , the [C II ] emission is mostlycoming from neutral regions rather than the ionized. Withthe f neutral[CII] constraints, we use the [O I ] /[C II ] ra- tio as a tracer for the neutral gas density and gas temperature(thus the gas pressure), without imposing any assumptions ofinterstellar medium (ISM) structure.4.2. Dust temperature and CO (12–11) detection
We infer T dust from an empirical fitting that connects the[O I ] /[C II ] neutral line ratio to T dust through the FIR color( S /S ) (D´ıaz-Santos et al. 2017, using Equation 6 and 2in the paper). It gives 43 K and 54 K for the SMG and theQSO, respectively and they are well-matched with the fittedSED dust temperature in Salom´e et al. (2012). For this cal-culation we have assumed [O I ] /[O I ] = 10, and theneutral fraction of the [C II ] emission based on the [N II ] /[C II ] line ratio as explained above. We also note thatthe high neutral gas fraction of the [C II ] emission is also cou-pled with the warm dust temperature. The [O I ] /[O I ] ratio below 10 would imply an optically thick emission (Tie-lens & Hollenbach 1985) and many local galaxies (except for ENSE , WARM NEUTRAL GAS IN
BR1202-0725
SYSTEM Figure 3.
CO SLEDs for the QSO and the SMG (from this workand Salom´e et al. 2012) relative to z ∼ QSOs (Carniani et al.2019) and local AGNs and starburst galaxies (Mashian et al. 2015),where CO (2–1) luminosities are available for normalization. TheCO SLEDs for the SMG is shifted rightward by 0.2 for clarity. a few extraordinary galaxies, e.g., Arp220, IRAS17208-0014with self-absorption of [O I ] ) have [O I ] /[O I ] ra-tios higher than that. For the optically thin case, the inferreddust temperature would be higher. The detection of [O I ] will be very challenging from ground-based telescopes forthe BR1202-0725 system; the line would fall into an ALMABand 10 coverage, where the transmission is below 0.2 at abest precipitable water vapor condition and the line [O I ] may also suffer from self-absorption for these dusty popu-lation. Nevertheless, the remarkable agreement between thedust temperature from different approaches strengthens theview that for both the SMG and the QSO, the observed high[O I ] /[C II ] ratios are closely connected to their warmdust temperatures and hence high kinetic gas temperatures.As another, supportive evidence, we report the first de-tection of CO (12–11) line in the SMG and the QSO (seeAppendix A for images and the spectra). In Figure 3, weshow the CO spectral energy distribution (CO SLED) for var-ious targets from local and high- z galaxy studies (Mashianet al. 2015; Carniani et al. 2019). For the CO SLEDs of theBR1202-0725 pair, we took values from Salom´e et al. (2012)for J upper = 2 , , , and ( J upper = 2 values were orig-inally from Carilli et al. (2002) but corrected for the VLAbandwidth). While we defer a modeling of the CO SLED,the CO SLEDs of the SMG and the QSO are similar to thoseof z ∼ QSOs, local Seyferts (NGC 1068, NGC 4945) or warm ULIRGs (NGC 4418). These similarities provide ad-ditional supporting evidence that the QSO and the SMG havesimilar gas properties of warm and dense gas.4.3.
Comparison with a cosmological simulation
Finally, we compare with a cosmological simulation (Lupiet al. 2020) which allows us to infer the physical propertiesfrom the line ratio. They performed a cosmological zoom-insimulation targeting M vir ∼ × M (cid:12) halo at z = 6 and investigated the far-infrared fine-structure lines. Whilethe target redshift and galaxies ( M (cid:63) ∼ M (cid:12) and SFR ≈ M (cid:12) yr − at z = 6 ) are different from the star-burstingpair (SFR ≈ M (cid:12) yr − ) at z = 4 . , it provides an in-sight regarding the [O I ] /[C II ] line ratio. They ran threedifferent models, two (CloudyVAR and CloudyFIX) usingCLOUDY (Ferland et al. 1998) and one (Krome) KROME(Grassi et al. 2014) to test how photoinoization equilibriumand thermal state assumptions affect the FIR emission prop-erties. The thermodynamics and chemistry are fully coupledin the Krome model, whereas in the two Cloudy models theyare not. CloudyFIX assume a constant temperature to takeinto account any dynamical effects while for CloudyVAR thetemperature is variable according to the radiation attenuation.Krome does not take into account the chemical network e.g.,CO, while CloudyVAR and CloudyFIX do.In the left panel of Figure 2, we plot the data points fromthree different models presented in Lupi et al. (2020). Weconverted the simulated SFR into L FIR using the Kennicutt(1998) recipe to overplot in Figure 2. This conversion maycontain large uncertainties, depending on the assumptionse.g., IMF, star-formation history, contribution of the old stars.As noted in Kennicutt (1998), the conversion can have uncer-tainties of ±
30% (starburst galaxies) and the inferred L FIR can result in up to a factor of two to three lower value fornormal spiral galaxies. While the converted L FIR is differentby two orders of magnitude compared to the QSO and theSMG, the observed [O I ] -to-[C II ] line ratios are consistentwith those predicted by the CloudyFIX and Krome models.Lupi et al. (2020) discussed the origin of the differencesbetween different models. As the [O I ] line is strongly de-pendent on the gas temperature, they claimed that the lower[O I ] -to-[C II ] line ratio in CloudyVAR can be explainedby the lower temperature predicted in the model comparedto the others. Both Krome and CloudyFIX simulations showboth the higher temperature and gas density distribution inthe (luminosity-weighted) density-temperature phase dia-gram than CloudyVAR. Meanwhile, the difference betweenKrome and CluodyFIX could be coming from the nature (ora caveat) of the Krome model where the chemical network(e.g., CO formation, highly ionized species) is not fully takeninto account. In the Cloudy models, the calculation is post-processed that all input values are already attenuated than M INJU
M. L
EE ET AL .the intrinsic flux that would affect the chemistry. Further, inboth Cloudy models, gas shocks are not taken into accountand a temperature may not be fully consistent with the hy-drodynamics in the simulation. We should note again thatthe simulated galaxy is different from our galaxies in thatthey have different galaxy properties (e.g., M star , SFR) andredshift. As the simulation does not give any informationon how dusty galaxy they are, another uncertainty comes infor the conversion between SFR and L FIR . Despite the dif-ferent caveats in the models and different galaxy propertiesbetween the observed and simulated galaxies, the compar-ison between the observations and simulations suggest theexistence of dense and warm neutral gas in the BR1202-0725system.Taking all together, the high [O I ] /[C II ] ratios in the SMGand the QSO reasonably indicate the existence of dense andwarm neutral gas. The SMG shares ISM properties with theQSO, where the black hole accretion is actively happening.While, at this moment, it is difficult to conclude whether ourobservational fact challenges the starburst-QSO evolutionaryscenario (e.g., Hopkins et al. 2008), it is tempting to say thatthe highly obscured starburst (the SMG) and the QSO havesimilar ISM properties and the SMG might have a hiddenAGN, even though both galaxies have not encountered a finalcoalescence. 4.4. Multi-phase properties
Figure 4 shows the [N II ] /[N II ] versus [O I ] /[C II ] ratios of the QSO and the SMG. We gatheredall data in the literature from local and high- z galaxy stud-ies when they are available (Cormier et al. 2015; Spinoglioet al. 2015; Fern´andez-Ontiveros et al. 2016; De Breuck et al.2019; Novak et al. 2019; Yang et al. 2019; Li et al. 2020).The figure compares the ionized gas density on the verticalaxis versus the neutral gas density and temperature on thehorizontal axis.Two dust-obscured star-forming galaxies (BR1202-SMGand SPT0418-47) have relatively low [N II ] /[N II ] ratios compared to its high [O I ] /[C II ] ratios, which aresimilar to the other high- z samples. As shortly discussed inLee et al. (2019), the optical depth can affect the line fluxesat high frequencies and, in particular, it may lead to a fainter[N II ] 122 flux, thus lowering the [N II ] ratio. The [O I ] and[C II ] lines are closer in frequency and both of them at lowerfrequency thus may be less affected than the [N II ] line ratio.We also discussed in the previous paper that [N II ] may nottrace very dense ionized gas because a combination of the[N II ] lines can trace gas density up to ≈
500 cm − . In thisregard, it may be worth noting on one exceptional data pointof NGC 4151, which exhibits a low [N II ] ratio at the given[O I ] -to-[C II ] ratio in the local samples (i.e., an x-cross just Figure 4.
Luminosity line ratios of [N II ] /[N II ] versus[O I ] /[C II ] . Together with our work, we plot local stud-ies from Fern´andez-Ontiveros et al. 2016 (crosses), Spinoglio et al.2015 (filled circles), and one SPT galaxy at z = 4 . from De Breucket al. (2019) having all the four lines detected. Two lower limits ofthe [N II ] 122/205 ratio from dwarf galaxies from Cormier et al.(2015) are also plotted. We place two lines indicating the [O I ]/[C II ] ratio from z ∼ − QSO observations without [N II ] / constraints (Yang et al. 2019; Li et al. 2020) and upper limitof the [O I ] /[C II ] ratio and lower limit of the [N II ] ratio fromNovak et al. (2019). below SPT0418-47). It is a Seyfert galaxy where its electrondensity is found to be high ( ≈ cm − ) if it is derivedbased on the [Ne V]. Considering this, the low values in the[N II ] ratios in these two dust-obscured star-forming galaxiesmay also indicate the existence of even denser gas that cannotbe traced by [N II ] .For galaxies detected in all four lines (a total number of 43including the lower limit constraints from dwarf galaxies),there is a possible positive correlation, but statistically notsignificant, between the ionized gas density (traced by the[N II ] ratio) and neutral gas density and temperature (tracedby [O I ] /[C II ] ; Spearman’s correlation coefficient = 0.24with p -value 0.13). However, if we exclude three ‘outliers’of BR1202 SMG, SPT0418-47 and NGC 4151 whose [N II ]ratios are low compared to their high [O I ] -to-[C II ] ratio, weget a more significant correlation signature (i.e., Spearman’scoefficient of 0.37 with a p -value of 0.02). Of the zerothorder, the positive correlation is naively expected if the [N II ] / and [O I ] /[C II ] ratios trace density of theH II region and PDR which are physically connected witheach other. By taking into account the correlation between ENSE , WARM NEUTRAL GAS IN
BR1202-0725
SYSTEM L FIR and [O I ] /[C II ] and the one between [N II ] and [O I ]/[C II ] , extreme SFRs for high- z galaxies can be attributedto the existence of dense gas in both phases, ionized andneutral. We defer more sophisticated models to explain theobserved line ratios and ISM structures and conditions to afuture article.To conclude, we reported the detection of [O I ] froma compact group of BR1202-0725 system at z = 4 . . Thisadds two more galaxies in addition to three in the currentlyavailable detection for galaxies at z > . We find high [O I ]/[C II ] ratios compared to local galaxies for all high- z galax-ies which exhibit high FIR luminosities. The high [O I ]/[C II ] ratios and the joint analysis with previous detectionof [N II ] lines for both of the QSO and the SMG suggestthe presence of warm and dense neutral gas in these highlystar-forming galaxies. The detection of the [O I ] line in bothsystems with a relatively short amount of integration withALMA demonstrates the great potential of this line as a densegas tracer for high-z galaxies. Yet, we are still probing highlystar-forming exemplars, the [O I ] line detections of ‘normal’galaxies are also foreseen in future observations. ACKNOWLEDGMENTSWe thank to the anonymous referee for constructivecomments that contributed to the improvement of thiswork. This paper makes use of the following ALMAdata: ADS/JAO.ALMA Facilities:
ALMA
Software: astropy (Astropy Collaboration et al. 2013),CASA (McMullin et al. 2007)APPENDIX A. DETECTION OF CO(12-11)We detect CO (12–11) line emissions both from the SMGand the QSO. Figure 5 shows the integrated line maps andthe spectrum. Three LAEs are not detected in CO (12–11). The line fluxes, FWHM and luminosities are summa-rized in Table 1. For the SMG, it seems that there is asub-component associated to it in the southern area which ispeaked at ∼ − km s − with a long tail redward (Fig-ure 6). We could not identify a similar signature in the[C II ] map, though if it is real, it may be relevant to the[C II ] bridge component connected with the QSO (Carilliet al. 2013). LAE 1 is known to have an offset between theLy α and the [CII] emissions, which is (cid:48)(cid:48) . . Accordingly,we have checked whether the CO (12–11) subcomponent isassociated with the Ly α emission instead of the [C II ] emis-sion. The CO (12–11) subcomponent is offset from the [C II ] emission by (cid:48)(cid:48) . , which is larger than the Ly α –[C II ] sep-aration. Therefore, if the emission is real, this might not bedirectly associated with the Ly α emission from LAE 1, butwith a halo or outflowing component from the SMG, if any.The extended Ly α emissions (Drake et al. 2020) reach out toa region where the subcomponent is detected. Future, deeperobservations would need to verify this. B. HCN (6–5) AND HCO + (6–5) OBSERVATIONSAt a resolution of (cid:48)(cid:48) . × (cid:48)(cid:48) . , there is no clear de-tection feature around the expected velocity range (Fig-ure 7). The spectrum is after the continuum subtraction using imcontsub (fitorder = 1 ). For a circular aperture of (cid:48)(cid:48) . ,the 3 σ upper limit placed by this ALMA observations are < . Jy km s − and < . Jy km s − , for the QSO andthe SMG, respectively, assuming the same line widths mea-sured from the CO (12-11) emission (Table 1) and using thenoise level measured at 100 km s − resolution.REFERENCES Changing to fitorder = 0 does not change the result.
Astropy Collaboration, Robitaille, T. P., Tollerud, E. J., et al. 2013,A&A, 558, A33
INJU
M. L
EE ET AL . B´ethermin, M., Fudamoto, Y., Ginolfi, M., et al. 2020, A&A, 643,A2Brauher, J. R., Dale, D. A., & Helou, G. 2008, ApJS, 178, 280Carilli, C. L., Riechers, D., Walter, F., et al. 2013, ApJ, 763, 120Carilli, C. L., Kohno, K., Kawabe, R., et al. 2002, AJ, 123, 1838Carniani, S., Marconi, A., Biggs, A., et al. 2013, A&A, 559, A29Carniani, S., Marconi, A., Maiolino, R., et al. 2017, A&A, 605,A105Carniani, S., Gallerani, S., Vallini, L., et al. 2019, MNRAS, 489,3939Carniani, S., Ferrara, A., Maiolino, R., et al. 2020, MNRAS, 499,5136Chabrier, G. 2003, PASP, 115, 763Cormier, D., Madden, S. C., Lebouteiller, V., et al. 2015, A&A,578, A53Croxall, K., Smith, J. D. T., Pellegrini, E., et al. 2017, ArXive-prints, arXiv:1707.04435De Breuck, C., Weiß, A., B´ethermin, M., et al. 2019, A&A, 631,A167Decarli, R., Walter, F., Carilli, C., et al. 2014, ApJL, 782, L17D´ıaz-Santos, T., Armus, L., Charmandaris, V., et al. 2017, ApJ,846, 32
QSO SMGQSO SMG
Figure 5.
Upper: CO (12–11) line maps for QSO (left) and SMG(right). The contours are drawn starting from 4 σ in steps of 1 σ upto 10 σ , i.e., [4,5,6, ..., 10] σ . Negative component contours at -4 σ isalso drawn as dashed lines. The synthesized beam ( (cid:48)(cid:48) . × (cid:48)(cid:48) . )is shown on the bottom left on each panel. The panel size is (cid:48)(cid:48) in width. Bottom : the CO (12–11) line spectrum using the sameimage in the upper panel for the QSO (left) and SMG (right) usinga circular aperture of (cid:48)(cid:48) . . Shaded area velocity range to create theline intensity map. The spectral resolution is set to 50 km s − . Drake, A. B., Walter, F., Novak, M., et al. 2020, arXiv e-prints,arXiv:2007.14221Elbaz, D., Daddi, E., Le Borgne, D., et al. 2007, A&A, 468, 33Farrah, D., Lebouteiller, V., Spoon, H. W. W., et al. 2013, ApJ,776, 38Ferland, G. J., Korista, K., Verner, D. A., et al. 1998, PASP, 110,761Fern´andez-Ontiveros, J. A., Spinoglio, L., Pereira-Santaella, M.,et al. 2016, ApJS, 226, 19Grassi, T., Bovino, S., Schleicher, D. R. G., et al. 2014, MNRAS,439, 2386Herrera-Camus, R., Sturm, E., Graci´a-Carpio, J., et al. 2018, ApJ,861, 94 Figure 6.
The spectra at the position of the southern componentclose to the SMG taking a circular aperture of (cid:48)(cid:48) . . The normalizedspectra of the SMG (using the same aperture size) is also shown asa reference. Figure 7.
Non-detection spectrum of HCN (6–5) and HCO + (6–5)lines for the QSO (left) and the SMG (right). ENSE , WARM NEUTRAL GAS IN
BR1202-0725
SYSTEM11