Submillimeter H2O and H2O+emission in lensed ultra- and ...and hyper-luminous infrared galaxies at z ... Modelling reveals that the far-infrared radiation ﬁelds have warm dust temperature

AampA 595 A80 (2016)DOI 1010510004-6361201628160ccopy ESO 2016

AstronomyampAstrophysics

Submillimeter H2O and H2O+ emission in lensed ultra-and hyper-luminous infrared galaxies at z sim2ndash4

C Yang (杨辰涛)1 2 3 4 5 A Omont4 5 A Beelen2 E Gonzaacutelez-Alfonso6 R Neri7 Y Gao (高煜)1 P van der Werf8A Weiszlig9 R Gavazzi4 5 N Falstad10 A J Baker11 R S Bussmann12 A Cooray13 P Cox14 H Dannerbauer15

S Dye16 M Gueacutelin7 R Ivison17 18 M Krips7 M Lehnert4 5 M J Michałowski17D A Riechers12 M Spaans19 and E Valiante20

(Affiliations can be found after the references)

Received 19 January 2016 Accepted 20 July 2016

ABSTRACT

We report rest-frame submillimeter H2O emission line observations of 11 ultra- or hyper-luminous infrared galaxies (ULIRGs or HyLIRGs) atz sim 2ndash4 selected among the brightest lensed galaxies discovered in the Herschel-Astrophysical Terahertz Large Area Survey (H-ATLAS) Usingthe IRAM NOrthern Extended Millimeter Array (NOEMA) we have detected 14 new H2O emission lines These include five 321ndash312 ortho-H2Olines (Eupk = 305 K) and nine J = 2 para-H2O lines either 202ndash111 (Eupk = 101 K) or 211ndash202 (Eupk = 137 K) The apparent luminosities of theH2O emission lines are microLH2O sim 6ndash21 times 108 L (3 lt micro lt 15 where micro is the lens magnification factor) with velocity-integrated line fluxes rangingfrom 4ndash15 Jy km sminus1 We have also observed CO emission lines using EMIR on the IRAM 30 m telescope in seven sources (most of those havenot yet had their CO emission lines observed) The velocity widths for CO and H2O lines are found to be similar generally within 1σ errors inthe same source With almost comparable integrated flux densities to those of the high-J CO line (ratios range from 04 to 11) H2O is found tobe among the strongest molecular emitters in high-redshift HyULIRGs We also confirm our previously found correlation between luminosity ofH2O (LH2O) and infrared (LIR) that LH2O sim LIR

11ndash12 with our new detections This correlation could be explained by a dominant role of far-infraredpumping in the H2O excitation Modelling reveals that the far-infrared radiation fields have warm dust temperature Twarm sim 45ndash75 K H2O columndensity per unit velocity interval NH2O∆V amp 03 times 1015 cmminus2 kmminus1 s and 100 microm continuum opacity τ100 gt 1 (optically thick) indicating thatH2O is likely to trace highly obscured warm dense gas However further observations of J ge 4 H2O lines are needed to better constrain thecontinuum optical depth and other physical conditions of the molecular gas and dust We have also detected H2O+ emission in three sources Atight correlation between LH2O and LH2O+ has been found in galaxies from low to high redshift The velocity-integrated flux density ratio betweenH2O+ and H2O suggests that cosmic rays generated by strong star formation are possibly driving the H2O+ formation

Key words galaxies high-redshift ndash galaxies ISM ndash infrared galaxies ndash submillimeter galaxies ndash radio lines ISM ndash ISM molecules

1 Introduction

After molecular hydrogen (H2) and carbon monoxide (CO)the water molecule (H2O) can be one of the most abundantmolecules in the interstellar medium (ISM) in galaxies It pro-vides some important diagnostic tools for various physicaland chemical processes in the ISM (eg van Dishoeck et al2013 and references therein) Prior to the Herschel Space Ob-servatory (Pilbratt et al 2010) in extragalactic sources non-maser H2O rotational transitions were only detected by the In-frared Space Observatory (ISO Kessler et al 1996) in the formof far-infrared absorption lines (Gonzaacutelez-Alfonso et al 20042008) Observations of local infrared bright galaxies by Her-schel have revealed a rich spectrum of submillimeter (submm)H2O emission lines (submm H2O refers to rest-frame submil-limeter H2O emission throughout this paper if not otherwisespecified) Many of these lines are emitted from high-excitationrotational levels with upper-level energies up to Eupk = 642 K

Herschel is an ESA space observatory with science instruments pro-vided by European-led Principal Investigator consortia and with impor-tant participation from NASA The reduced spectra as FITS files are only available at the CDS viaanonymous ftp to cdsarcu-strasbgfr (130791285) or viahttpcdsarcu-strasbgfrviz-binqcatJA+A595A80

(eg van der Werf et al 2010 Gonzaacutelez-Alfonso et al 20102012 2013 Rangwala et al 2011 Kamenetzky et al 2012Spinoglio et al 2012 Meijerink et al 2013 Pellegrini et al2013 Pereira-Santaella et al 2013) Excitation analysis of theselines has revealed that they are probably excited through absorp-tion of far-infrared photons from thermal dust emission in warmdense regions of the ISM (eg Gonzaacutelez-Alfonso et al 2010)Therefore unlike the canonical CO lines that trace collisionalexcitation of the molecular gas these H2O lines represent a pow-erful diagnostic of the far-infrared radiation field

Using the Herschel archive data Yang et al (2013 hereafterY13) have undertaken a first systematic study of submm H2Oemission in local infrared galaxies H2O was found to be thestrongest molecular emitter after CO within the submm bandin those infrared-bright galaxies even with higher flux densitythan that of CO in some local ULIRGs (velocity-integrated fluxdensity of H2O(321ndash312) is larger than that of CO(5ndash4) in fourgalaxies out of 45 in the Y13 sample) The luminosities of thesubmm H2O lines (LH2O) are near-linearly correlated with totalinfrared luminosity (LIR integrated over 8ndash1000 microm) over threeorders of magnitude The correlation is revealed to be a straight-forward result of far-infrared pumping H2O molecules are ex-cited to higher energy levels through absorbing far-infrared pho-tons then the upper level molecules cascade toward the lines we

Article published by EDP Sciences A80 page 1 of 23

AampA 595 A80 (2016)

observed in an almost constant fraction (Fig 1) Although thegalaxies dominated by active galactic nuclei (AGN) have some-what lower ratios of LH2OLIR there does not appear to be a linkbetween the presence of an AGN and the submm H2O emission(Y13) The H2O emission is likely to trace the far-infrared radi-ation field generated in star-forming nuclear regions in galaxiesexplaining its tight correlation with far-infrared luminosity

Besides detections of the H2O lines in local galaxies fromspace telescopes redshifted submm H2O lines in high-redshiftlensed Ultra- and Hyper-Luminous InfraRed Galaxies (ULIRGs1013 L gt LIR ge 1012 L HyLIRGs LIR ge 1013 L) can also bedetected by ground-based telescopes in atmospheric windowswith high transmission Strong gravitational lensing boosts theflux and allows one to detect the H2O emission lines easilySince our first detection of submm H2O in a lensed Herschelsource at z = 23 (Omont et al 2011) using the IRAM NOrthernExtended Millimeter Array (NOEMA) several individual detec-tions at high-redshift have also been reported (Lis et al 2011van der Werf et al 2011 Bradford et al 2011 Combes et al2012 Lupu et al 2012 Bothwell et al 2013 Omont et al 2013Vieira et al 2013 Weiszlig et al 2013 Rawle et al 2014) Thesenumerous and easy detections of H2O in high-redshift lensedULIRGs show that its lines are the strongest submm molecularlines after CO and may be an important tool for studying thesegalaxies

We have carried out a series of studies focussing on submmH2O emission in high-redshift lensed galaxies since our firstdetection Through the detection of J = 2 H2O lines inseven high-redshift lensed HyULIRGs reported by Omont et al(2013 hereafter O13) a slightly super-linear correlation be-tween LH2O and LIR (LH2O prop LIR

12) from local ULIRGs andhigh-redshift lensed HyULIRGs has been found This resultmay imply again that far-infrared pumping is important for H2Oexcitation in high-redshift extreme starbursts The average ra-tios of LH2O to LIR for the J = 2 H2O lines in the high-redshiftsources tend to be 18 plusmn 09 times higher than those seen locally(Y13) This shows that the same physics with infrared pump-ing should dominate H2O excitation in ULIRGs at low and highredshift with some specificity at high-redshift probably linkedto the higher luminosities

Modelling provides additional information about the H2Oexcitation For example through LVG modelling Riechers et al(2013) argue that the excitation of the submm H2O emission inthe z sim 63 submm galaxy is far-infrared pumping dominatedModelling of the local Herschel galaxies of Y13 has been car-ried out by Gonzaacutelez-Alfonso et al (2014 hereafter G14) Theyconfirm that far-infrared pumping is the dominant mechanismresponsible for the submm H2O emission (except for the ground-state emission transitions such as para-H2O transition 111ndash000)in the extragalactic sources Moreover collisional excitation ofthe low-lying (J le 2) H2O lines could also enhance the radiativepumping of the (J ge 3) high-lying lines The ratio between low-lying and high-lying H2O lines is sensitive to the dust tempera-ture (Td) and H2O column density (NH2O) From modelling theaverage of local star-forming- and mild-AGN-dominated galax-ies G14 show that the submm H2O emission comes from regionswith NH2O sim (05ndash2)times1017 cmminus2 and a 100 microm continuum opac-ity of τ100 sim 005ndash02 where H2O is mainly excited by warmdust with a temperature range of 45ndash75 K H2O lines thus pro-vide key information about the properties of the dense cores ofULIRGs that is their H2O content the infrared radiation fieldand the corresponding temperature of dust that is warmer thanthe core outer layers and dominates the far-infrared emission

Observations of the submm H2O emission together with ap-propriate modelling and analysis therefore allows us to studythe properties of the far-infrared radiation sources in great de-tail So far the excitation analysis combining both low- andhigh-lying H2O emission has only been done in a few case stud-ies Using H2O excitation modelling considering both collisionand far-infrared pumping Gonzaacutelez-Alfonso et al (2010) andvan der Werf et al (2011) estimate the sizes of the far-infraredradiation fields in Mrk 231 and APM 08279+5255 (APM 08279hereafter) which are not resolved by the observations directlyand suggest their AGN dominance based on their total enclosedenergies This again demonstrates that submm H2O emission isa powerful diagnostic tool which can even transcend the angularresolution of the telescopes

The detection of submm H2O emission in the Herschel-ATLAS1 (Eales et al 2010 H-ATLAS hereafter) sourcesthrough gravitational lensing allows us to characterise thefar-infrared radiation field generated by intense star-forming ac-tivity and possibly AGN and learn the physical conditions inthe warm dense gas phase in extreme starbursts in the early Uni-verse Unlike standard dense gas tracers such as HCN whichis weaker at high-redshift compared to that of local ULIRGs(Gao et al 2007) submm H2O lines are strong and even com-parable to high-J CO lines in some galaxies (Y13 O13) There-fore H2O is an efficient tracer of the warm dense gas phasethat makes up a major fraction of the total molecular gas massin high-redshift HyULIRGs (Casey et al 2014) The success-ful detections of submm H2O lines in both local (Y13) and thehigh-redshift universe (O13) show the great potential of a sys-tematic study of H2O emission in a large sample of infraredgalaxies over a wide range in redshift (from local up to z sim 4)and luminosity (LIR sim 1010ndash1013 L) However our previoushigh-redshift sample was limited to seven sources and to oneJ = 2 para-H2O line (Eupk = 100ndash127 K) per source (O13)In order to further constrain the conditions of H2O excitationto confirm the dominant role of far-infrared pumping and tolearn the physical conditions of the warm dense gas phase inhigh-redshift starbursts it is essential to extend the studies tohigher excitation lines We thus present and discuss here the re-sults of such new observations of a strong J = 3 ortho-H2O linewith Eupk = 304 K in six strongly lensed H-ATLAS galaxiesat zsim 28ndash36 where a second lower-excitation J = 2 para-H2Oline was also observed (Fig 1 for the transitions and the corre-sponding Eup)

We describe our sample observation and data reduction inSection 2 The observed properties of the high-redshift submmH2O emission are presented in Sect 3 Discussions of the lens-ing properties LH2O-LIR correlation H2O excitation compari-son between H2O and CO AGN contamination will be givenin Sect 4 Section 5 describes the detection of H2O+ lines Wesummarise our results in Sect 6 A flat ΛCDM cosmology withH0 = 71 km sminus1 Mpcminus1 ΩM = 027 ΩΛ = 073 (Spergel et al2003) is adopted throughout this paper

2 Sample and observation

Our sample consists of eleven extremely bright high-redshiftsources with F500 microm gt 200 mJy discovered by the H-ATLASsurvey (Eales et al 2010) Together with the seven similar

1 The Herschel-ATLAS is a project with Herschel which is an ESAspace observatory with science instruments provided by European-ledPrincipal Investigator consortia and with important participation fromNASA The H-ATLAS website is httpwwwh-atlasorg

A80 page 2 of 23

C Yang et al H2O excitation in lensed HyULIRGs at z sim 2ndash4

Energy (K)

ortho-H2O para-H

para-H2O+ ortho-H

110101

211202

Fig 1 Energy level diagrams of H2O and H2O+ shown in black and redrespectively Dark blue arrows are the submm H2O transitions we haveobserved in this work Pink dashed lines show the far-infrared pumpingpath of the H2O excitation in the model we use with the wavelength ofthe photon labeled The light blue dashed arrow is the transition frompara-H2O energy level 220 to 211 along the cascade path from 220 to111 Rotational energy levels of H2O and H2O+ as well as fine structurecomponent levels of H2O+ are also shown in the figure

sources reported in our previous H2O study (O13) theyinclude all the brightest high-redshift H-ATLAS sources(F500 microm gt 170 mJy) but two imaged at 880 microm with SMA byBussmann et al (2013 hereafter B13) In agreement with the se-lection according to the methods of Negrello et al (2010) thedetailed lensing modelling performed by B13 has shown that allof them are strongly lensed but one G09v1124 (Ivison et al2013 see below) The sample of our present study is thuswell representative of the brightest high-redshift submillime-ter sources with F500 microm gt 200 mJy (with apparent total in-frared luminosity sim5ndash15 times 1013 L and z sim 15ndash42) found byH-ATLAS in its equatorial (ldquoGAMArdquo) and north-galactic-pole(ldquoNGPrdquo) fields in sim300 deg2 with a density sim005 degminus2 Inour previous project (O13) we observed H2O in seven stronglylensed high-redshift H-ATLAS galaxies from the B13 sampleIn this work in order to observe the high-excitation ortho-H2O(321ndash312) line with rest frequency of 1162912 GHz withthe IRAMNOEMA we selected the brightest sources at 500 micromwith z amp 28 so that the redshifted lines could be observed in areasonably good atmospheric window at νobs 300 GHz Eightsources with such redshift were selected from the B13 H-ATLASsample

B13 provide lensing models magnification factors (micro) andinferred intrinsic properties of these galaxies and list their COredshifts which come from Harris et al (2012) Harris et al(in prep) Lupu et al (in prep) Krips et al (in prep) and Riech-ers et al (in prep)

In our final selection of the sample to be studied in theH2O(321ndash312) line we then removed two sources SDP 81 andG12v230 that were previously observed in H2O (O13 andalso ALMA Partnership Vlahakis et al 2015 for SDP 81) be-cause the J = 2 H2O emission is too weak andor the inter-ferometry could resolve out some flux considering the lensingimage The observed high-redshift sample thus consists of twoGAMA-field sources G09v197 and G12v243 and four sourcesin the H-ATLAS NGP field NCv1143 NAv1195 NAv1177and NBv178 (Tables 1 and 2) Among the six remaining sourcesat redshift between 28 and 36 only one NBv178 has beenobserved previously in a low-excitation line para-H2O(202ndash111)(O13) Therefore we have observed both para-H2O line 202ndash111

or 211ndash202 and ortho-H2O(321ndash312) in the other five sources inorder to compare their velocity-integrated flux densities

In addition we also observed five sources mostly at lowerredshifts in para-H2O lines 202ndash111 or 211ndash202 (Tables 1 and 2) tocomplete the sample of our H2O low-excitation study They arethree strongly lensed sources G09v140 NAv156 and SDP11a hyper-luminous cluster source G09v1124 (Ivison et al 2013)and a z sim 37 source NCv1268 for which we did not proposea J = 3 H2O observation considering its large linewidth whichcould bring difficulties in line detection

As our primary goal is to obtain a detection of the submmH2O lines we carried out the observations in the compact Dconfiguration of NOEMA The baselines extended from 24 to176 m resulting in a synthesised beam with modestlow resolu-tion of sim10primeprime times 09primeprime to sim56primeprime times 33primeprime as shown in Table 1 TheH2O observations were conducted from January 2012 to Decem-ber 2013 in good atmospheric conditions (seeing of 03 primeprimendash15 primeprime)stability and reasonable transparency (PWV le 1 mm) The to-tal on source time was sim15ndash8 h per source 2 mm 13 mm and08 mm bands covering 129ndash174 201ndash267 and 277ndash371 GHzrespectively were used All the central observation frequencieswere chosen based on previous redshifts given by B13 accordingto the previous CO detections (Table 2) In all cases but one thefrequencies of our detections of H2O lines are consistent withthese CO redshifts The only exception is G09v140 where ourH2O redshift disagrees with the redshift of z = 20894 plusmn 00009given by Lupu et al (in prep) which is quoted by B13 We findz = 20925plusmn 00001 in agreement with previous CO(3ndash2) obser-vations (Riechers et al in prep) We used the WideX correlatorwhich provided a contiguous frequency coverage of 36 GHz indual polarisation with a fixed channel spacing of 195 MHz

The phase and bandpass were calibrated by measur-ing standard calibrators that are regularly monitored at theIRAMNOEMA including 3C 279 3C 273 MWC349 and0923+392 The accuracy of the flux calibration is estimated torange from sim10 in the 2 mm band to sim20 in the 08 mmband Calibration imaging cleaning and spectra extraction wereperformed within the GILDAS2 packages CLIC and MAPPING

To compare the H2O emission with the typical moleculargas tracer CO we also observed the sources for CO lines usingthe EMIR receiver at the IRAM 30 m telescope The CO datawill be part of a systematic study of molecular gas excitation inH-ATLAS lensed HyULIRGs and a full description of the dataand the scientific results will be given in a following paper (Yanget al in prep) The global CO emission properties of the sourcesare listed in Table 3 where we list the CO fluxes and linewidthsA brief comparison of the emission between H2O and CO lineswill be given in Sect 43

3 Results

A detailed discussion of the observation results for each sourceis given in Appendix A including the strength of the H2O emis-sion the image extension of H2O lines and the continuum(Fig A1) the H2O spectra and linewidths (Fig 2) and their com-parison with CO (Table 3) We give a synthesis of these resultsin this section

2 See httpwwwiramfrIRAMFRGILDAS for more informationabout the GILDAS softwares

A80 page 3 of 23

AampA 595 A80 (2016)

Table 1 Observation log

IAU name Source RA Dec RApk Decpk H2O line νobs Beam ton(J2000) (J2000) (J2000) (J2000) (GHz) (primeprime) (h)

H-ATLAS J0830510+013224 G09v197 08305102 +01322488 08305117 +01322439 211ndash202 162286 56times 33 3508305117 +01322409 321ndash312 250952 26times 11 31

H-ATLAS J1135263minus014605 G12v243 11352636 minus01460556 11352627 minus01460644 202ndash111 239350 23times 10 6911352628 minus01460643 321ndash312 281754 22times 11 15

H-ATLAS J1256327+233625 NCv1143 12563270 +23362486 12563256 +23362792 211ndash202 164739 31times 29 1512563256 +23362769 321ndash312 254745 21times 10 15

H-ATLAS J1326301+334410 NAv1195 13263012 +33440990 13263014 +33440911 202ndash111 250045 20times 17 3813263014 +33440909 321ndash312 293334 10times 09 31

H-ATLAS J1330084+245900 NBv178 13300856 +24585830 13300856 +24585855 321ndash312 282878 17times 11 42

H-ATLAS J0849334+021443 G09v1124-W 08493336 +02144230 08493359 +02144468 211ndash202 220537 18times 12 84G09v1124-T 08493295 +02143970H-ATLAS J0853589+015537 G09v140 08535890 +01553700 08535884 +01553775 211ndash202 243425 18times 10 19H-ATLAS J0910431minus000321 SDP11 09104309 minus00032251 09104306 minus00032210 202ndash111 354860 19times 15 38H-ATLAS J1251354+261457 NCv1268 12513546 +26145752 12513538 +26145812 211ndash202 160864 29times 26 77H-ATLAS J1344294+303036 NAv156 13442952 +30303405 13442946 +30303401 211ndash202 227828 17times 17 23

Notes RA and Dec are the J2000 Herschel coordinates which were taken as the centres of the NOEMA images displayed in Fig A1 RApk andDecpk are the J2000 coordinates of the NOEMA dust continuum image peaks νobs is the central observed frequency The rest-frame frequenciesof para-H2O 202ndash111 211ndash202 and ortho-H2O 321ndash312 lines are 987927 GHz 752033 GHz and 1162912 GHz respectively (the rest-frame fre-quencies are taken from the JPL catalogue httpspecjplnasagov) ton is the on-source integration time The source G09v1124 whichis not resolved by SPIRE is a cluster that consists of two main components eastern component W (G09v1124-W) and western component T(G09v1124-T) as described in Ivison et al (2013) (see also Fig A3)

Table 2 Previously observed properties of the sample

Source z F250 F350 F500 F880 rhalf ΣSFR f14 GHz Td micro microLIR(mJy) (mJy) (mJy) (mJy) (kpc) (103 M yrminus1 kpcminus2) (mJy) (K) (1013 L)

G09v197 3634 260 plusmn 7 321 plusmn 8 269 plusmn 9 855 plusmn 40 085 091 plusmn 015 plusmn015 44 plusmn 1 69 plusmn 06 153 plusmn 43G12v243 3127 290 plusmn 7 295 plusmn 8 216 plusmn 9 486 plusmn 23 ndash ndash plusmn015 ndash ndash (83 plusmn 17)NCv1143 3565 214 plusmn 7 291 plusmn 8 261 plusmn 9 972 plusmn 65 040 208 plusmn 077 061 plusmn 016 40 plusmn 1 113 plusmn 17 128 plusmn 43NAv1195 2951 179 plusmn 7 279 plusmn 8 265 plusmn 9 652 plusmn 23 157 021 plusmn 004 plusmn014 36 plusmn 1 41 plusmn 03 74 plusmn 20NAv1177 2778 264 plusmn 9 310 plusmn 10 261 plusmn 10 501 plusmn 21 ndash ndash plusmn015 ndash ndash (55 plusmn 11)NBv178 3111 273 plusmn 7 282 plusmn 8 214 plusmn 9 592 plusmn 43 055 109 plusmn 141 067 plusmn 020 43 plusmn 1 130 plusmn 15 107 plusmn 39

G09v1124-Wa2410 242 plusmn 7 293 plusmn 8 231 plusmn 9 500 plusmn 35 ndash ndash plusmn015 40 plusmn 1 1 33 plusmn 03

G09v1124-Ta ndash ndash plusmn015 36 plusmn 1 15 plusmn 02 27 plusmn 08G09v140 2089b 389 plusmn 7 381 plusmn 8 241 plusmn 9 614 plusmn 29 041 077 plusmn 030 075 plusmn 015 36 plusmn 1 153 plusmn 35 65 plusmn 25

SDP11 1786 417 plusmn 6 378 plusmn 7 232 plusmn 8 306 plusmn 24 089 022 plusmn 008 066 plusmn 014 41 plusmn 1 109 plusmn 13 62 plusmn 19NCv1268 3675 145 plusmn 7 201 plusmn 8 212 plusmn 9 789 plusmn 44 093 031 plusmn 014 110 plusmn 014 39 plusmn 1 110 plusmn 10 95 plusmn 27NAv156 2301 481 plusmn 9 484 plusmn 13 344 plusmn 11 731 plusmn 24 150 014 plusmn 008 112 plusmn 027 38 plusmn 1 117 plusmn 09 113 plusmn 31

Notes z is the redshift inferred from previous CO detection quoted by B13 (see the references therein) F250 F350 and F500 are the SPIRE fluxdensities at 250 350 and 500 microm respectively (Pascale et al 2011) F880 is the SMA flux density at 880 microm rhalf and ΣSFR are the intrinsic half-lightradius at 880 microm and the lensing-corrected surface SFR (star formation rate) density (Sect 42) f14 GHz is the 14 GHz band flux densities from theVLA FIRST survey Td is the cold-dust temperature taken from B13 (note that the errors quoted for Td are significantly underestimated since theuncertainties from differential lensing and single-temperature dust SED assumption are not fully considered) micro is the lensing magnification factorfrom B13 except for G09v1124 which is adopted from Ivison et al (2013) microLIR is the apparent total infrared luminosity mostly inferred fromB13 The microLIR in brackets are not listed in B13 thus we infer them from single modified black body dust SED fitting using the submm photometrydata listed in this table (a) The cluster source G09v1124 includes two main components G09v1124-W to the east and G09v1124-T to the west(Fig A3) and the values of these two rows are quoted from Ivison et al (2013) (b) our H2O observation gives z = 2093 for G09v140 This valueis slightly different from the value of 2089 quoted by B13 from Lupu et al (in prep) obtained by CSOZ-Spec but consistent with CO(3ndash2)observation by Riechers et al (in prep)

31 General properties of the H2O emissions

To measure the linewidth velocity-integrated flux density andthe continuum level of the spectra from the source peak and fromthe entire source we extract each spectrum from the CLEANedimage at the position of the source peak in a single synthesisbeam and the spectrum integrated over the entire source Thenwe fit them with Gaussian profiles using MPFIT (Markwardt2009)

We detect the high-excitation ortho-H2O(321ndash312) in fiveout of six observed sources with high signal to noise ratios

(SN gt 9) and velocity-integrated flux densities comparable tothose of the low-excitation J = 2 para-H2O lines (Table 4 andFigs 2 and A1) We also detect nine out of eleven J = 2 para-H2O lines either 202ndash111 or 211ndash202 with SN ge 6 in terms oftheir velocity-integrated flux density plus one tentative detectionof H2O(202ndash111) in SDP11 We present the values of velocity-integrated H2O flux density detected at the source peak in a sin-gle synthesised beam IH2O

pk and the velocity-integrated H2Oflux density over the entire source IH2O (Table 4) The detectedH2O lines are strong with IH2O = 37ndash146 Jy km sminus1 Even con-sidering gravitational lensing correction this is consistent with

A80 page 4 of 23

Fig 2a Spatially integrated spectra of H2O in the six sources with both J = 2 para-H2O and J = 3 ortho-H2O lines observed The red linesrepresent the Gaussian fitting to the emission lines The H2O(202ndash111) spectrum of NBv178 is taken from O13 Except for H2O(321ndash312) inNAv1195 all the J = 2 and J = 3 H2O lines are well detected with a high SN and similar profiles in both lines for the same source

our previous finding that high-redshift HyULIRGs are verystrong H2O emitters with H2O flux density approaching that ofCO (Tables 3 and 4 and Sect 43) The majority of the images(711 for J = 2 lines and 34 for J = 3) are marginally re-solved with IH2O

pkIH2O sim 04ndash07 They show somewhat lensedstructures The others are unresolved with IH2O

pkIH2O gt 08All continuum emission flux densities (S ν(ct)pk for the emissionpeak and S ν(ct) for the entire source) are very well detected(SN ge 30) with a range of total flux density of 9ndash64 mJy forS ν(ct) Figure A1 shows the low-resolution images of H2O and

the corresponding dust continuum emission at the observing fre-quencies Because the positions of the sources were derived fromHerschel observation which has a large beamsize (gt17primeprime) com-paring to the source size the position of most of the sourcesare not perfectly centred at these Herschel positions as seen inthe maps The offsets are all within the position error of the Her-schel measurement (Fig A1) G09v1124 is a complex HyLIRGsystem including two main components eastern G09v1124-Wand western G09v1124-T as described in Ivison et al (2013)In Fig A3 we identified the two strong components separated

A80 page 5 of 23

AampA 595 A80 (2016)

Fig 2b Spatially integrated spectra of H2O of the five sources with only one J = 2 para-H2O line observed The red lines represent the Gaussianfitting to the emission lines Except for the H2O line in G09v1124 all the J = 2 H2O lines are well detected

Table 3 Observed CO line properties using the IRAM 30 mEMIR

Source CO line ICO ∆VCO(Jy km sminus1) (km sminus1)

G09v197 5ndash4 95 plusmn 12 224 plusmn 326ndash5 104 plusmn 23 292 plusmn 86

NCv1143 5ndash4 131 plusmn 10 273 plusmn 276ndash5 110 plusmn 10 284 plusmn 27

NAv1195 5ndash4 110 plusmn 06 281 plusmn 16NAv1177 3ndash2 68 plusmn 04 231 plusmn 15

5ndash4 110 plusmn 06 230 plusmn 16NBv178 5ndash4 103 plusmn 08 614 plusmn 53

6ndash5 97 plusmn 10 734 plusmn 85G09v140 4ndash3 75 plusmn 21 198 plusmn 51NAv156 5ndash4 177 plusmn 66 432 plusmn 182

Notes ICO is the velocity-integrated flux density of CO ∆VCO is thelinewidth (FWHM) derived from fitting a single Gaussian to the lineprofile

about 10primeprime in agreement with Ivison et al (2013) The J = 2H2O and dust continuum emissions in NBv178 NCv1195G09v140 SDP 11 and NAv156 as well as the J = 3 ortho-H2Oand the corresponding dust continuum emissions in G09v197NCv1143 and NAv1177 are marginally resolved as shown inFig A1 Their images are consistent with the correspondingSMA images (B13) in terms of their spatial distribution Therest of the sources are not resolved by the low-resolution syn-thesised beams The morphological structure of the H2O emis-sion is similar to the continuum for most sources as shown inFig A1 The ratio S ν(ct)pkS ν(ct) and S ν(H2O)pkS ν(H2O) arein good agreement within the error However for NCv1143 inwhich S ν(ct)pkS ν(ct) = 055 plusmn 001 and S ν(H2O)pkS ν(H2O) =074 plusmn 016 the J = 3 ortho-H2O emission appears more com-pact than the dust continuum Generally it seems unlikely that

we have a significant fraction of missing flux for our sourcesNevertheless the low angular resolution (sim1primeprime at best) limits thestudy of spatial distribution of the gas and dust in our sourcesA detailed analysis of the images for each source is given inAppendix A

The majority of the sources have H2O (and CO) linewidthsbetween 210 and 330 km sminus1 while the four others rangebetween 500 and 700 km sminus1 (Table 4) Except NCv1268 whichshows a double-peaked line profile all H2O lines are well fit bya single Gaussian profile (Fig 2) The line profiles between theJ = 2 and J = 3 H2O lines do not seem to be significantly differ-ent as shown from the linewidth ratios ranging from 126plusmn 014to 084 plusmn 016 The magnification from strong lensing is verysensitive to the spatial configuration in other words differentiallensing which could lead to different line profiles if the differ-ent velocity components of the line are emitted at different spa-tial positions Since there is no visible differential effect betweentheir profiles it is possible that the J = 2 and J = 3 H2O linesare from similar spatial regions

In addition to H2O within the 36 GHz WideX band we havealso tentatively detected H2O+ emission in 3 sources NCv1143G09v197 and G15v2779 (see Sect 5)

32 Lensing properties

All our sources are strongly gravitationally lensed (exceptG09v1124 see Appendix A11) which increases the line fluxdensities and allows us to study the H2O emission in an afford-able amount of observation time However the complexity ofthe lensed images complicates the analysis As mentioned abovemost of our lensed images are either unresolved or marginally re-solved Thus we will not discuss here the spatial distribution ofthe H2O and dust emissions through gravitational lensing mod-elling However we should keep in mind that the correction of

A80 page 6 of 23

Table 4 Observed properties of H2O emission lines

Source H2O line νH2O S ν(ct)pk S ν(ct) S pkH2O S H2O IH2O

pk IH2O ∆VH2O microLH2O

(GHz) ( mJybeam ) (mJy) ( mJy

beam ) (mJy) ( Jy km sminus1

beam ) (Jy km sminus1) (km sminus1) (108 L)

G09v197 211ndash202 162255 89 plusmn 02 94 plusmn 02 149 plusmn 22 150 plusmn 21 38 plusmn 04 41 plusmn 04 257 plusmn 27 74 plusmn 07321ndash312 250947 217 plusmn 03 361 plusmn 03 78 plusmn 19 150 plusmn 26 24 plusmn 04 37 plusmn 04 234 plusmn 34 104 plusmn 10

NCv1143 211ndash202 164741 112 plusmn 01 133 plusmn 02 174 plusmn 13 187 plusmn 13 56 plusmn 03 58 plusmn 03 293 plusmn 15 101 plusmn 05321ndash312 254739 348 plusmn 05 635 plusmn 05 239 plusmn 43 321 plusmn 41 52 plusmn 06 80 plusmn 07 233 plusmn 22 213 plusmn 18

NAv1195 202ndash111 250034 140 plusmn 04 258 plusmn 04 66 plusmn 25 116 plusmn 25 21 plusmn 06 40 plusmn 06 328 plusmn 51 67 plusmn 10321ndash312 (293334) 172 plusmn 05 412 plusmn 05 lt42 lt73 lt15 lt26 330a lt50

NAv1177 202ndash111 261489 265 plusmn 06 355 plusmn 06 168 plusmn 49 212 plusmn 49 44 plusmn 09 54 plusmn 09 241 plusmn 41 82 plusmn 12321ndash312 307856 382 plusmn 04 620 plusmn 04 148 plusmn 26 252 plusmn 31 46 plusmn 05 73 plusmn 06 272 plusmn 24 129 plusmn 11

NBv178 202ndash111b 240290 154 plusmn 03 369 plusmn 04 50 plusmn 10 123 plusmn 32 27 plusmn 03 67 plusmn 13 510 plusmn 90 122 plusmn 24

321ndash312 282863 292 plusmn 02 426 plusmn 02 88 plusmn 10 106 plusmn 10 48 plusmn 04 67 plusmn 05 607 plusmn 43 143 plusmn 10G09v1124-W 211ndash202 (220537) 642 plusmn 015 76 plusmn 02 lt14 lt16 lt12c lt14c 850c lt13c

G09v1124-T 408 plusmn 015 49 plusmn 02 lt17 lt20 lt10c lt12c 550c lt10c

G09v140 211ndash202 243182 169 plusmn 02 306 plusmn 03 175 plusmn 20 277 plusmn 19 49 plusmn 04 82 plusmn 04 277 plusmn 14 57 plusmn 03SDP11 202ndash111 354930 292 plusmn 13 521 plusmn 13 148 plusmn 84 403 plusmn 117 52 plusmn 20 92 plusmn 20 214 plusmn 41 63 plusmn 11

NCv1268 211ndash202 161013 66 plusmn 01 100 plusmn 01 52 plusmn 11 90 plusmn 12 37 plusmn 04 70 plusmn 07 731 plusmn 75 128 plusmn 12NAv156 211ndash202 227822 140 plusmn 06 227 plusmn 06 158 plusmn 33 232 plusmn 30 78 plusmn 11 146 plusmn 13 593 plusmn 56 120 plusmn 11

Notes νH2O is the observed central frequency of H2O lines and the values in brackets are the H2O line frequencies inferred from the CO redshiftsfor the undetected sources S ν(ct)pk and S ν(ct) are the peak and spatially integrated continuum flux density respectively S pk

H2O is the peak H2Oline flux and S H2O is the total line flux IH2O

pk and IH2O are the peak and spatially integrated velocity-integrated flux density of the H2O lines∆VH2O is the H2O linewidth microLH2O is the apparent luminosity of the observed H2O line (a) The linewidth of the undetected H2O(321ndash312) inNAv1195 has been set to 330 km sminus1 by assuming that the widths of the H2O(321ndash312) and H2O(202ndash111) lines are roughly the same (b) the data ofpara-H2O(202ndash111) in NBv178 is taken from O13 (c) the 2σ upper limits of IH2O are derived by assuming that the H2O linewidths are similar tothose of the CO lines (Ivison et al 2013)

the magnification is a crucial part of our study In addition dif-ferential lensing could have a significant influence when compar-ing H2O emission with dust and even comparing different tran-sitions of same molecular species (Serjeant 2012) especially forthe emission from close to the caustics

In order to infer the intrinsic properties of our sample es-pecially LH2O as in our first paper O13 we adopted the lensingmagnification factors micro (Table 2) computed from the modellingof the 880 microm SMA images (B13) As shown in the Appendixthe ratio of S ν(ct)pkS ν(ct) and S ν(H2O)pkS ν(H2O) are in goodagreement within the uncertainties Therefore it is unlikely thatthe magnification of the 880 microm continuum image and H2O canbe significantly different However B13 were unable to providea lensing model for two of our sources G12v243 and NAv1177because their lens deflector is unidentified This does not affectthe modelling of H2O excitation and the comparison of H2O andinfrared luminosities since the differential lensing effect seemsto be insignificant as discussed in Sects 4 and Appendix A

4 Discussion

41 LH2 O ndash LIR correlation and LH2 O ndash LIR ratio

Using the formula given by Solomon et al (1992) we derive theapparent H2O luminosities of the sources microLH2O (Table 4) fromIH2O For the ortho-H2O(321ndash312) lines microLH2O varies in the rangeof 6ndash22 times 108 L while the microLH2O of the J = 2 lines are a factorsim12ndash2 weaker (Table 4) as discussed in Sect 42

Using the lensing magnification correction (taking the val-ues of micro from B13) we have derived the intrinsic H2O lumi-nosities (Table 5) The error of each luminosity consists of theuncertainty from both observation and the gravitational lensingmodelling After correcting for lensing the H2O luminosities of

our high-redshift galaxies appear to be one order of magnitudehigher than those of local ULIRGs as well as their infrared lumi-nosities (Table 5) so that many of them should rather be consid-ered as HyLIRGs than ULIRGs Though the ratio of LH2OLIR inour high-redshift sample is close to that of local ULIRGs (Y13)with somewhat a statistical increase in the extreme high LIR end(Fig 3)

As displayed in Fig 3 for H2O of the three observed linesbecause we have extended the number of detections to 21 H2Olines distributed in 16 sources and 3 transitions we may inde-pendently study the correlation of LH2O(202ndash111) and LH2O(211ndash202)with LIR while we had approximately combined the two lines inO13

As found in O13 the correlation is slightly steeperthan linear (LH2O sim L12

IR ) To broaden the dynamical rangeof this comparison we also included the local ULIRGsfrom Y13 together with a few other H2O detectionsin high-redshift HyULIRGs for example HLSJ 0918(HLSJ 0918286+514223) (Combes et al 2012 Rawle et al2014) APM 08279 (van der Werf et al 2011) SPT 0538(SPT-S J05381650308) (Bothwell et al 2013) and HFLS3(Riechers et al 2013 with the magnification factor fromCooray et al 2014) (Fig 3) In the fitting however we excludedthe sources with heavy AGN contamination (Mrk 231 andAPM 08279) or missing flux resolved out by the interferometry(SDP 81) We also excluded the H2O(321ndash312) line of HFLS3considering its unusual high LH2O(321ndash312)LIR ratio as discussedabove that could bias our fitting We have performed a linearregression in log-log space using the Metropolis-HastingsMarkov Chain Monte Carlo (MCMC) algorithm samplerthrough linmix_err (Kelly 2007) to derived the α in

LH2O prop LαIR (1)

A80 page 7 of 23

AampA 595 A80 (2016)

Table 5 IR luminosity H2O line luminosity and global dust temperature of the entire sample

Source H2O Transition LIR LH2O(211ndash202) LH2O(202ndash111) LH2O(321ndash312)

(1012 L) (107 L) (107 L) (107 L)G09v197 211ndash202 321ndash312 221 plusmn 59 107 plusmn 14 ndash 150 plusmn 19G12v243 202ndash111 321ndash312 832 plusmn 166micro ndash 884 plusmn 107micro 1432 plusmn 115microNCv1143 211ndash202 321ndash312 114 plusmn 31 90 plusmn 14 ndash 189 plusmn 33NAv1195 202ndash111 321ndash312 180 plusmn 46 ndash 164 plusmn 30 lt123NAv1177 202ndash111 321ndash312 550 plusmn 110micro ndash 820 plusmn 128micro 1291 plusmn 108microNBv178 202ndash111 321ndash312 82 plusmn 22 ndash 94 plusmn 21 110 plusmn 15

G09v1124-W 211ndash202 331 plusmn 32 lt129 ndash ndashG09v1124-T 211ndash202 145 plusmn 18 lt69 ndash ndash

G09v140 211ndash202 42 plusmn 13 37 plusmn 09 ndash ndashSDP11 202ndash111 57 plusmn 16 ndash 58 plusmn 14 ndash

NCv1268 211ndash202 86 plusmn 23 115 plusmn 15 ndash ndashNAv156 211ndash202 97 plusmn 26 103 plusmn 12 ndash ndashSDP81 202ndash111 61 ndash 33 ndash

NAv1144 211ndash202 11 97 ndash ndashSDP9 211ndash202 52 70 ndash ndash

G12v230 202ndash111 16 ndash 13 ndashSDP17b 202ndash111 16 ndash 20 ndash

G15v2779 211ndash202 21 266 ndash ndash

Notes LIR is the intrinsic total infrared luminosity (8ndash1000 microm) taken from B13 The intrinsic H2O luminosities are inferred from microLH2O using microin B13 The first group of the sources are the ones with both J = 2 and J = 3 H2O lines observed the next group are the sources with only J = 2H2O observed and the last group are the previous published sources in O13

Fig 3 Correlation between LIR and LH2O in local ULIRGs and high-redshift HyULIRGs The black points represent local ULIRGs from Y13The blue points with solid error bars are the H-ATLAS source in this work together with some previously published sources Red points withdashed error bars are excluded from the fit as described in the text Upper limits are shown in arrows The light blue lines show the results ofthe fitting The insets are the probability density distributions of the fitted slopes α We find tight correlations between the luminosity of the threeH2O lines and LIR namely LH2O prop LIR

11minus12

A80 page 8 of 23

The fitted parameters are α = 106plusmn019 116plusmn013 and 106plusmn022 for H2O line 202ndash111 211ndash202 and 321ndash312 respectivelyComparing with the local ULIRGs the high-redshift lensed oneshave higher LH2OLIR ratios (Table 6) These slopes confirm ourfirst result derived from 7 H2O detections in (O13) The slightsuper-linear correlations seem to indicate that far-infrared pump-ing play an important role in the excitation of the submm H2Oemission This is unlike the high-J CO lines which are deter-mined by collisional excitation and follow the linear correla-tion between the CO line luminosity and LIR from the local tothe high-redshift Universe (Liu et al 2015) As demonstrated inG14 using the far-infrared pumping model the steeper than lin-ear growth of LH2O with LIR can be the result of an increas-ing optical depth at 100 microm (τ100) with increasing LIR In localULIRGs the ratio of LH2OLIR is relatively low while most ofthem are likely to be optically thin (τ100 sim 01 G14) On theother hand for the high-redshift lensed HyULIRGs with highvalues of LIR the continuum optical depth at far-infrared wave-lengths is expected to be high (see Sect 42) indicating that theH2O emission comes from very dense regions of molecular gasthat are heavily obscured

Similar to what we found in the local ULIRGs (Y13) wefind again an anti-correlation between Td and LH2O(321ndash312)LIRThe Spearmanprimes rank correlation coefficient for the fiveH2O(321ndash312) detected H-ATLAS sources is ρ = minus09 with atwo-sided significance of its deviation from zero p = 004However after including the non-detection of H2O(321ndash312) inNAv1195 the correlation is much weaker that is to say ρ minus05 and p sim 032 No significant correlation has been foundbetween Td and LH2O(202ndash111)LIR (ρ = minus01 and p = 087) norLH2O(211ndash202)LIR (ρ = minus03 and p = 045) As explained inG14 in the optically thick and very warm galaxies the ratioof LH2O(321ndash312)LIR is expected to decrease with increasing TdAnd this anti-correlation can not be explained by optically thinconditions However a larger sample is needed to increase thestatistical significance of this anti-correlation

Although it is important to stress that the luminosity ofH2O is a complex result of various physical parameters suchas dust temperature gas density H2O abundance and H2O gasdistribution relative to the infrared radiation field etc it isstriking that the correlation between LH2O and LIR stays lin-ear from local young stellar objects (YSOs) in which the H2Omolecules are mainly excited by shocks and collisions to localULIRGs (far-infrared pumping dominated) extending sim12 or-ders of magnitudes (San Joseacute-Garciacutea et al 2016) implying thatH2O indeed traces the SFR proportionally similarly to the densegas (Gao amp Solomon 2004) in the local infrared bright galax-ies However for the high-redshift sources the LH2O emissionsare somewhat above the linear correlations which could be ex-plained by their high τ100 (or large velocity dispersion) Asshown in Table 6 HFLS3 with a τ100 gt 1 has extremely largeratios of LH2OLIR which are stronger than the average of ourH-ATLAS sources by factors sim2 for the J = 2 lines and sim4for J = 3 (see Fig 3) The velocity dispersions of its H2O linesare sim900 km sminus1 (with uncertainties from 18 to 36) which islarger than all our sources For optically thick systems larger ve-locity dispersion will increase the number of absorbed pumpingphotons and boost the ratio of LH2OLIR (G14)

For the AGN-dominated sources (ie APM 08279G09v1124-W and Mrk 231) as shown in Fig 3 most ofthem (except for the H2O(321ndash312) line of Mrk 231) are well be-low the fitted correlation (see Sect 44) This is consistent withthe average value of local strong-AGN-dominated sources TheJ 3 H2O lines are far-infrared pumped by the 75 and 101 microm

Fig 4 Velocity-integrated flux density distribution of H2O normalisedto IH2O(202ndash111) adapted from Y13 Local averaged values are shown inblack dashed line and marks Among them AGN-dominated sourcesare shown in red and star-forming dominated galaxies are shown in blueSome individual sources are also shown in this plot as indicated by thelegend Green diamonds are the high-redshift lensed HyULIRGs fromthis work HFLS3 is a z = 63 high-redshift galaxy from Riechers et al(2013)

photons thus the very warm dust in strong-AGN-dominatedsources is likely to contribute more to the LIR than the J 3H2O excitation (see also Y13)

42 H2O excitation

We have detected both J = 2 and J = 3 H2O lines in five sourcesout of six observed for J = 3 ortho-H2O lines By comparingthe line ratios and their strength relative to LIR we are able toconstrain the physical conditions of the molecular content andalso the properties of the far-infrared radiation field

To compare the H2O excitation with local galaxies we plotthe velocity-integrated flux density of ortho-H2O(321ndash312) nor-malised by that of para-H2O(202ndash111) in our source on top ofthe local and high-redshift H2O SLEDs (spectral line energydistributions) in Fig 4 All the six high-redshift sources are lo-cated within the range of the local galaxies with a 1σ disper-sion of sim02 Yet for the z = 634 extreme starburst HFLS3the value of this ratio is at least 17 times higher than the av-erage value of local sources (Y13) and those of our lensedhigh-redshift HyULIRGs at 3σ confidence level (Fig 4) Thisprobably traces different excitation conditions namely the prop-erties of the dust emission as it is suggested in G14 that theflux ratio of H2O(321ndash312) over H2O(202ndash111) is the most di-rect tracer of the hardness of the far-infrared radiation fieldwhich powers the submm H2O excitation However the line ra-tios are still consistent with the strong saturation limit in thefar-infrared pumping model with a Twarm amp 65 K The largescatter of the H2O line ratio between 321ndash312 and 202ndash111 in-dicates different local H2O excitation conditions As far-infraredpumping is dominating the H2O excitation the ratio thereforereflects the differences in the far-infrared radiation field forexample the temperature of the warmer dust that excites theH2O gas and the submm continuum opacity It is now clear thatfar-infrared pumping is the prevailing excitation mechanism forthose submm H2O lines rather than collisional excitation (G14)in infrared bright galaxies in both the local and high-redshift

A80 page 9 of 23

AampA 595 A80 (2016)

Table 6 Ratio between infrared and H2O luminosity and the velocity-integrated flux density ratio between different H2O transitions

Source H2O Transition TdLH2O(211ndash202)

LH2O(202ndash111)

LH2O(321ndash312)

IH2O(321ndash312)

IH2O (211ndash202)

IH2O(321ndash312)

IH2O (202ndash111)

(K) (times10minus6) (times10minus6) (times10minus6)G09v197 211ndash202 321ndash312 44 plusmn 1 48 plusmn 14 ndash 68 plusmn 20 09 plusmn 01 (08 plusmn 02)G12v243 202ndash111 321ndash312 (39 plusmn 2) ndash 106 plusmn 25 153 plusmn 33 ndash 12 plusmn 02NCv1143 211ndash202 321ndash312 40 plusmn 1 79 plusmn 25 ndash 166 plusmn 54 14 plusmn 01 (11 plusmn 04)NAv1195 202ndash111 321ndash312 36 plusmn 1 ndash 91 plusmn 29 lt68 ndash lt07NAv1177 202ndash111 321ndash312 (32 plusmn 1) ndash 149 plusmn 38 235 plusmn 51 ndash 13 plusmn 02NBv178 202ndash111 321ndash312 43 plusmn 1 ndash 114 plusmn 47 134 plusmn 49 ndash 10 plusmn 02

G09v1124-W 211ndash202 40 plusmn 1 lt39 ndash ndash ndash ndashG09v1124-T 211ndash202 36 plusmn 1 lt48 ndash ndash ndash ndash

G09v140 211ndash202 36 plusmn 1 88 plusmn 35 ndash ndash ndash ndashSDP11 202ndash111 41 plusmn 1 ndash 102 plusmn 38 ndash ndash ndash

NCv1268 211ndash202 39 plusmn 1 134 plusmn 39 ndash ndash ndash ndashNAv156 211ndash202 38 plusmn 1 107 plusmn 31 ndash ndash ndashSDP81 202ndash111 34 plusmn 1 ndash 54 ndash ndash ndash

NAv1144 211ndash202 39 plusmn 1 97 ndash ndash ndash ndashSDP9 211ndash202 43 plusmn 1 135 ndash ndash ndash ndash

G12v230 202ndash111 41 plusmn 1 ndash 81 ndash ndash ndashSDP17b 202ndash111 38 plusmn 1 ndash 125 ndash ndash ndash

G15v2779 211ndash202 41 plusmn 1 77 ndash ndash ndash ndash

HFLS3 202ndash111 211ndash202 321ndash312 56+9minus12 203 222 573 18 plusmn 06 22 plusmn 05

APM 08279 202ndash111 211ndash202 321ndash312 220 plusmn 30 22 60 64 19 plusmn 03 09 plusmn 01HLSJ 0918 202ndash111 38 plusmn 3 114 ndash ndash ndash ndashSPT 0538 202ndash111 39 plusmn 2 ndash 403 ndash ndash ndash

local strong-AGN 202ndash111 211ndash202 321ndash312 ndash 38 64 67 11 plusmn 04 09 plusmn 03local H ii+mild-AGN 202ndash111 211ndash202 321ndash312 ndash 58 92 108 14 plusmn 04 11 plusmn 03

Notes The luminosity ratios between each H2O line and their total infrared and the velocity-integrated flux density ratio of different H2O tran-sitions Td is the cold-dust temperature taken from B13 except for the ones in brackets which are not listed B13 that we infer them from singlemodified black-body dust SED fitting using the submmmm photometry data listed in Table 2 All the errors quoted for Td are significantly underes-timated especially because they do not include possible effects of differential lensing and make the assumption of a single-temperature Line ratiosin brackets are derived based on the average velocity-integrated flux density ratios between 211ndash202 and 202ndash111 lines in local infrared galaxies Thelocal strong-AGN sources are the optically classified AGN-dominated galaxies and the local H ii+mild-AGN sources are star-forming-dominatedgalaxies with possible mild AGN contribution (Y13) The first group of the sources are from this work and the sources in the second group arethe previously published sources in O13 the third group contains the previously published high-redshift detections from other works HFLS3(Riechers et al 2013) APM 08279 (van der Werf et al 2011) HLSJ 0918 (Combes et al 2012 Rawle et al 2014) and SPT 0538 (Bothwell et al2013) the last group shows the local averaged values from Y13

Universe The main path of far-infrared pumping related to thelines we observed here are 75 and 101 microm as displayed in Fig 1Therefore the different line ratios are highly sensitive to thedifference between the monochromatic flux at 75 and 101 micromWe may compare the global Td measured from far-infrared andsubmm bands (B13) It includes both cold and warm dust con-tribution to the dust SED in the rest-frame which is howeverdominated by cold dust observed in SPIRE bands It is thus notsurprising that we find no strong correlation between Td andIH2O(321ndash312)IH2O(202ndash111) (r sim minus03) The Rayleigh-Jeans tail ofthe dust SED is dominated by cooler dust which is associatedwith extended molecular gas and less connected to the submmH2O excitation As suggested in G14 it is indeed the warmerdust (Twarm as shown by the colour legend in Fig 5) dominatingat the Wien side of the dust SED that corresponds to the excita-tion of submm H2O lines

To further explore the physical properties of the H2O gascontent and the far-infrared dust radiation related to the submmH2O excitation we need to model how we can infer keyparameters such as the H2O abundance and those determiningthe radiation properties from the observed H2O lines For thispurpose we use the far-infrared pumping H2O excitation modeldescribed in G14 to fit the observed LH2O together with the cor-responding LIR and derive the range of continuum optical depth

at 100 microm (τ100) warm dust temperature (Twarm) and H2O col-umn density per unit of velocity interval (NH2O∆V) in the fivesources with both J = 2 and J = 3 H2O emission detectionsDue to the insufficient number of the inputs in the model whichare LH2O of the two H2O lines and LIR we are only able toperform the modelling by using the pure far-infrared pumpingregime Nevertheless our observed line ratio between J = 3and J = 2 H2O lines suggests that far-infrared pumping is thedominant excitation mechanism and the contribution from col-lisional excitation is minor (G14) The plusmn1σ contours from χ2

fitting are shown in Fig 5 for each warm dust temperature com-ponent (Twarm = 35ndash115 K) per source It is clear that with twoH2O lines (one J = 2 para-H2O and ortho-H2O(312ndash312)) wewill not be able to well constrain τ100 and NH2O∆V As shownin the figure for Twarm 75 K both very low and very highτ100 could fit the observation data together with high NH2O∆V while the dust with Twarm amp 95 K are likely favouring high τ100In the low continuum optical depth part in Fig 5 as τ100 de-creases the model needs to increase the value of NH2O∆V togenerate sufficient LH2O to be able to fit the observed LH2OLIRThis has been observed in some local sources with low τ100 suchas in NGC 1068 and NGC 6240 There are no absorption featuresin the far-infrared but submm H2O emission have been detectedin these sources (G14) The important feature of such sources is

A80 page 10 of 23

Table 7 Parameters derived from far-infrared pumping model of H2O

Source τ100 Twarm NH2O∆V NH2O(K) (cmminus2 kmminus1 s) (cmminus2)

G09v197 18 45ndash55 (03ndash06)times 1015 (03ndash11)times 1017

G12v243 ndash 45ndash95 amp07times 1015 amp07times 1017

NCv1143 72 45ndash55 (20ndash20)times 1015 (20ndash60)times 1017

NAv1177 ndash 45ndash75 amp10times 1015 amp10times 1017

NBv178 25 45ndash75 amp06times 1015 amp06times 1017

Notes τ100 is derived from Eq (2) with errors of a few units (see text)while Twarm and NH2O∆V are inferred from the H2O excitation modelNH2O is calculated by taking a typical ∆V value range of 100ndash300 km sminus1

as suggested by G14

the lack of J ge 4 H2O emission lines Thus the observation ofhigher excitation of H2O will discriminate between the low andhigh τ100 regimes

Among these five sources favoured key parameters aresomewhat different showing the range of properties we canexpect for such sources Compared with the other fourHyULIRGs G09v197 is likely to have the lowest Twarm as onlydust with Twarm sim 45minus55 K can fit well with the data NCv1143and NAv1177 have slightly different diagnostic which yieldshigher dust temperature as Twarm sim 45ndash75 K while NBv178 andG12v243 tend to have the highest temperature range Twarm sim

45ndash95 K The values of Twarm are consistent with the fact thatH2O traces warm gas We did not find any significant differ-ences between the ranges of NH2O∆V derived from the mod-elling for these five sources although G09v197 tends to havelower NH2O∆V (Table 7) As shown in Sect 44 there is no ev-idence of AGN domination in all our sources the submm H2Olines likely trace the warm dust component that connect to theheavily obscured active star-forming activity However due tothe lack of photometry data on the Wien side of the dust SEDswe will not be able to compare the observed values of Twarm di-rectly with the ones derived from the modelling

By adopting the 100 microm dust mass absorption coefficientfrom Draine (2003) of κ100 = 271 cm2 gminus1 we can derive the dustopacity by

τ100 = κ100 σdust = κ100

( Mdust

)= κ100

2πr2half

where σdust is the dust mass column density Mdust is the dustmass A is the projected surface area of the dust continuumsource and rhalf is the half-light radius of the source at submmAs shown in Table 2 among the five sources in Fig 5 the val-ues of Mdust and rhalf in G09v197 NCv1143 and NBv178 havebeen derived via gravitational lensing (B13) Consequently thederived approximate dust optical depth at 100 microm in these threesources is τ100 asymp 18 72 and 25 respectively One should notethat the large uncertainty in both the κ100 and rhalf of thesehigh-redshift galaxies can bring a factor of few error budgetNevertheless by adopting a gas-to-dust mass ratio of X = 100(eg Magdis et al 2011) we can derive the gas depletion timeusing the following approach

tdep =Mgas

Xτ100

ΣSFRκ100asymp 18 times 104

τ100ΣSFR

M yrminus1 kpcminus2

Myr (3)

where Mgas is the total molecular gas mass and ΣSFR is thesurface SFR density derived from LIR using Kennicutt (1998)

calibration by assuming a Salpeter IMF (B13 and Table 2) Theimplied depletion time scale is tdep asymp 35ndash60 Myr with errorswithin a factor of two in which the dominant uncertainties arefrom the assumed gas-to-dust mass ratio and the half-light ra-dius The tdep is consistent with the values derived from dense gastracers like HCN in local (U)LIRGs (eg Gao amp Solomon 2004Garciacutea-Burillo et al 2012) As suggested in G14 the H2O andHCN likely to be located in the same regions indicate that theH2O traces the dense gas as well Thus the τ100 derived aboveis likely also tracing the far-infrared radiation source that pow-ers the submm H2O emissions B13 also has found that theseH-ATLAS high-redshift HyULIRGs are expected to be opti-cally thick in the far-infrared By adding the constrain from τ100above we can better derive the physical conditions in the sourcesas shown in Table 7

From their modelling of local infrared galaxies G14 find arange of Twarm = 45ndash75 K τ100 = 005ndash02 and NH2O∆V =(05ndash2) times 1015 cmminus2 kmminus1 s The modelling results for ourhigh-redshift sources are consistent with those in local galaxiesin terms of Twarm and NH2O∆V However the values of τ100 wefound at high-redshift are higher than those of the local infraredgalaxies This is consistent with the higher ratio between LH2Oand LIR at high-redshift (Y13) which could be explained byhigher τ100 (G14) However as demonstrated in an extreme sam-ple a very large velocity dispersion will also increase the valueof LH2OLIR within the sources with τ100 gt 1 Thus the higherratio can also be explained by larger velocity dispersion (notincluding systemic rotations) in the high-redshift HyULIRGsCompared with local ULIRGs our H-ATLAS sources are muchmore powerful in terms of their LIR The dense warm gas re-gions that H2O traces are highly obscured with much more pow-erful far-infrared radiation fields which possibly are close to thelimit of maximum starbursts Given the values of dust temper-ature and dust opacity the radiation pressure Prad sim τ100σTdc(σ is Stefan-Boltzmannprimes constant and c the speed of light) ofour sources is about 08times 10minus7 erg cmminus3 If we assume a H2 den-sity nH2 of sim106 cm minus3 and take Tk sim 150 K as suggested in G14the thermal pressure Pth sim nH2 kBTk sim 2 times 10minus8 erg cmminus3 (kBis the Boltzmann constant and Tk is the gas temperature) As-suming a turbulent velocity dispersion of σv sim 20ndash50 km sminus1

(Bournaud et al 2015) and taking molecular gas mass densityρ sim 2micronH2 (2micro is the average molecular mass) would yield forthe turbulent pressure Pturb sim ρσ

2v3 sim 4 times 10minus6 erg cmminus3 This

might be about an order of magnitude larger than Prad and twoorders of magnitude larger than Pth but we should note that allvalues are very uncertain especially Pturb which could be uncer-tain by at maximum a factor of a few tens Therefore keepingin mind their large uncertainties turbulence andor radiation arelikely to play an important role in limiting the star formation

43 Comparison between H2O and CO

The velocity-integrated flux density ratio between submm H2Oand submm CO lines with comparable frequencies is 002ndash003in local PDRs such as Orion and M 82 (Weiszlig et al 2010) Butthis ratio in local ULIRGs (Y13) and in H-ATLAS high-redshiftHyULIRGs is much higher from 04 to 11 (Tables 3 and 4)The former case is dominated by typical PDRs where CO linesare much stronger than H2O lines while the latter sources showsclearly a different excitation regime in which H2O traces thecentral core of warm dense and dusty molecular gas which isabout a few hundred parsec (Gonzaacutelez-Alfonso et al 2010) in di-ameter in local ULIRGs and highly obscured even at far-infrared

A80 page 11 of 23

AampA 595 A80 (2016)

Fig 5 Parameter space distribution of the H2O far-infrared pumping excitation modelling with observed para-H2O 202ndash111 or 211ndash202 and ortho-H2O(321ndash312) in each panel plusmn1σ contours are shown for each plot Different colours with different line styles represent different temperaturecomponents of the warm dust as shown in the legend The explored warm dust temperature range is from 35 K to 115 K The temperature contoursthat are unable to fit the data are not shown in this figure From the figure we are able to constrain the τ100 Twarm and NH2O∆V for the five sourcesHowever there are strong degeneracies Thus we need additional information such as the velocity-integrated flux densities of J ge 4 H2O linesto better constrain the physical parameters

Generally submm H2O lines are dominated by far-infraredpumping that traces strong far-infrared dust continuum emissionwhich is different from the regime of molecular gas traced bycollisional excited CO lines In the active star-forming nucleusof the infrared-bright galaxies the far-infrared pumped H2O isexpected to trace directly the far-infrared radiation generated bythe intense star formation which can be well correlated with thehigh-J CO lines (Liu et al 2015) Thus there is likely to be acorrelation between the submm H2O and CO emission Fromour previous observations most of the H2O and CO line pro-files are quite similar from the same source in our high-redshiftlensed HyULIRGs sample (Fig 2 of O13) In the present workwe again find similar profiles between H2O and CO in terms oftheir FWHM with an extended sample (Table 3 and 4) In both

cases the FWHMs of H2O and CO are generally equal withintypical 15σ errors (see special discussion for each source inAppendix A)

As the gravitational lensing magnification factor is sen-sitive to spatial alignment the similar line profiles couldthus suggest similar spatial distributions of the two gastracers However there are a few exceptional sourcessuch as SDP 81 (ALMA Partnership Vlahakis et al 2015) andHLSJ0918 (Rawle et al 2014) In both cases the H2O linesare lacking the blue velocity component found in the CO lineprofiles Quite different from the rest sources in SDP 81 andHLSJ0918 the CO line profiles are complicated with multi-ple velocity components Moreover the velocity-integrated fluxdensity ratios between these CO components may vary following

A80 page 12 of 23

the excitation level (different Jup) Thus it is important to anal-yse the relation between different CO excitation components(from low-J to high-J) and H2O Also high resolution obser-vation is needed to resolve the multiple spatial gas componentsand compare the CO emission with H2O and dust continuumemission within each component

44 AGN content

It is still not clear how a strong AGN could affect the excita-tion of submm H2O in both local ULIRGs and high-redshiftHyULIRGs Nevertheless there are some individual stud-ies addressing this question For example in APM 08279van der Werf et al (2011) found that AGN is the main powersource that excites the high-J H2O lines and also enriches thegas-phase H2O abundance Similar conclusion has also beendrawn by Gonzaacutelez-Alfonso et al (2010) that in Mrk 231 theAGN accounts for at least 50 contribution to the far-infraredradiation that excites H2O From the systematic study of lo-cal sources (Y13) slightly lower values of LH2OLIR are foundin strong-AGN-dominated sources In the present work the de-creasing ratio of LH2OLIR with AGN is clearly shown in Fig 3where Mrk 231 G09v1124-W and APM 08279 are below thecorrelation by factors between 2 and 5 with less than 30 uncer-tainties (except the H2O(321ndash123) line of Mrk 231)

In the far-infrared pumping regime the buried AGN will pro-vide a strong far-infrared radiation source that will pump theH2O lines However the very warm dust powered by the AGNwill increase the value of LIR faster than the number of ge75 micromphotons that is dominating the excitation of J le 3 H2O lines(eg Kirkpatrick et al 2015) If we assume that the strength ofthe H2O emission is proportional to the number of pumpingphotons then in the strong-AGN-dominated sources the ratioof LH2OLIR will decrease since much warmer dust is presentMoreover strong radiation from the AGN could dissociate theH2O molecules

To evaluate the AGN contribution to the H-ATLAS sourceswe extracted the 14 GHz radio flux from the FIRST radio sur-vey (Becker et al 1995) listed in Table 2 By comparing thefar-infrared and radio emission using the q parameter (Condon1992) q equiv log(LFIR375 times 1012 W) minus log(L14 GHz1 W Hzminus1)we derive values of q from 19 to 25 in our sources These val-ues follow the value 23plusmn 01 found by Yun et al (2001) for nonstrong-radio AGN This may suggest that there is also no signif-icant indication of a high radio contribution from AGN Thisis also confirmed by the Wide-field Infrared Survey Explorer(WISE Wright et al 2010) which does not detect our sourcesat 12 microm and 22 microm However rest-frame optical spectral obser-vations show that G09v1124-W is rather a powerful AGN (Oteoet al in prep) which is the only identified AGN-dominatedsource in our sample

5 Detection of H2O+ emission lines

H2O can be formed through both solid-state and gas-phasechemical reactions (van Dishoeck et al 2013) On dust-grainmantles surface chemistry dominates the formation of H2Omolecules Then they can be released into the ISM gas throughsublimation In the gas phase H2O can be produced through tworoutes the neutral-neutral reaction usually related to shockscreates H2O via O + H2 minusminusrarr OH + H OH + H2 minusminusrarr H2O + Hat high temperature (amp300 K) At lower temperature (100 K)the ion-neutral reactions in photon-dominated regions (PDRs)cosmic-ray-dominated regions and X-ray-dominated regions

(eg Meijerink amp Spaans 2005) generate H2O from O H+ H+3

and H2 with intermediates such as O+ OH+ H2O+ and H3O+and finally H3O+ + e minusminusrarr H2O + H However classical PDRs arenot likely linked to these highly excited submm H2O emissions(Y13) Therefore H2O+ lines are important for distinguishingbetween shock- or ion-chemistry origin for H2O in the early Uni-verse indicating the type of physical regions in these galaxiesshock-dominated regions cosmic-ray-dominated regions or X-ray-dominated regions Indeed they can be among the most di-rect tracers of the cosmic-ray orand X-ray ionization rate (egGeacuterin et al 2010 Neufeld et al 2010 Gonzaacutelez-Alfonso et al2013) of the ISM which regulates the chemistry and influ-ences many key parameters for example X-factor (Bell et al2007) that connects the CO luminosity to the H2 mass More-over the significant detections of H2O+ emission in high-redshiftHyULIRGs could help us understanding H2O formation in theearly Universe

When observing our sources with redshift z amp 33 it ispossible to cover all the following lines with the NOEMAWideX bandwidth para-H2O(211ndash202) at 752 GHz and fourortho-H2O+ lines (two intertwined fine structure doublets of twodifferent lines whose frequencies almost coincide by chance)202ndash111 (52minus32) at 7421 GHz 211ndash202 (52minus32) at 7423 GHz202ndash111 (32minus32) at 7463 GHz and 211ndash202 (52minus52) at 7465 GHzin the 36 GHz band simultaneously (the rest-frame frequen-cies are taken from the CDMS catalogue httpwwwastrouni-koelndecdms see energy level diagram of H2O+ inFig 1 and the full spectra in Fig 6) Additionally within thisrange we can also cover the H18

2 O(211ndash202) line at 7453 GHzThere are three sources of our sample that have been observedin such a frequency setup NCv1143 NCv1268 and G09v197We have also included the source G15v2779 from our previousobservation (O13) in which we have covered both H2O(211ndash202)at 752 GHz and H2O+ lines around 746 GHz We have detectedboth main lines of H2O+ in NCv1143 and tentatively detectedone line in G09v197 and G15v2779 (Fig 6) For NCv1268due to the large noise level and the complex line profile we werenot able to really identify any H2O+ line detection

As shown in Fig 6 in NCv1143 the dominant H2O+ finestructure lines 211ndash202 (52minus52) at 7465 GHz and 202ndash111 (52minus32)at 7421 GHz are well detected The velocity-integrated flux den-sities of the two lines from a two-Gaussian fit are 19 plusmn 03and 16 plusmn 02 Jy km sminus1 respectively These are the approxi-mate velocity-integrated flux densities of the dominant H2O+

lines 211ndash202 (52minus52) and 202ndash111 (52minus32) if neglecting the mi-nor contributions from H2O+ lines 202ndash111 (32minus32) at 7462 GHzand 211ndash202 (52minus32) at 7423 GHz However the H2O+ line pro-file at 7465 GHz is slightly wider than the H2O line (Fig 6)probably due to a contribution from the fairly weak fine struc-ture line H2O+(202ndash111) (32minus32) at 7463 GHz The ratio be-tween total velocity-integrated flux density of the H2O+ linesand H2O(211ndash202) is 060 plusmn 007 (roughly 03 for each dominantH2O+ line) being consistent with the average value from thelocal infrared galaxies (Y13)3 In order to derive the velocity-integrated flux density of each fine structure doublets around742 and 746 GHz we have also performed a four-Gaussian fitwith fixed line positions (equal to νrest(1 + z)) and linewidth(equals to that of H2O(211ndash202)) We find the velocity-integrated

3 As suggested by Gonzaacutelez-Alfonso et al (2013) due to thevery limited spectral resolution of HerschelSPIRE FTS the ortho-H2O+(202ndash111) (32minus32) line at 7465 GHz quoted in Y13 is actuallydominated by ortho-H2O+(211ndash202) (52minus52) considering their likely ex-citation and relative strength

A80 page 13 of 23

AampA 595 A80 (2016)

Fig 6 Left panel from top to bottom are the full NOEMA spectrum at νrest sim 750 GHz of NCv1143 G09v197 and G15v2779 respec-tively The reference frequency is the redshifted frequency of the line H2O(211ndash202) The frequencies of the main H2O+(211ndash202) (52minus52) andH2O+(202ndash111) (52minus32) lines are indicated by grey vertical dashed lines The three dashed squares in the spectrum of NCv1143 show the position ofeach zoom-in spectrum of the H2O+ (or the H18

2 O) as displayed in the right panel indicated by the A B or C The superposed blue dashed histogramsrepresents the spectra of H2O(211ndash202) centred at the frequencies of the H2O+ lines Note that in many cases the observed frequency ranges (yel-low histograms) do not include the full expected profiles for the H2O+ lines The red curve represents the Gaussian fitting to the spectra We havedetected both H2O+ lines in NCv1143 and tentatively detected H2O+(202ndash111) (52minus32) in G09v197 and H2O+(211ndash202) (52minus52) in G15v2779Right panel from top to bottom are the spectra dominated by lines of H2O+(211ndash202) (52minus52) H2O+(202ndash111) (32minus32) and H18

2 O(211ndash202) respec-tively displayed as the filled yellow histograms The reference frequency is the frequency of each of these lines Weaker H2O+(202ndash111) (32minus32)and H2O+(211ndash202) (52minus32) components are indicated by additional grey vertical dashed lines The superposed blue dashed histograms representthe spectra of para-H2O(211ndash202) in NCv1143 centred at each line frequency The red curve represents the Gaussian fitting to the spectra and thegreen dashed curves are the decomposed Gaussian profiles for each fine structure line The violet error bar indicates the plusmn1σ uncertainties of thespectrum

flux densities of the two fine structure lines of H2O+(211ndash202)are 16 plusmn 05 and 03 plusmn 04 Jykm sminus1 while they are 16 plusmn 04and 02 plusmn 05 Jykm sminus1 for the two fine structure lines ofH2O+(202ndash111) (Table 8) We should note that these fitting re-sults have much larger uncertainties due to the blending Never-theless they are consistent with the earlier fitting results with-out de-blending The similarity of the velocity-integrated fluxdensities between the H2O+(202ndash111) and H2O+(211ndash202) lines isin good agreement with the regime of far-infrared pumping assubmm H2O (Gonzaacutelez-Alfonso et al 2013) As a first approxi-mation if these H2O+ lines are optically thin and we ignore theadditional pumping from ortho-H2O+ 202 to ortho-H2O+ J = 3energy levels the statistical equilibrium applied to energy level202 52 implies that all population arriving per second at 202 52should be equal to all population leaving the level per second

After subtracting the Gaussian profiles of all the H2O+

lines in the spectrum we find a 3σ residual in terms of the

velocity-integrated flux density around 7453 GHz (I = 06 plusmn02 Jy km sminus1 see Fig 6) This could be a tentative detection ofthe H18

2 O(211ndash202) line at 745320 GHz The velocity-integratedflux density ratio of H18

2 O(211ndash202) over H2O(211ndash202) inNCv1143 would hence be sim01 If this tentative detection wasconfirmed it would show that ALMA could easily study suchlines But sophisticated models will be needed to infer isotoperatios

The spectrum of the H2O(211ndash202) line in G09v197 coversboth the two main H2O+ fine structure lines (Fig 6) How-ever due to the limited sensitivity we have only tentatively de-tected the H2O+(202ndash111) (52minus32) line just above 3σ (neglect-ing the minor contribution from H2O+(211ndash202) (52minus32)) and thevelocity-integrated flux density is 14plusmn04 Jy km sminus1 using a sin-gle Gaussian fit We did not perform any line de-blending forthis source considering the data quality The H2O+ line profile isin good agreement with that of the H2O (blue dashed histogram

A80 page 14 of 23

Table 8 Observed ortho-H2O+ fine structure line parameters of thehigh-redshift H-ATLAS lensed HyLIRGs

Source H2O+ transition νrest νline IH2O+

(GHz) (GHz) (Jy km sminus1)NCv1143 211ndash202 (52minus52) 7465 16353 16 plusmn 05

202ndash111 (32minus32) 7463 16348 02 plusmn 05211ndash202 (52minus32) 7423 16261 03 plusmn 04202ndash111 (52minus32) 7421 16256 16 plusmn 04

G09v197 202ndash111 (52minus32) 7421 16014 14 plusmn 04G15v2779 211ndash202 (52minus52) 7465 14235 12 plusmn 03

Notes The H2O+ (202ndash111) (52minus32) line in G09v197 is blended by(211ndash202) (52minus32) and H2O+ (211ndash202) (52minus52) line in G15v2779 isblended by (202ndash111) (32minus32) However the contribution from the lat-ter in each case is small likely less than 20 as shown in the caseof the H2O+ lines in NCv1143 Note that the quoted uncertainties donot include the missing parts of the spectra cut by the limited observedbandwidth (Fig 6)

in Fig 7) The velocity-integrated flux density of the undetectedH2O+(211ndash202) (52minus52) line could also be close to this value asdiscussed in the case of NCv1143 yet somewhat lower andnot detected in this source More sensitive observation shouldbe conducted to further derive robust line parameters

We have also tentatively detected the H2O+(211ndash202) (52minus52)line in G15v2779 (SN sim 4 by neglecting the minor contribu-tion from the H2O+(202ndash111) (32minus32) line) The line profile isin good agreement with that of H2O(211ndash202) (blue dashed his-togram in Fig 6) The velocity-integrated flux density derivedfrom a double-peak Gaussian fit is 12 plusmn 03 Jy km sminus1 (we didnot perform any line de-blending for the H2O+ doublet consider-ing the spectral noise level) There could be a minor contributionfrom the H2O+(202ndash111) (32minus32) line to the velocity-integratedflux density However such a contribution is likely to be negli-gible as in the case of NCv1143 The contribution is also withinthe uncertainty of the velocity-integrated flux density Neverthe-less the position of H2O+ has a small blue-shift compared withH2O but note that the blue part of the line is cut by the limitedobserved bandwidth (yellow histogram)

After including the local detections of H2O+ lines fromY13 (Table B1) we find a tight linear correlation between theluminosity of H2O and the two main lines of H2O+ (slopesequal to 103 plusmn 006 and 091 plusmn 007 see Fig 7) Howeverone should keep in mind that because the local measure-ment done by Herschel SPIREFTS (Naylor et al 2010) hasrather low spectral resolution neither H2O+(211ndash202) (52minus32)and H2O+(202ndash111) (52minus32) nor H2O+(211ndash202) (52minus52) andH2O+(202ndash111) (32minus32) can be spectroscopically resolved In thecorrelation plot (Fig 7) and Table B1 we use the total luminos-ity from the 742 GHz and 746 GHz lines by assuming the contri-bution from H2O+(211ndash202) (52minus32) and H2O+(202ndash111) (32minus32)to the velocity-integrated flux density of the line at 742 GHz and746 GHz is small (sim18) and does not vary significantly be-tween different sources Hence the velocity-integrated flux den-sity ratio between each of the two dominant H2O+ fine structurelines and H2O in NCv1143 G15v2779 and G09v197 is sim03(uncertainties are less than 30) which is consistent with localgalaxies as shown in the figure This ratio is much larger thanthe abundance ratio of H2O+H2Osim 005 found in Arp 220 ananalogue of high-redshift ULIRGs (Rangwala et al 2011)

As discussed above the AGN contribution to the excita-tion of the submm lines of most of our sources appears to beminor Thus the formation of H2O+ is likely dominated bycosmic-ray ionization rather than X-ray ionization Given the

Fig 7 Correlation between the luminosity of J = 2 ortho-H2O+ andpara-H2O(211ndash202) The fitted function is LH2O+ prop LH2O

α We found avery good correlation between LH2O+ and LH2O with a slope close toone Black points are from the local ULIRGs as listed in Table B1 Darkblue ones are high-redshift starbursts from this work Black solid linesindicate the χ2 fitting results while the grey dashed lines and the greyannotations represent the average ratio between the LH2O+ and LH2O

average luminosity ratio of H2O+H2O sim 03 plusmn 01 shown inFig 7 Meijerink et al (2011) suggest a cosmic-ray ionizationrate of 10minus14ndash10minus13 sminus1 Such high cosmic-ray ionization ratesdrive the ambient ionization degree of the ISM to 10minus3ndash10minus2rather than the canonical 10minus4 Therefore in the gas phase anion-neutral route likely dominates the formation of H2O How-ever H2O can also be enriched through the water-ice sublimationthat releases H2O into the gas-phase ISM As the upper partsim90 K of the possible range for Twarm is close to the sublima-tion temperature of water ice Hence the high H2O abundance(NH2O 03 times 1017 cmminus2 see Sect 42) observed is likely to bethe result of ion chemistry dominated by high cosmic-ray ion-ization andor perhaps water ice desorption However furtherobservation of H2O+ lines of different transitions and a largersample is needed to constrain the contribution to H2O formationfrom neutral-neutral reactions dominated by shocks

6 Conclusions

In this paper we report a survey of submm H2O emission at red-shift z sim 2ndash4 by observing a higher excited ortho-H2O(321ndash312)in 6 sources and several complementary J = 2 para-H2O emis-sion lines in the warm dense cores of 11 high-redshift lensed ex-treme starburst galaxies (HyULIRGs) discovered by H-ATLASSo far we have detected an H2O line in most of our observa-tions of a total sample of 17 high-redshift lensed galaxies inother words we have detected both J = 2 para-H2O and J = 3ortho-H2O lines in five and in ten other sources only one J = 2para-H2O line In these high-redshift HyULIRGs H2O is thesecond strongest molecular emitter after CO within the submmband as in local ULIRGs The spatially integrated H2O emis-sion lines have a velocity-integrated flux density ranging from 4to 15 Jy km sminus1 which yields the apparent H2O emission lumi-nosity microLH2O ranging from sim6ndash22times108 L After correction forgravitation lensing magnification we obtained the intrinsic LH2Ofor para-H2O lines 202ndash111 211ndash202 and ortho-H2O(321ndash312)The luminosities of the three H2O lines increase with LIR asLH2O prop LIR

11ndash12 This correlation indicates the importance offar-infrared pumping as a dominant mechanism of H2O excita-tion Comparing with J = 3 to J = 6 CO lines the linewidthsbetween H2O and CO are similar and the velocity-integratedflux densities of H2O and CO are comparable The similarity inline profiles suggests that these two molecular species possiblytrace similar intense star-forming regions

Using the far-infrared pumping model we have analysed theratios between J = 2 and J = 3 H2O lines and LH2OLIR in5 sources with both J H2O lines detected We have derived the

A80 page 15 of 23

AampA 595 A80 (2016)

ranges of the warm dust temperature (Twarm) the H2O columndensity per unit velocity interval (NH2O∆V) and the optical depthat 100 microm (τ100) Although there are strong degeneracies thesemodelling efforts confirm that similar to those of local ULIRGsthese submm H2O emissions in high-redshift HyULIRGs tracethe warm dense gas that is tightly correlated with the massivestar forming activity While the values of Twarm and NH2O (byassuming that they have similar velocity dispersion ∆V) are sim-ilar to the local ones τ100 in the high-redshift HyULIRGs islikely to be greater than 1 (optically thick) which is larger thanτ100 = 005ndash02 found in the local infrared galaxies Howeverwe notice that the parameter space is still not well constrained inour sources through H2O excitation modelling Due to the lim-ited excitation levels of the detected H2O lines we are only ableto perform the modelling with pure far-infrared pumping

The detection of relatively strong H2O+ lines opens the pos-sibility to help understanding the formation of such large amountof H2O In these high-redshift HyULIRGs the H2O formationis likely to be dominated by ion-neutral reactions powered bycosmic-ray-dominated regions The velocity-integrated flux den-sity ratio between H2O+ and H2O (IH2O+IH2O sim 03) is remark-ably constant from low to high-redshift reflecting similar con-ditions in HyULIRGs However more observations of H2O+

emissionabsorption and also OH+ lines are needed to furtherconstrain the physical parameters of the cosmic-ray-dominatedregions and the ionization rate in those regions

We have demonstrated that the submm H2O emission linesare strong and easily detectable with NOEMA Being a uniquediagnostic the H2O emission offers us a new approach to con-strain the physical conditions in the intense and heavily ob-scured star-forming regions dominated by far-infrared radiationat high-redshift Follow-up observations of other gas tracers forinstance CO HCN H2O+ and OH+ using the NOEMA IRAM30m and JVLA will complement the H2O diagnostic of the struc-ture of different components dominant physical processes starformation and chemistry in high-redshift HyULIRGs

With unprecedented spatial resolution and sensitivity theimage from the ALMA long baseline campaign observationof SDP 81 (also known as H-ATLAS J0903116+003906ALMA Partnership Vlahakis et al 2015 Dye et al 2015Rybak et al 2015) shows the resolved structure of the dustCO and H2O emission in the z = 3 ULIRG With carefulreconstruction of the source plane images ALMA will help toresolve the submm H2O emission in high-redshift galaxies intothe scale of giant molecular clouds and provide a fresh view ofdetailed physics and chemistry in the early Universe

Acknowledgements We thank our referee for the very detail comments and sug-gestions which have improved the paper This work was based on observationscarried out with the IRAM Interferometer NOEMA supported by INSUCNRS(France) MPG (Germany) and IGN (Spain) The authors are grateful to theIRAM staff for their support CY thanks Claudia Marka and Nicolas Billot fortheir help of the IRAM 30 mEMIR observation CY also thanks Zhi-Yu Zhangand Ivaacuten Oteo for insightful discussions CY AO and YG acknowledgesupport by NSFC grants 11311130491 11420101002 and CAS Pilot B pro-gram XDB09000000 CY and YG also acknowledge support by NSFC grants11173059 CY AO AB and YG acknowledge support from the Sino-French LIA-Origin joint exchange program EG-A is a Research Associate atthe Harvard-Smithsonian Center for Astrophysics and thanks the Spanish Minis-terio de Economiacutea y Competitividad for support under projects FIS2012-39162-C06-01 and ESP2015-65597-C4-1-R and NASA grant ADAP NNX15AE56GRJI acknowledges support from ERC in the form of the Advanced Investi-gator Programme 321302 COSMICISM US participants in H-ATLAS ac-knowledge support from NASA through a contract from JPL Italian partici-pants in H-ATLAS acknowledge a financial contribution from the agreementASI-INAF I009100 SPIRE has been developed by a consortium of institutesled by Cardiff Univ (UK) and including Univ Lethbridge (Canada) NAOC(China) CEA LAM (France) IFSI Univ Padua (Italy) IAC (Spain) Stockholm

Observatory (Sweden) Imperial College London RAL UCL-MSSL UKATCUniv Sussex (UK) and Caltech JPL NHSC Univ Colorado (USA) This de-velopment has been supported by national funding agencies CSA (Canada)NAOC (China) CEA CNES CNRS (France) ASI (Italy) MCINN (Spain)SNSB (Sweden) STFC UKSA (UK) and NASA (USA) CY is supported bythe China Scholarship Council grant (CSC No 201404910443)

ReferencesALMA Partnership Vlahakis C Hunter T R Hodge J A et al 2015 ApJ

808 L4Becker R H White R L amp Helfand D J 1995 ApJ 450 559Bell T A Viti S amp Williams D A 2007 MNRAS 378 983Bothwell M S Aguirre J E Chapman S C et al 2013 ApJ 779 67Bournaud F Daddi E Weiszlig A et al 2015 AampA 575 A56Bradford C M Bolatto A D Maloney P R et al 2011 ApJ 741 L37Bussmann R S Peacuterez-Fournon I Amber S et al 2013 ApJ 779 25 (B13)Calanog J A Fu H Cooray A et al 2014 ApJ 797 138Casey C M Narayanan D amp Cooray A 2014 Phys Rep 541 45Combes F Rex M Rawle T D et al 2012 AampA 538 L4Condon J J 1992 ARAampA 30 575Cooray A Calanog J Wardlow J L et al 2014 ApJ 790 40Draine B T 2003 ARAampA 41 241Dye S Furlanetto C Swinbank A M et al 2015 MNRAS 452 2258Eales S Dunne L Clements D et al 2010 PASP 122 499Gao Y amp Solomon P M 2004 ApJ 606 271Gao Y Carilli C L Solomon P M amp Vanden Bout P A 2007 ApJ 660

L93Garciacutea-Burillo S Usero A Alonso-Herrero A et al 2012 AampA 539 A8Geacuterin M de Luca M Black J et al 2010 AampA 518 L110Gonzaacutelez-Alfonso E Smith H A Fischer J amp Cernicharo J 2004 ApJ 613

247Gonzaacutelez-Alfonso E Smith H A Ashby M L N et al 2008 ApJ 675

303Gonzaacutelez-Alfonso E Fischer J Isaak K et al 2010 AampA 518 L43Gonzaacutelez-Alfonso E Fischer J Graciaacute-Carpio J et al 2012 AampA 541 A4Gonzaacutelez-Alfonso E Fischer J Bruderer S et al 2013 AampA 550 A25Gonzaacutelez-Alfonso E Fischer J Aalto S amp Falstad N 2014 AampA 567 A91

(G14)Harris A I Baker A J Frayer D T et al 2012 ApJ 752 152Ivison R J Swinbank A M Smail I et al 2013 ApJ 772 137Kamenetzky J Glenn J Rangwala N et al 2012 ApJ 753 70Kelly B C 2007 ApJ 665 1489Kennicutt Jr R C 1998 ARAampA 36 189Kessler M F Steinz J A Anderegg M E et al 1996 AampA 315 L27Kirkpatrick A Pope A Sajina A et al 2015 ApJ 814 9Lis D C Neufeld D A Phillips T G Gerin M amp Neri R 2011 ApJ 738

L6Liu D Gao Y Isaak K et al 2015 ApJ 810 L14Lupu R E Scott K S Aguirre J E et al 2012 ApJ 757 135Magdis G E Daddi E Elbaz D et al 2011 ApJ 740 L15Markwardt C B 2009 in Astronomical Data Analysis Software and Systems

XVIII eds D A Bohlender D Durand amp P Dowler ASP Conf Ser 411251

Meijerink R amp Spaans M 2005 AampA 436 397Meijerink R Spaans M Loenen A F amp van der Werf P P 2011 AampA 525

A119Meijerink R Kristensen L E Weiszlig A et al 2013 ApJ 762 L16Naylor D A Baluteau J-P Barlow M J et al 2010 in SPIE Conf Ser

7731 16Negrello M Hopwood R De Zotti G et al 2010 Science 330 800Neufeld D A Goicoechea J R Sonnentrucker P et al 2010 AampA 521 L10Omont A Neri R Cox P et al 2011 AampA 530 L3Omont A Yang C Cox P et al 2013 AampA 551 A115 (O13)Pascale E Auld R Dariush A et al 2011 MNRAS 415 911Pellegrini E W Smith J D Wolfire M G et al 2013 ApJ 779 L19Pereira-Santaella M Spinoglio L Busquet G et al 2013 ApJ 768 55Pilbratt G L Riedinger J R Passvogel T et al 2010 AampA 518 L1Rangwala N Maloney P R Glenn J et al 2011 ApJ 743 94Rawle T D Egami E Bussmann R S et al 2014 ApJ 783 59Riechers D A Bradford C M Clements D L et al 2013 Nature 496 329Rybak M McKean J P Vegetti S Andreani P amp White S D M 2015

MNRAS 451 L40San Joseacute-Garciacutea I Mottram J C van Dishoeck E F et al 2016 AampA

585 A103Serjeant S 2012 MNRAS 424 2429Solomon P M Downes D amp Radford S J E 1992 ApJ 387 L55Spergel D N Verde L Peiris H V et al 2003 ApJS 148 175

A80 page 16 of 23

Spinoglio L Pereira-Santaella M Busquet G et al 2012 ApJ 758 108van der Werf P P Isaak K G Meijerink R et al 2010 AampA 518 L42van der Werf P P Berciano Alba A Spaans M et al 2011 ApJ 741 L38van Dishoeck E F Herbst E amp Neufeld D A 2013 Chemical Reviews 113

9043Vieira J D Marrone D P Chapman S C et al 2013 Nature 495 344Weiszlig A Requena-Torres M A Guumlsten R et al 2010 AampA 521 L1Weiszlig A De Breuck C Marrone D P et al 2013 ApJ 767 88Wright E L Eisenhardt P R M Mainzer A K et al 2010 AJ 140

1868Yang C Gao Y Omont A et al 2013 ApJ 771 L24 (Y13)Yun M S Reddy N A amp Condon J J 2001 ApJ 554 803

1 Purple Mountain ObservatoryKey Lab of Radio Astronomy Chi-nese Academy of Sciences 210008 Nanjing PR Chinae-mail yangchtpmoaccn

2 Institut drsquoAstrophysique Spatiale CNRS Univ Paris-Sud Univer-siteacute Paris-Saclay Bacirct 121 91405 Orsay Cedex France

3 Graduate University of the Chinese Academy of Sciences 19AYuquan Road Shijingshan District 10049 Beijing PR China

4 CNRS UMR 7095 Institut dprimeAstrophysique de Paris 75014 ParisFrance

5 UPMC Univ Paris 06 UMR 7095 Institut drsquoAstrophysique deParis 75014 Paris France

6 Universidad de Alcalaacute Departamento de Fisica y MatemaacuteticasCampus Universitario 28871 Alcalaacute de Henares Madrid Spain

7 Institut de Radioastronomie Millimeacutetrique (IRAM) 300 rue de laPiscine 38406 Saint-Martin-drsquoHegraveres France

8 Leiden Observatory Leiden University PO Box 9513 2300 RALeiden The Netherlands

9 Max Planck Institut fuumlr Radioastronomie Auf dem Hgel 69 53121Bonn Germany

10 Department of Earth and Space Sciences Chalmers University ofTechnology Onsala Space Observatory 43992 Onsala Sweden

11 Department of Physics and Astronomy Rutgers The State Univer-sity of New Jersey 136 Frelinghuysen Road PiscatawayNJ 08854-8019 USA

12 Astronomy Department Cornell University 220 Space SciencesBuilding Ithaca NY 14853 USA

13 Department of Physics and Astronomy University of CaliforniaIrvine Irvine CA 92697 USA

14 Joint ALMA Observatory 3107 Alonso de Coacuterdova VitacuraSantiago Chile

15 Universitat Wien Institut fuumlr Astrophysik Tuumlrkenschanzstrasse 171180 Wien Austria

16 School of Physics and Astronomy University of Nottingham Uni-versity Park Nottingham NG7 2RD UK

17 Institute for Astronomy University of Edinburgh Royal Observa-tory Blackford Hill Edinburgh EH9 3HJ UK

18 European Southern Observatory Karl Schwarzschild Straszlige 285748 Garching Germany

19 Kapteyn Astronomical Institute University of Groningen 9700 AVGroningen The Netherlands

20 School of Physics and Astronomy Cardiff University The ParadeCardiff CF24 3AA UK

A80 page 17 of 23

AampA 595 A80 (2016)

Appendix A Individual sources

In the Appendix we describe the observational results ofeach source including the lensing model the H2O spectrummapping of the H2O and continuum emission (by showing low-resolution NOEMA H2O and continuum images) the ratio be-tween velocity-integrated flux densities of different H2O transi-tions and the comparison between H2O and CO emission

A1 G09v197 at z = 3634

The galaxy G09v197 has the second largest redshift inour sample obtained by CARMA (Riechers et al inprep) In the SMA 880 microm image (B13) similar to SDP81(ALMA Partnership Vlahakis et al 2015 Dye et al 2015) itdisplays a triple arc structure with an angular size scale of sim2primeprimeHowever there are two foreground deflectors at two differentredshifts making this complex mass distribution a very unusualcase B13 estimate a lensing amplification micro = 69 plusmn 06

We have observed both para-H2O(211ndash202) and ortho-H2O(321ndash312) lines at 162 GHz and 251 GHz respectively Thesource is clearly unresolved at 162 GHz but marginally resolvedat 251 GHz as displayed in Fig A1 The ratio between thepeak and the spatially integrated flux density of the continuum(S ν(ct)pkS ν(ct)) is 095 plusmn 003 and 060 plusmn 001 at 162 GHz and251 GHz respectively The H2O emission line peak to spatiallyintegrated flux densiity ratio (S pk

H2OS H2O) for J = 2 and J = 3are 10 plusmn 02 and 05 plusmn 02 respectively Therefore the spatialconcentrations of H2O and continuum image are in good agree-ment within the uncertainties

Both the continuum and the H2O lines are well detected inthis source The two H2O lines are well fitted by single Gaussianprofiles with similar linewidths (257 plusmn 27 and 234 plusmn 34 km sminus1Fig 2 and Table 4) The difference in linewidth (23 km sminus1) issmaller than the errors of the linewidth Therefore there is nosignificant difference between the spectra of the two transitionsThe ratio between IH2O(321ndash312) and IH2O(211ndash202) is 091 plusmn 012which is the lowest of our five detected sources in both linesHowever this ratio remains consistent with the observations oflocal galaxies (Y13 and Fig 4) by taking the uncertainty intoaccount

From our CO line observations we find a line FWHM of∆VCO = 224 plusmn 32 and 292 plusmn 86 km sminus1 for the CO(5ndash4) andCO(6ndash5) lines respectively which are within 1σ to the H2OFWHMs The observed ratio of IH2OICO for both the CO(5ndash4)and CO(6ndash5) lines is about 04 with less than 25 uncertainty

We have tentatively detected an H2O+ line in this source aswell (see Sect 5)

A2 G12v243 at z = 3127

The source is marginally resolved in the SMA 880 microm image(Fig 2 of B13) with a size sim15primeprime but there is no obviousstrongly lensed structures such as multiple images It is not yetpossible to build a lensing model for this source because thesearch for a deflector by B13 has been unsuccessful

Both para-H2O(202ndash111) and ortho-H2O(321ndash312) lines arewell detected as shown in Fig A1 and the source is unresolvedconsistent with the SMA image The ratios of S ν(ct)pkS ν(ct) for239 GHz and 282 GHz are 071 plusmn 002 and 087 plusmn 001 respec-tively while the S pk

H2OS H2O for H2O(202ndash111) and H2O(321ndash312)are 06 plusmn 02 and 10 plusmn 02 respectively

The two H2O lines are both well fitted by a single Gaussianprofile The FWHMs are 201plusmn 27 and 221plusmn 20 km sminus1 for para-H2O(202ndash111) and ortho-H2O(321ndash312) respectively The differ-ence is within the 1σ uncertainty

The velocity-integrated flux density ratio of high-lying overlow-lying H2O line of this source IH2O(321ndash312)IH2O(202ndash111) = 12plusmn02 which is slightly lower than that of Arp 220 as shown inFig 4 The linewidths of the H2O lines (201 plusmn 27 and 221 plusmn20 km sminus1) are the narrowest ones among our sources The val-ues are also very close to the CO(1ndash0) linewidth (210plusmn30 km sminus1Harris et al 2012) Their similarity indicates that there is notlikely any strong differential lensing effect between the CO andH2O emissions in this case

A3 NCv1143 at z = 3565

Having a redshift of z = 3565 from CO observation by CARMA(Riechers et al in prep) this source is one of the brightest (atsubmm) in our sample It is resolved by the SMA 880 microm beam(B13) with a size sim 2primeprime featured by two components at the north-east and southwest directions With a single deflector the lensingmodel estimates a magnification factor of micro = 69 plusmn 06

As displayed in Fig A1 both the lines and the continuumare very strong and well detected The ratio S ν(ct)pkS ν(ct) =086 plusmn 002 shows that the source is unresolved at 165 GHz(for observing para-H2O line 211ndash202) At 255 GHz the ratiosS ν(ct)pkS ν(ct) = 055 plusmn 001 and S pk

H2OS H2O = 07 plusmn 02 in-dicate that the source is partially resolved consistent with theSMA result

Both the H2O(202ndash111) and H2O(321ndash312) lines can be fittedby single Gaussian profiles The ratio of IH2O(321ndash312)IH2O(211ndash202)is 136 plusmn 013 close to the mean ratio found in the nearby star-forming-dominated galaxies (Y13 and Fig 4) The linewidthsof H2O(211ndash202) and H2O(321ndash312) are 293 plusmn 15 and 233 plusmn22 km sminus1 respectively Although the former is larger theyare compatible within an uncertainty of 16σ Also the H2Olinewidth agrees well with the CO(5ndash4) and CO(6ndash5) linewidths(273plusmn27 and 284plusmn27 km sminus1 see Table 3) Therefore the line ra-tios are unlikely to be affected by differential magnification Theobserved ratio of IH2OICO is 04ndash05 and 06ndash07 (uncertaintiesare within 13) for the J = 2 para-H2O and J = 3 ortho-H2Orespectively

We have also detected ortho-H2O+(211ndash202) and ortho-H2O+(202ndash111) fine structure lines together with para-H2O(211ndash202) in this source The further discussion of theH2O+ spectra and its interpretation are given in Sect 5

A4 NAv1195 at z = 2951

As quoted in B13 the redshift of this source was first obtainedby the CO observation (Harris et al in prep) Its SMA imageshows a typical lensed feature with two components separatedby sim2primeprime along the northwest and southeast direction The lensingmodel suggests a modest magnification factor micro = 41 plusmn 03

We have robust detections of H2O(202ndash111) and the contin-uum emission at 250 GHz and 293 GHz (Fig A1) However theH2O(321ndash312) line isat odds with the other five sources There-fore we only show the image of the dust continuum emission atthis frequency in Fig A1 The source is clearly resolved into twocomponents in the three images and the northwest componentis about 4 times stronger than the southeast one in the contin-uum images in agreement with the SMA image (B13) For thecontinuum at 250 GHz the peak to total integrated flux density

A80 page 18 of 23

162 GHz 251 GHz

239 GHz 282 GHz

164 GHz 255 GHz

Fig A1 Mapping of the H2O emission lines and the corresponding continuum emission (frequencies have been shown accordingly in thewhite text) in the sources with both para J = 2 and ortho J = 3 H2O lines observed The contours of the continuum emission start from6σ in step of 10σ and the contours of the H2O emission start from 3σ in step of 1σ Asymmetric negative contours are shown in whitedashed lines For each observation the 1σ contours for the continuum (mJy beamminus1) and the H2O emission line (Jy km sminus1 beamminus1) are as fol-lows G09v197 H2O(211ndash202) (017057) G09v197 H2O(321ndash312) (025038) G12v243 H2O(202ndash111) (029048) G12v243 H2O(321ndash312)(030053) NCv1143 H2O(211ndash202) (016036) and NCv1143 H2O(321ndash312) (042072)

ratio S ν(ct)pkS ν(ct) = 054 plusmn 002 and for H2O(202ndash111) theratio S pk

H2OS H2O equals to 06 plusmn 03 Therefore the spatial dis-tributions of dust and the H2O emission are likely to be similarin this source In the observation at 293 GHz S ν(ct)pkS ν(ct) =042 plusmn 001 due to a smaller synthesis beam (Table 1)

Figure 2 shows the spectra corresponding to the two ob-servations of NAv1195 The H2O(202ndash111) line can be fit-ted by a single Gaussian profile with a linewidth equal to328 plusmn 51 km sminus1 We have not detected the higher excita-tion H2O(321ndash312) line as mentioned above By assuming thesame linewidth as the lowerminusJ H2O line we can infer a2σ detection limit of 256 Jy km sminus1 This yields a flux ratioof H2O(321ndash312)H2O(202ndash111) 06 This value is significantlylower than that in the five other sources where it ranges from 075to 160 (errors are within 25) but it remains close to the low-est values measured in local galaxies (Y13) as shown in Table 6and Fig 4 This low ratio of H2O lines probably originates from

different excitation conditions especially for the far-infrared ra-diation field since the line H2O(321ndash312) is mainly populatedthrough far-infrared pumping via absorbing 75 microm photons (seeSect 5) The CO(5ndash4) line of the source has a linewidth of281 plusmn 16 km sminus1 which is comparable with the H2O line profileThe observed ratio of IH2OICO (CO(5ndash4)) is le04

A5 NAv1177 at z = 2778

NOEMA observation of the CO line in this source gives a red-shift of z = 2778 (Krips et al in prep) The SMA 880 microm imageshows a compact structure with two peaks sim1primeprime away along theeastwest direction and the western component is the dominantone (Fig 2 of B13) However due to the absence of deflector inthe foreground optical image from SDSS and lack of the deepoptical and near-infrared images the lensing properties are stillunknown (B13)

A80 page 19 of 23

AampA 595 A80 (2016)

250 GHz 293 GHz

261 GHz 308 GHz

240 GHz 283 GHz

Fig A2 (See Fig A1 caption) For each observation the 1σ contour for the continuum (mJy beamminus1) and the H2O emission line(Jy km sminus1 beamminus1) are as follows NCv1195 H2O(202ndash111) (034051) NCv1195 H2O(321ndash312) (048ndash) NAv1177 H2O(202ndash111) (058065)NAv1177 H2O(321ndash312) (038058) NBv178 H2O(202ndash111) (028030) NBv178 H2O(321ndash312) (021029)

As displayed in Fig A1 both the lines of H2O(202ndash111)and H2O(321ndash312) and the corresponding dust continuum arewell detected The ratios S ν(ct)pkS ν(ct) are 075 plusmn 002 and062 plusmn 001 for observation at 261 GHz and 308 GHz respec-tively One should notice that the direction of the synthesisedbeam is perpendicular to the alignment of the two componentsin the image thus the source is marginally resolved in theH2O(202ndash111) and the corresponding dust continuum imagesFor the H2O(321ndash312) observation at higher frequency we cansee the partially resolved feature The peak to total flux ra-tios of H2O are S pk

H2OS H2O sim 08 plusmn 03 and 06 plusmn 01 for theH2O(202ndash111) and H2O(321ndash312) lines respectively indicatingsimilar spatial distribution compared with the dust emission

The H2O spectra displayed in Fig 2 show single Gaussianprofiles with FWHM = 241 plusmn 41 and 272 plusmn 24 km sminus1 (Table 4)The profiles of the two H2O lines are similar within the uncer-tainties The line ratio IH2O(321ndash312)IH2O(202ndash111) = 134plusmn024 Thisvalue is close to that found in Arp 220 and it is the largest ratioin our sample We have also detected the CO(3ndash2) and CO(5ndash4)lines using the IRAM 30 m telescope in this source (Table 3)

the linewidth of CO is around 230 plusmn 16 km sminus1 which is similarto the width of the detected H2O lines The ratio of IH2OICO isfrom 05 to 11 with less than 20 uncertainties

A6 NBv178 at z = 3111

The CO redshift of NBv178 was obtained by Riecherset al (in prep) via CARMA z = 3111 and the data of theH2O(202ndash111) line were obtained by O13 The source is resolvedin the SMA 880 microm image (B13) with a somewhat complex mor-phology and the size is sim25primeprime There are three main componentsin the image The two strong components located at northwestand southeast direction of the image and the weakest compo-nent close to the southeast The derived lensing magnificationis micro = 135 plusmn 15 which is the second largest among our sam-ple In the near-infrared images the source has a similar three-component Einstein ring-like structure with a slightly smallermagnification micro = 108+03

minus02 (Calanog et al 2014)Our NOEMA images of both continuums and H2O lines as

shown in Fig A1 are consistent with the SMA 880 microm image

A80 page 20 of 23

243 GHz 355 GHz

161 GHz 228 GHz

221 GHz

Fig A3 (See Fig A1 caption) For each observation the 1σ contour for the continuum (mJy beamminus1) and the H2O emission line(Jy km sminus1 beamminus1) are as follows G09v1124 H2O(211ndash202) (017ndash) G09v140 H2O(211ndash202) (019032) SDP11 H2O(202ndash111) (130104)NCv1268 H2O(211ndash202) (013039) and NAv156 H2O(211ndash202) (053102)

The images are resolved into two main parts while the southerncomponent is extended along the western side The continuumand H2O line images have fairly high SN From the observa-tion of H2O(202ndash111) at 241 GHz (O13 note that this observationwas performed at higher resolution with a 14primeprime times 10primeprime beam) thevalues of S ν(ct)pkS ν(ct) and S pk

H2OS H2O agree well which are042 plusmn 001 and 04 plusmn 01 respectively The continuum image at283 GHz gives S ν(ct)pkS ν(ct) = 069 plusmn 001 and the image ofH2O(321ndash312) gives S pk

H2OS H2O = 08 plusmn 01 The similarity ofthe peak to spatially integrated flux density ratios suggest thatthe spatial distribution of H2O and submm dust continuum arelikely to be consistent Additionally the images of J = 2 andJ = 3 images are also consistent within the uncertainty

NBv178 has a very broad linewidth compared with theother sources As shown in Fig 2 the linewidth of H2O(202ndash111)and H2O(321ndash312) are 510 plusmn 90 and 607 plusmn 43 km sminus1 respec-tively The two lines have similar profiles The source has aIH2O(321ndash312)IH2O(202ndash111) ratio equal to 1 within the range of thelocal galaxies (Fig 4) The CO(5ndash4) and CO(6ndash5) observations(Table 3) give linewidths of 614plusmn53 and 734plusmn85 km sminus1 which

are wider than the H2O lines The observed ratio of IH2OICO isabout 07 with less than 25 uncertainty for J = 2 H2O

A7 G09v140 at z = 2093

A CO redshift of G09v140 z = 20894 was obtained byCSOZ-Spec (Lupu et al in prep) But through our H2O ob-servation we find a redshift of z = 2093 Our value is consis-tent with the CO(3ndash2) observation by Riechers et al (in prep)and we have adopted this value SMA observation of the 880 micromdust continuum shows two close components with a separationof sim1primeprime along the east and west direction The lensing model es-timates micro = 153 plusmn 35 which is the largest magnification inour sample The Keck near-infrared image of the source sug-gests a magnification of micro = 114+09

minus10 for the stellar component(Calanog et al 2014)

The H2O(211ndash202) line is well detected as well as the cor-responding dust continuum As shown by the images of theH2O and dust continuum (Fig A1) the source is partially re-solved by the synthesised beam The two component (east and

A80 page 21 of 23

AampA 595 A80 (2016)

west) structure is consistent with the 880 microm image and thewestern component is stronger than the eastern one Both ratiosof S ν(ct)pkS ν(ct) and S pk

H2OS H2O are found to be 06 (uncer-tainty lt13) However the eastern component is not detectedin the H2O image

The H2O(211ndash202) line can be fitted with a single Gaussianprofile with a FWHM of 277 plusmn 14 km sminus1 (Fig 2) Howeverthe line has a steeper decline on the red side of the spectrumThe CO(4ndash3) observation gives a linewidth of 198 plusmn 51 km sminus1which is 07 plusmn 02 times narrower than that of the H2O line Thevelocity-integrated flux density of the H2O is larger than that ofthe CO(4ndash3) in this source with a ratio of IH2OICO = 11 plusmn 03

A8 SDP11 at z = 1786

The CO observation by Lupu et al (2012) found z = 1786 TheSMA 880 microm image displays a typical strongly lensed morphol-ogy with two components north and south respectively Thesize of the source is sim2primeprime The gravitational magnification esti-mated by B13 is micro = 109 plusmn 13

As shown in Fig A1 the source is partially resolved Thedust continuum image shows an extended structure along thenorth and south direction with the brightest peak in the northernpart The noisy images are consistent with the SMA 880 microm ob-servation The ratio of S ν(ct)pkS ν(ct) is 056 plusmn 003 The imageshows two distinctive components in the north and south direc-tion This structure also agrees with the high resolution SMA880 microm image Additionally S pk

H2OS H2O = 04 plusmn 02 suggeststhat the H2O image is slightly resolved compared with the dustemission This diffierence may come from their different spatialdistribution

Although the noise level of its spectrum is the highest amongour sources because of the high frequency namely 355 GHz theH2O(202ndash111) line is marginally detected with SN = 46 Thelinewidth is 214 plusmn 41 km sminus1 (Fig 2)

A9 NCv1268 at z = 3675

The redshift of NCv1268 was obtained by the CO observationof Krips et al (in prep) A typical strongly lensed morphologywas found by the SMA 880 microm observation (B13) with a strongarc-like component in the south direction The structure has asize sim25primeprime The B13 lensing model estimates micro = 130 plusmn 15

The source is marginally resolved by the NOEMA synthesisbeam (Fig A1) When comparing the flux ratios between thedust and H2O emission from the peak and the spatially integratedvalues they give S ν(ct)pkS ν(ct) = 066plusmn001 and S pk

H2OS H2O =

06 plusmn 01 The values of dust emission and the H2O image are ingood agreement

NCv1268 is the only source in which we have detecteda double-peaked line profile from our new observations witha slightly stronger blue velocity component (Fig 2) The totallinewidth is very large 731 plusmn 75 km sminus1

A10 NAv156 at z = 2301

Harris et al (2012) give a CO redshift of z = 23010 for thissource The SMA 880 microm dust continuum image shows a classicstrongly lensed morphology with multiple images It consists ofan arc-like component in the western direction and a point-likecomponent toward the east They are separated sim2primeprime The lensmodel implies that the magnification factor micro = 117plusmn09 (B13)

As shown in Fig A1 the source is marginally resolvedwith an extended morphology in the eastern part The structuresdisplayed in the dust and H2O images are similar The ratiosS ν(ct)pkS ν(ct) = 062 plusmn 003 and S pk

H2OS H2O = 07 plusmn 02 alsosuggest their similarity within the errors Most of the fluxes areconcentrated in the western part which agrees with the SMA880 microm image

The H2O(211ndash202) line can be fitted by a single Gaussianwith a large linewidth equal to 593plusmn56 km sminus1 The CO(4ndash3) lineobservation by NOEMA (Oteo et al in prep) gives a linewidthof 621 plusmn 47 km sminus1 The linewidths of CO and H2O are in verygood agreement Our noisy detection of CO(5ndash4) at IRAM 30 m(Table 3) gives a ratio of IH2OICO = 08 plusmn 03

A11 G09v1124 at z = 2410

The redshift of this source is measured by CO observation(Harris et al 2012) This multiple source with two main com-ponents each with intrinsic LIR gt 1013 L separated by 10primeprime(Fig A1) was studied in detail by (Ivison et al 2013 see alsoB13 and Oteo et al in prep) It is special in our sample since thetwo main sources are from two very different HyLIRGs ratherthan multiple images of a single source generated by strong grav-itational lensing The eastern component G09v1124-W whichcontains a powerful AGN (Oteo et al in prep) is unlensed andthe western component G09v1124-T is only weakly lensed witha magnification factor micro = 15plusmn02 Thus throughout the discus-sions we treat G09v1124-W and G09v1124-T as two distinctsources (see Tables 2 4 5 and 6)

Probably because of this too small lensing magnification andthe smaller values of each microLIR (Table 2) we have only detectedthe dust continuum emission in this source The H2O(211ndash202)line is undetected The 2σ upper limits of the velocity-integratedflux density of the H2O(211ndash202) line show that the values ofIH2O are more than three times smaller than in the other sourcesAs seen in Table 4 and Fig 3 the ratio LH2OLIR is smaller thanall our other sources for G09v1124-W probably because of itsstrong AGN However for G09v1124-T this ratio albeit smallmight be comparable to that of G09v197 The dust continuum at221 GHz follows the same structure as the previously publishedobservations (Fig A1 and Ivison et al 2013) Both the easterncomponent (W) and the western one (T) are marginally resolvedby the synthesised beam The peak to total continuum flux ratiosare S ν(ct)pkS ν(ct) = 084 plusmn 003 and 083 plusmn 004 respectively

Appendix B H2O+ detections in local ULIRGs

The study using the Herschel SPIREFTS spectra of 167 lo-cal galaxies has revealed several emission and absorptionlines of H2O+ which are ortho-H2O+ lines 211ndash202 (52minus52)202ndash111 (52minus32) 111ndash000 (32minus12) 111ndash000 (12minus12) (Y13 see alsoTable B1) All J ge 2 H2O+ lines are in emission Table B1gives values of the H2O+ flux and luminosity for those amongthe Y13 sample where H2O+ lines are (tentatively) detectedwith SN amp 25 However for the H2O+(111ndash000) lines whichconnect the ground state they are often found to be in emis-sion in AGN-dominated sources while they are in absorption instar-forming-dominated ones A full description of the datasetfor this Herschel SPIREFTS survey will be given in Liu et alin prep At high-redshift prior to our study the J = 2 ortho-H2O+ doublet lines seem to have only been tentatively detectedin two sources SPT0346-52 (Weiszlig et al 2013) and HFLS3(Riechers et al 2013)

A80 page 22 of 23

Table B1 Beam-matched H2O+ H2O line and infrared luminositiesfrom local detections (Herschel SPIREFTS archive) and high-redshiftHerschel lensed galaxies

Source 202ndash111 (52minus32) 211ndash202 (52minus52) LIRminusbeam LH2O106 L 106 L 1011 L 106 L

Local ULIRGsESO320-G030 04 plusmn 01 ndash 15 09CGCG049-057 03 plusmn 01 07 plusmn 02 19 16NGC 2623 05 plusmn 02 07 plusmn 03 35 20Arp299-A 08 plusmn 02 07 plusmn 03 54 13Arp220 37 plusmn 08 35 plusmn 09 162 152IRAS13120-5453 36 plusmn 12 24 plusmn 10 282 107IRASF17207-0014 68 plusmn 22 63 plusmn 22 246 155Mrk231 26 plusmn 10 30 plusmn 15 324 93

High-redshift lensed galaxiesNCv1143 287 plusmn 72 286 plusmn 72 1136 897G09v197 358 plusmn 90 ndash 2114 1066G15v2779 ndash 558 plusmn 159 2700 2660HFLS3 ndash 2618 plusmn 2003 3640 7378

Notes Luminosity of H2O+ fine structure lines 202ndash111 (52minus32)(7421 GHz) 211ndash202 (52minus52) (7465 GHz) and H2O(211ndash202) As for thelocal sources the Herschel SPIREFTS beam is smaller than the entiresource for some galaxies here LIRminusbeam represents the LIR from the samebeam area as the measured emission line (a full description will be givenby Liu et al in prep) LH2O is the luminosity of H2O(211ndash202) HFLS3data are taken from Riechers et al (2013)

A80 page 23 of 23

Introduction

Sample and observation

Results

General properties of the H2O emissions

Lensing properties

Discussion

LH2OndashLIR correlation and LH2OndashLIR ratio

H2O excitation

Comparison between H2O and CO

AGN content

Detection of H2O+ emission lines

Conclusions

References

Individual sources

G09v197 at z=3634

G12v243 at z=3127

NCv1143 at z=3565

NAv1195 at z=2951

NAv1177 at z=2778

NBv178 at z=3111

G09v140 at z=2093

SDP11 at z=1786

NCv1268 at z=3675

NAv156 at z=2301

G09v1124 at z=2410

H2O+ detections in local ULIRGs