The Optical Identification of a Primeval Galaxy at z >~ 4.4

7 Pages·0.23 MB·English

by  Adriano Fontana

Mon.Not.R.Astron.Soc.000,000–000 (0000) Printed1February2008
⋆
The optical identification of a primeval galaxy at z ∼> 4.4
Adriano Fontana1, Stefano Cristiani2,3, Sandro D’Odorico3, Emanuele Giallongo1
and Sandra Savaglio3,4
1 Osservatorio Astronomico di Roma, 00040 Monte Porzio, Italy
6 2 Dipartimento di Astronomia dell’ Universit`a, 35122 Padova, Italy
9 3 European Southern Observatory, 85748 Garching bei Mu¨nchen, Germany
9 4 Istituto di Astrofisica Spaziale del CNR, 00044 Frascati, Italy
1

n
a
J
1February2008

6
1
ABSTRACT

1
v
WehaveobtainedwiththeSUSICCDcameraontheESO3.5mNTTdeepimagesinthe
6
BVRIbandsofthefieldcenteredontheQSOBRI1202-0725(zem =4.694).Inthefinal
8
combinedframesthe stellarimageshaveFWHMof1,1,0.6and0.65arcsecrespectively.
0 The R and I images show clearly a galaxy 2.2′′ from the QSO, corresponding to 13h−1
1 50
0 kpc at z ∼ 4.5. Possible identification with the metal absorption systems seen in the
6 line of sight to the QSO, including the highest redshift damped system known to date
9 atz =4.383,arediscussed.Weconcludethatitscolourscanbereconciledonlywiththe
/ spectrum of a primeval galaxy at z ∼> 4.4, making it the most distant galaxy detected
h
so far. From its magnitudes and models of young galaxy evolution we deduce that it is
p
- forming stars at a rate ∼ 30M⊙ yr−1 and has an estimated age of the order of 108 yr
o or less, implying that the bulk of the stellar population formed at z <6.
r
t
s
a
:
v
i 1 INTRODUCTION been equally unsuccessful at large redshifts (Djorgovski et
X
al. 1993, Pahre & Djorgovski 1995).
Thedetectionofprimevalgalaxies(PG’s),i.e.high-redshift,
r An alternative searching technique is based on ultra-
a younggalaxiesundergoingtheirfirstepisodesofstarforma-
deep imaging in different colours in the fields of high red-
tion, is essential to provide the observational underpinning
shift QSOswhich show Lyman Limit and Damped Lyman-
to modern theories of galaxy formation and evolution. A
α Absorption Systems (DLAS) in their absorption spectra.
stronghintontheirexistenceuptoredshifts≈4isprovided
This method has been successful in a few cases (Steidel &
by the observation of metal enriched absorption systems in
Hamilton 1992, Steidel et al. 1995, Giavalisco et al. 1994),
the spectra of high-redshift QSOs, but their actual detec-
detecting galaxies at redshift z=3−3.5.
tionatz≥1hasprovedtobeadifficulttask.Searchingfor
We have applied this approach to study the field of
the optical counterparts of powerful radio-sources has led
the 18.7 R magnitude,z = 4.694 QSO BRI 1202-0725. The
to the discovery of a number of high redshift galaxies (Mc-
quasar was found during the APM optical survey for QSO
Carthy1993;Lacyetal.1994),buttheseobjectsarehardly
with z > 4 (Irwin, McMahon & Hazard1991) and its coor-
informative on the general population of PGs because the
dinates are given in McMahon et al(1994). Giallongo et al.
continuumiscontaminatedbynon-stellarprocessesandthe (1994)obtainedaspectrumofthisQSOat≈40kms−1 and
emissionfluxesandrelativeintensitiesarenotthoseofstan-
detectedaDLASatz≈4.38,thehighestredshiftsystemof
dard star formation regions. Deep spectroscopic surveys at
thistypeknowntodate.Luetal.(1995)andWampleretal
opticalwavelengths(e.g.Glazebrooketal.1995a,b,Cramp-
.(1995) have studied the metal absorption lines associated
ton et al. 1995), have led to the identification of galaxies
to this DLAS and both derived metal abundances (for Fe
typically at redshift up to z ∼ 1. Field surveys based on and O respectively) which are of the order of 10−2 of the
the search of highly redshifted Lyman-α line emission or
solar value.
other emission lines as a signature of star formation have
2 OBSERVATIONS AND ANALYSIS
⋆ BasedonobservationscollectedattheEuropeanSouthernOb- The field of BRI 1202-07 was observed during three photo-
servatory,LaSilla,Chile metric nights, 23–26 Apr 1995, with the SUSI direct CCD
2 A. Fontana et al.
cameraattheNasmythfocusoftheESONTT.Fourbroad– optical bands. The characteristic spectral features of such
bandfilterswereusedcorrespondingtotheBVI passbands an object are, in fact, the Lyman continuum break at 912
of theJohnson-Kron-Cousins system (JKC) and to ther of ˚A (rest–frame) and a relatively flat spectrum longward of
theThuanandGunnsystem.Inadditionanarrowband(67 it (Bruzual & Charlot 1993). Additional attenuation of the
˚A)filtercenteredatλ=6560˚A,whichincludestheLyman- emitted flux shortward of 1215 ˚A (rest–frame) occurs be-
αemissionatthez =4.38DLAsystem,hasbeenused.Sev- cause of the absorption by the population of intergalactic
eral dithered images were obtained in thefour broad bands Lyman-α clouds. An interesting comparison can be made
and in the narrow filter. The total integration times are with a spectrum of a PG with constant star formation rate
15000,10800,7200,7800 and 7800 seconds for the four broad (SFR)andanageof0.1Gyrkindlycomputedandredshifted
bands and the narrow filter observations respectively. The at 4.4 by Madau according to his standard model (Madau
FWHM of the stellar images in the combined frames are 1995).Convolvingthissyntheticspectrum(Fig.2a)withour
1,1,0.6, 0.65 and 1 arcsec. instrumental passbands and normalizing the galaxy flux to
The photometric calibration was obtained from several I′ = 24.1, we derive the following magnitudes: R′ = 24.5,
standard stars observed on the same nights and at simi- V′ =25.9, and B′ =29.8, consistent with the observations.
lar airmasses. Total magnitudes have been obtained in an This agreement implies a low Lyman-α emission, which is
aperture of fixed size (1.8′′), applying a seeing–dependent consistent with the upper limit on the Lyman-α equivalent
aperture correction. width <150 ˚A derived from thenarrow band frame.
In view of the spectral peculiarities expected for PG’s, At the same time, the colours are not compatible with
we have calibrated B′V′R′I′ magnitudes in the “natural” those of a galaxy at significantly lower redshift. In the case
system definedby our instrumentalpassbands. of a galaxy that is actively forming stars, the UV radiation
The zero-points of our instrumental system have been emittedat wavelengthslongward ofitsLymanbreakmoves
adjusted to give the same BVRI magnitudes of the JKC inthebluerbandsastheredshiftdecreases,resultinginV′−
systemforstellarobjectswithB−V =V −R=R−I =0. R′andB′−R′coloursmuchflatterthanthoseobserved.For
Thecolourtransformationsforstellarobjectsturnouttobe instancea0.7Gyrold,star–forminggalaxywouldinvariably
I′=I,B′−V′ =0.91×(B−V),V′−R′=1.17×(V −R), have (De Robertis and McCall 1995) V − R < 0.5, and
R′−I′=0.74×(R−I).Themagnitudeshavebeencorrected B−V <0.5 in the redshift range 1<z <3. In the case of
for galactic absorption which at the position of the QSO is anoldstellar populationat intermediateredshifts,ofwhich
E(B-V)=0.03. thelarge V′−R′ colour mightbeanindication, population
Figure 1 shows the region surrounding BRI 1202-0725 synthesis models (De Robertis & McCall 1995, Bressan et
intheB′V′R′ bands.Afaint,nonsymmetricgalaxy clearly al. 1994) predict a much larger R−I colour than observed
stands out in the R′ and I′ frames NW of the QSO at a in our PG candidate. According to the models, in fact, in
P.A.=328◦ and a distance of 2.2′′, corresponding † to 18– the redshift interval 1–3 the V −R colour varies between
13 kpc in the redshift range z = 1−4.7. Its magnitudes, 2 and 2.5, but the R−I colour is larger than 1.5. These
measured after the QSO subtraction, are: B′ ≥ 27.5, V′ = arguments apply to the metal systems detected at z < 3,
26.2±0.4, R′=24.3±0.1, I′ =24.1±0.2. The lower limit whichthencannotbeassociatedwiththedetectedgalaxy.In
onB′ isa1σ limit.Fromthenarrowbandframe(7800s)we particular, at z=1.7 an active (passive) galaxy would result
set a 1σ flux upperlimit fNB <6.7×10−17 erg s−1 cm−2. (DeRobertis&McCall1995,Bressanetal.1994)inV−R=
This galaxy has also been independently detected in 0 (2.1) and R−I =0.3 (2.1).
deep K band exposures at the Keck telescope (Djorgovski Wefinally notethat thefaint B′ andV′ magnitudesof
1995),withanestimatedKmagnitudeof23±0.5(Djorgov- thisgalaxycannotbeinterpretedasduetoastrongredden-
sky,privatecommunication). inginalowredshiftobject. TheobservedR′−I′=0.2and
I′−K ≈ 1 colours would imply, in presence of a substan-
tial amount of dust, an intrinsic spectrum with abnormally
negative R′−I′ and I′−K colours.
3 REDSHIFT IDENTIFICATION
The observed colours can also be used to discriminate
In discussing the possible redshift of the galaxy we have betweengalaxies atredshiftsaround4.5.Attheseredshifts,
detected,wemusttakeintoaccountthatthelineofsightto evenasmalldifferenceinredshiftsproducessignificantvari-
theQSOislikelytobepopulatedbymanygroupsofgalaxies ations of the observed colours (see fig. 2b), due to changes
at different redshifts. This can be inferred by the presence of the Lyman forest absorption in the various passbands.
of several metal absorption systems. Beside the DLAS at At redshifts lower than 4.4 the V′ flux would considerably
z = 4.383 shown by Giallongo et al (1994), Wampler et al. increase, while at higher redshifts, the flux in the R′ band
(1995) discuss two systems at z =4.672 and z =4.687 and would be attenuated by the Lyman forest. The same PG
list 6 other metal systems with redshift between 1.75 and thatat z ∼4.4is consistent with theobservations,at z=4
4.48. (withthesamefluxnormalization intheIband)wouldstill
The peculiar colours of the detected galaxy, B′−R′ > have R′ = 24.3, but V′ = 25.3, i.e. 0.9 mag. brighter than
3.1, V′−R′ =1.9, R′−I′ =0.2, R′−K =1.3, agree with observed.Atz=4.7(withthesameintergalacticabsorption
thoseexpectedforaz≈4.4starforminggalaxy,whoseUV asinBRI1202-0725),itwouldhaveV′=26.5butR′=25.1,
radiation from young massive stars is redshifted into the i.e.0.8mag.fainterthanobservedintheRband.Neverthe-
less, a Lyman-α emission might increase the R′ flux and
reconcile the predictions with the observations. At z =4.7,
† Ω=1andHo=50kms−1Mpc−1areadoptedthroughoutthe the required strength of such a line, causing a ∆R′ =−0.8
paper mag., is fLyα ∼ 2×10−16 erg s−1 cm−2, still lower by an
>
A Primeval galaxy at z ∼ 4.4 3
Figure1. Upperleft:a1hGunn-rcombinedimageofthefieldoftheQSOBRI1202-0725.Thiscombinationofthesingleframesofbest
image quality has an image quality of 0.5′′ FWHM. Lower left and lower right: combined images in the B′ (15000s), and V′ (10800s)
bands,respectively,bothwith1′′ FWHM.Upperright:a2h Gunn-rcombinedimage(FWHM0.65′′)inwhichascaledstellarPSFhas
beensubtracted tothe QSOandtothestarinthe upperrightcorner.Thegalaxyangularsizeis1.7”. Thefieldis22′′×22′′,Northis
atthetop,Easttotheleft.
4 A. Fontana et al.
Figure 2. a) Comparisonbetween a model spectrum of a primeval galaxies at z =4.38 and the observed fluxes.
The dotted lineshows theintrinsicspectrum, the solidline the spectrum withthe intergalactic Lymanabsorption
included.Theupperhorizontalaxisshowstherestframewavelengths.Theobservedfluxisshownontheleftvertical
axisandtherest–frameluminosityontherightone.Theconversionfromouropticalmagnitudesintomonochromatic
fluxeshasbeencalibratedonstandardstarswithnocorrectionforthepeculiarspectrumofthisgalaxy.Themodel
spectrum has been normalized to the observed I flux. Fluxes have been plotted at the peak of filter transmission,
andtheerrorbarsinthexdirectionshowtheFWHMofthefilters.b)Sameasa,butforaprimevalgalaxyatz=4
(solidline)andoneatz=4.7(dotted line),bothwiththeintergalacticLymanabsorptionincluded.
order of magnitude than the upper limit inferred from the s−1 ˚A−1,correspondstoaSFR≈30M⊙ yr−1 inbothcases.
negative detection of the [O II] 3727 ˚A line in the K band The rest-frame blue magnitude is MB ∼ −21, of the order
(Pahre&Djorgovsky1995).Redshiftsmuchhigherthan4.7 of the characteristic magnitude M∗ of the local luminosity
B
can be excluded, since the Lyman-α emission would shift function of bluegalaxies.
out of theR′ filter. From the I−K ∼1.3 colour we deduce an age ∼> 107
We conclude that we have detected a galaxy that is and≈108 yrforthesingleburstandtheconstantSFRcase
massively forming stars at z = 4.4−4.7: our deduction is
respectively. In both cases the converted mass in stars is
based on what it ispresently known orreasonably assumed Mstar ∼1−3×109M⊙.
on galaxy evolution and only a spectroscopic confirmation
willprovidetheultimateevidencethatourargumentiscor- If the galaxy is associated with one of the absorption
rect. systems, interesting estimates of the total gaseous mass
Mgas could be obtained. For instance, if it is associated
with the DLA, where almost all of the hydrogen is in the
4 DISCUSSION neutral form, Mgas can be estimated from the observed HI
column density NH =5×1020 cm−2 by assuming a radius
Thephotometricpropertiesof thisgalaxy canbecompared in the range 13−30 kpc, where the highest value is the
withstandardspectralevolutionmodels(Bruzual&Charlot upper limit in the impact parameter derived from known
1993, Charlot & Fall 1993) to infer its evolutionary status: DLA (Steidel 1995). The resulting mass is in the range
the UV luminosity is related to the instantaneous SFR oc- Mgas=2−9×109M⊙respectively,implyingatotalbaryonic
curringinthegalaxy,whiletheoverallshapeofthespectrum massnotsubstantially higherthan1010M⊙ andagas/stars
between the rest–frame UV emission and the 4000 ˚A break massratioof2orlarger.Theefficiencyofconversionofcold
is a function of the age of thegalaxy. gasintostarsisanimportantfreeparameterforgalaxyfor-
We have considered two extreme models for the SFR, mationmodels:attheconstantrateof30M⊙ yr−1,nomore
namelyasingleburstwithanexponentialdeclineonatime than20% ofthegaseousmassisconvertedintostarswithin
scale τ = 107 yr and a constant SFR. The conclusions do adynamicaltime in agreement with modelsthat reproduce
not change within the redshift range 4.4–4.7. The UV lu- the observed properties of local galaxies (Kauffman, White
minosity, as observed in the I–band, L1550 ≃ 2×1041 erg & Guiderdoni 1993).
>
A Primeval galaxy at z ∼ 4.4 5
Thisgalaxymightotherwisebenotassociatedwithany DeRobertisM.M.,andMcCallM.L.,1995,AJ,109,1497
identified absorption systems, and in particular it might be DjorgovskiS.G.,1995,inJ.R.WalshandI.J.Danzigereds,Science
in the environment of the quasar itself, which might be a withtheVLT,Springer-Verlag,Berlin,p.351
preferred site for galaxy formation. Indeed, extended line– Djorgovski S., Spinrad, H., McCarthy, P., and Strauss,M. A.,
1985,ApJL,299,L1
emitting companions of high redshift QSOs have been de-
Djorgovski S., Thompson D., Smith J.D., 1993, in B.Rocca-
tected in a number of cases (Djorgovski et al. 1985; Hu et
Volmerange, M.Dennefeld, B.Guiderdoni and Tran Thanh
al. 1991). Inmost cases, theyemit nocontinuum radiation,
Van eds, First Light in the Universe: Stars or QSOs?, Fron-
whichhasbeeninterpretedasanindicationofanongoingin-
tieres,p.67
teractionwiththequasaritself.Sincealltheseobjectsareas- GiallongoE.,D’OdoricoS.,FontanaA.,McMahonR.G.,Savaglio
sociatedwithradio–loudquasars,apossibleconnectionwith S., Cristiani S., Molaro P. and Trevese D., 1994, ApJL 425,
the same phenomena giving rise to the radio emission has L1
been suggested. As shown in the previous section, a strong GiavaliscoM.,MacchettoF.D.,SparksW.S.,1994,AA,288,103
Lymanα emission from this galaxy might reconcile its col- Glazebrook K., Ellis R.,Colless M.,Broadhurst T., Allington-
ors with those expected for a z =4.7 galaxy: however, it is SmithJ.,TanvirN.R.,TaylorK.,1995a, MNRAS,273,157
GlazebrookK.,Peacock J.A.,MillerL.andCollinsC.A.,1995b,
difficulttoascribetheoverallphotometricpropertiesofthis
MNRAS,275,169
galaxyastheresult ofreprocessing oftheQSOradiation or
Hu,E.M.,Songaila,A.Cowie,L.I.,andStockton,A.,1991,ApJ,
interactionwithit.Indeed,thecleardetectionintheIband,
368,28
wherenostrongemissionlineispresentatz=4.7,provides
Irwin M.J., McMahon R.G., Hazard C., 1991, in: Crampton
thestrongestevidencethatthisobjectshasatrulyUVcon- D. (ed) ASP Conf.Ser. Vol.21, The Space Distribution of
tinuum, possibly associated with an intense star–formation Quasars,Astron.Soc.Pac.,SanFrancisco,p.7
activity, which dominates any QSO reprocessed radiation. KauffmanG.,White S.D.M.and Guiderdoni B.,1993, MNRAS,
The strong Lymanα needed in this case is still compatible 264,201
with that expectedfrom a galaxy forming stars at a rate of LacyM.,MileyG.,RawlingsS.,SaundersR.,DickinsonM.,Gar-
30 M⊙ yr−1 (Charlot & Fall, 1993), though the large un- ringtonS.,MaddoxS.,PooleyG.,SteidelC.C.,BremerM.N.,
Cotter G., van Ojik R., Roettering H. and Warner P., 1994,
certainities in the models prevent an accurate comparison.
MNRAS,271,504
Furthermore, the absence of radio emission from this QSO
LuL.,SargentW.L.W.,WombleD.S.andBarlowT.,1995,ApJL
(McMahon et al 1994) argues against theexistence ofhigh
inthepress
energyphenomenaonseveralkpcscale, astypicalinradio–
MadauP.,1995,ApJ,441,18
loud objects. Our galaxy would therefore be similar to the McCarthyP.J.,1993,PASP105,1051
companion of the radio–quiet z =2.75 quasar Q1548+0917 McMahon, R.G., Omont, A., Gergeron, J., Kreysa, E., Haslam,
(Steidel,Sargent&Dickinson1991). InthiscaseaLymanα C.G.T.,1994,MNRAS,267,L9
emittinggalaxyataredshiftonly1000kms−1 greaterthan PahreM.A.,DjorgovskiS.G.,1995,ApJL,449,L1
thatof theQSOwas detected 5”apart from it.Also inthis Steidel C.C., 1995, in G.Meylan eds, QSO Absorption Lines,
object, indeed, the detection of the UV continuum and the Springer-Verlag,Berlin,p.139
SteidelC.C.,HamiltonD.,1992,AJ,104,941
absenceofhigh–ionizationemissionlinesfromthegalaxyar-
SteidelC.C.,SargentW.L.W.,DickinsonM.,1991,AJ,101,1187
gued against any direct connection with theQSO activity.
SteidelC.C.,Pettini M.,HamiltonD.,1995,ApJ,inpress
Whatsoever the redshift of this galaxy is, we conclude
WamplerE.J.,WilligerG.M.,BaldwinJ.A.,CarswellR.F.,Haz-
that its photometric properties are probably due to star–
ardC.andMcMahonR.G.,1995, A&Ainpress
formation activity, and thus may represent an example of
a high redshift galaxy undergoing its initial burst of star
formation. Anupperlimit oftheorderof 108 yrfortheage
ofthegalaxy impliesthatthebulkofthestellar population
formed at z < 6. This limit is not far from that of the
furthest QSOs known, and could be representative of the
epoch of thefirst galaxy formation.
Acknowledgments
WethankC.Hazard,M.J.IrwinandR.G.McMahonforpro-
viding the QSO coordinates in advance of publications, G.
Djorgovski forcommunicating hisestimate oftheKmagni-
tude of the galaxy, A. Bressan for providing the colours of
an elliptical galaxy as a function of redshift, and M. Vietri
forusefuldiscussions.WeareindebtedtoP.Madauforcom-
puting the redshifted spectrum of the star forming galaxy.
EG acknowledges partial financial support from ASI.
REFERENCES
BressanA.,ChiosiF.,andFagotto F.,ApJ.Suppl.,1994,94, 63
BruzualG.A.,CharlotS.,1993, ApJ,405,538
CharlotS.andFallS.M.,1993,ApJ,415,580
Crampton D., LeFevre O., Lilly S.J., Hammer F., 1995 ApJ in
thepress
2 A.Fontana et al.: The Optical Identi(cid:12)cation of a Primeval Galaxy
0 0
Fig.2 : Combined Gunn-r, B and V images of the
(cid:12)eld of the QSO BRI1202-0725.