On The Detection Of Ionizing Radiation Arising From Star-Forming Galaxies At Redshift z ~ 3-4 : Looking For Analogs Of "Stellar Reionizers"

1.3 MB·English

by  Eros Vanzella

Draftversion January30,2012
PreprinttypesetusingLATEXstyleemulateapjv.12/16/11
ON THE DETECTION OF IONIZING RADIATION ARISING FROM STAR–FORMING GALAXIES AT
REDSHIFT Z ∼3–4 : LOOKING FOR ANALOGS OF “STELLAR REIONIZERS”
Eros Vanzella1, Yicheng Guo2, Mauro Giavalisco2, Andrea Grazian3, Marco Castellano3, Stefano Cristiani1,
Mark Dickinson4, Adriano Fontana3, Mario Nonino1, Emanuele Giallongo3, Laura Pentericci3, Audrey
Galametz3, S. M. Faber5, Henry C. Ferguson6, Norman A. Grogin6, Anton M. Koekemoer6, Jeffrey Newman7,
Brian D. Siana8
1INAFOsservatorioAstronomicodiTrieste,ViaG.B.Tiepolo11,34131Trieste,Italy
2 2Depar3tmINeAntFoOfAsssetrrvoantoomrioy,AUsntrivoenrosmityicoofdMiRasosmacah,uVseiattsF,ra7s1c0atNio3r3t,h00P0l4e0asManotnStetpreoertz,ioA(mRhMer)s,tI,tMalyA01003
1 4NationalOpticalAstronomyObservatory,POBox26732,Tucson,AZ85726,USA
0 5UCO/LickObservatory,UniversityofCalifornia,SantaCruz,USA
2 6SpaceTelescopeScienceInstitute, Baltimore,USA,
7UniversityofPittsburgh,Pittsburgh,PA15260USAand
n 8CaliforniaInstitute ofTechnology, CA,USA
a
Draft version January 30, 2012
J

6 ABSTRACT
2
We use the spatially–resolved, multi–band photometry in the GOODS South field acquired by the

CANDELSprojectto constrainthe natureofcandidate Lymancontinuum(LyC)emitters atredshift
]
O z ∼ 3.7 identified using ultra–deep imaging below the Lyman limit (1-sigma limit of ≈30 AB in a
2′′diameteraperture). In18candidates,outofasampleof19withfluxdetectedat>3-sigmalevel,the
C
light centroid of the candidate LyC emission is offset from that of the LBG by up to 1.5′′. We fit the
.
h SEDof the LyC candidates to spectralpopulation synthesis models to measure photometric redshifts
p and the stellar population parameters. We also discuss the differences in the UV colors between the
- LBG and the LyC candidates, and how to estimate the escape fraction of ionizing radiation (fesc) in
o
cases, like in most of our galaxies, where the LyC emission is spatially offset from the host galaxy.
r
t In all but one case we conclude that the candidate LyC emission is most likely due to lower redshift
s interlopers. Based on these findings, we argue that the majority of similar measurements reported
a
in the literature need further investigation before it can be firmly concluded that LyC emission is
[
detected. Our only surviving LyC candidate is a LBG at z =3.795, which shows the bluest (B−V)

1 color among LBGs at similar redshift, a stellar mass of M ∼2×109M⊙, weak interstellar absorption
v lines and a flat UV spectral slope with no Lyα in emission. We estimate its f to be in the range
esc
2 25%-100%,depending on the dust and intergalactic attenuation.
4
Subject headings: galaxies: distances and redshifts - galaxies: high-redshift - galaxies: formation
6
5
1. 1. INTRODUCTION lier, or z >7).
0 At z ∼ 3 − 4 the direct detection of LyC emission
The origin of the ionizing radiation, whether from
2 from galaxies is still difficult, because ionizing radiation
galaxies or AGN, responsible for the re-ionization of the
1 is severely attenuated by neutral gas and by dust in the
universeathighredshift,z >6,andforkeepingitionized
: interstellarandcircumgalacticmedium (ISMandCGM)
v at later epochs, is still poorly constrained. The contri-
of the source itself, as well as by the intervening IGM.
i bution of quasars to the hydrogen ionizing background
X increases as we look back in time from z =0 to z ∼2 as As a result, the number of reported detections of star–
forming galaxies at high redshift, typically selected as
r thepeakofthequasarluminosityfunctionisapproached
a Lyman–break galaxies or with equivalent criteria, with
(e.g.,Fanidakisetal. 2011). Beyondredshift2theircon-
LyC emission is very small (Shapley et al. 2006; Iwata
tribution significantly decreases and the hydrogen pho-
et al. 2009; Nestor et al. 2011; S06, I09 and N11, here-
toionization rate is most probably maintained by stellar
after; Vanzella et al. 2010b, V10b, hereafter; Boutsia
emission (e.g., Siana et al. 2008; Faucher-Gigu`ere et al.
et al. 2011). The issue is further complicated by the
2009; Haardt & Madau 2011; but see Fiore et al. 2011).
relatively high probability of finding faint low–redshift
Thestar-forminggalaxiesatz >3arethereforethelead-
galaxiesatvery closeangularseparation,i.e., ∼ 1 arcsec
ingcandidatestoprovidetheremainingionizingphotons.
or less, froma high–redshift galaxywith brighterappar-
The severe IGM absorption at z > 4.5 prevents us from
ent magnitude (Vanzella et al. 2010a, V10a, hereafter).
observingdirectlytheLymancontinuum(LyC)emission
These interlopers, for which spectroscopic identification
(Meiksin2006;Inoue&Iwata2008). Thisisparticularly
isnotviable,canbewronglyinterpretedasspotsofLyC
true during the re–ionization of the Universe (z >7). It
emissionfrom the host galaxyif high–angularresolution
is therefore essential to identify diagnostics of the LyC
multi–band photometry is not available, as is often the
emitters at lower redshift, z ≃3−4 (at λ >1216˚A),
rest case, to recognize their nature.
and infer if galaxies with these characteristics are more
Recently, searches of LyC emitters with deep imag-
commonduring the epochofre–ionization(one Gyrear-
ing below the Lyman limit have yielded samples of can-
didates at redshift z∼ 3 where the region of emission
[email protected]
2 Vanzella et al.
of the candidate LyC ionization radiation is spatially offset relative to the centroid of the rest–frame far–UV
offset from that of the non–ionizing rest far–UV light lightoftheLBG(thenLyCatwavelengtharound≈1500
(dubbed “nLyC” in what follows), namely the main ˚A). A visual inspection of the HST/ACS images showed
bodyofthegalaxytypicallyobservedbetween1500˚Aand that in 9 of the 28 sources the U–band flux light can
2000˚A.ThedisplacementoftheseputativeLyC–emitting unambiguously be explained as coming from the outer
“blobs” is generally less than one arcsec, but in some isophotes of nearby sources, which most likely are fore-
casesvaluesashighas2′′ havebeenreported(I09;N11), ground interlopers (e.g., see Figures 4 and 5 of V10b).
corresponding to separation in the range of several kpc We have eliminated these galaxies from the sample of
to a few tens of kpc. In other words, the candidate LyC LyC candidates. The remaining 19 candidates are listed
emissionwouldcomefromregionsofthegalaxiesthatare inTab.1andareanalyzedhere. Theyhavecounterparts
wellseparatedfromthecenter,whichiswheremostofthe intheACSandWFC3imageswhichareclosetothetar-
LyCphotonsareproduced,andinsomecasessofarfrom geted host LBG with angular separations in the range
it to be classified as separate sources. If confirmed this 0.4′′–1.9′′. Image cutouts for these sources are shown
would provide important empirical constraints on how in V10b, here we show in Figure 2 and the Appendix
LyCescapesfromgalaxies. Futureobservationswiththe the more critical cases where the angular separation is
new HST/WFC3 in the LyC rest-frame at z ∼ 3 will smaller than 1′′. All but one of the U–band sources
help exploring this issue further. In the meantime, how- are fainter, in AB magnitudes, than their ACS optical
ever,motivatedbythe potentialimportance ifthe above (rest UV) nLyC counterparts. Tab. 1 reports the ratio
findings are confirmed, we use spatially resolved, multi– of f1500/fLyC for the 19 objects.
bandphotometry fromHSTto constrainthe nature of a The remaining4 U–banddetections of the initial sam-
sample of LyC emitter candidates that we have selected ple of32areco–spatialwith the ACSandWFC3 images
in the GOODS South, one of the fields targeted by the of the host sources, in the sense that the centroid of the
CANDELSproject(Groginetal. 2011;Koekemoeretal. U–band light falls within the errors (∼ 0.1′′) on the lo-
2011)fromultra–deepU–bandimagingbelowtheLyman cation derived from the optical HST/ACS images. As
Limitofstar–forminggalaxieswithknownspectroscopic discussed in V10b (and reported in their Table 3), 3 out
redshift z ≥ 3.4. We also report on the only one Ly- of 4 are detected in the 4 Msec X–ray Chandra images
man Break Galaxy in our sample at z = 3.795 (Ion1, (0.5-8 keV), and also show typical signatures of AGN
hereafter) with a clear LyC emission. in their optical spectra, such as high ionization emission
lines (e.g., C IV, N V). The forth source, which we call
2. ANALYSISOFASAMPLEOFCANDIDATESLYC Ion1, is our only robust candidate stellar LyC emitter
EMITTERSINGOODS–SOUTH
and we will describe it further in Sect. 6.
We have selected a sample of LyC emitter candidates
fromultra–deepimaging inthe U–bandofthe GOODS– 2.1. The Effect Of The Intergalactic Attenuation
South field (V10b). The U–band images, which are de-
The method we are adopting here is based on an
scribedinNonino et al. (2009),wereaimedto probe the
intermediate–band filter (FWHM = 350˚A, U–band),
spectral region below the Lyman limit of these galaxies
and is similar to the typical narrow–band imaging only
for possible emission of ionizing radiation. The images
in the cases where 3.4 < z < 3.5 (9 out of 19 LBGs),
are very deep, reaching 1-σ flux upper limit of about
i.e. the ∆λ between the Lyman limit of the galaxy and
30 mag (AB) for an unresolved source within a circular
the red cut-off of the filter is minimal. In this case, the
aperture of 2′′ diameter.
greater depth of our image compensate for the larger
A relatively large number of sources, selected as
noise due to the broader bandpass. For example, if
Lyman–break galaxies, have secure redshift identifica-
compared with the NB3640 narrow–band filter used by
tion in the GOODS-S field (Vanzella et al. 2008, 2009;
N11 (FWHM =100˚A), the depth of our U-band image
Popesso et al. 2009; Balestra et al. 2010). As described
nearly exactly compensate for the differet filter setup.
in V10b,we startfrom a spectroscopicsample of 122B–
Differences between the methods arise if we consider,
band dropouts at 3.4 ≤ z ≤ 4.5 selected by Giavalisco
as in our approach, a variable redshift. Indeed, the LyC
et al. (2004) from the GOODS/ACS images for which
region probed by our filter is redshift dependent, e.g., it
we have robust redshifts, i.e., Quality Flag QF=A. The
is λ <909,800,741˚Aatz =3.4,4.0,4.4,respectively.
lower limit of the redshift range is the lowest value such rest
that the 912˚A Lyman limit is outside, redward of the WehaveinvestigatedtheeffectoftheIGMattenuationas
a function of redshift by running MC simulations as in
system throughput in the U filter, while the upper is
V10b (where the IGM attenuation is extensively taken
chosen because the rapid increase of opacity of the IGM
into account). To this end, as a reference, we adopt
produces too small transparency at higher redshift to
a magnitude of i = 24 and a ratio f /f = 7
make analysis of z > 4.5 galaxies useful (see V10b). 7 775 1500 LyC
(V10b, Siana et al. 2007). We then apply 10000 differ-
outof122B–banddropoutsaredetectedintheChandra
ent IGM transmissions convolved with the U-band filter
4Msecimagesinthe CDF-S(Xue etal. 2011)andclas-
and add photometric noise to the estimated flux in the
sifiedasAGN. Amongthese 122sourceswe haveflagged
U-band (see (V10b)). In these conditions, at z =3.4 we
aspotentialLyCemittercandidatesthe32ofthemwhich
haveflux detectionatthe 2σ levelorlargerwithin acir- retrieve 82+−44% of the sources in their LyC (U-band) at
cular aperture of 1.2′′ diameter in the U–band image. S/N > 2 (88+3% if i = 22). As redshift increases,
−4 775
In this work we discuss in detail these U-band emitters the fraction of recovered sources decreases because the
(named Uem hereafter). IGM attenuation (see Figure 1). The inner box of Fig-
As we noticed in V10b, in 28 of the 32 sources the U– ure 1 shows how the IGM affects the fraction of recov-
band emission, i.e., the candidate LyC light, is spatially ered sources as a function of redshift (normalized to the
Observations of LyC emission from z∼3–4 star-forming galaxies 3
z =3.4 case). superposition, which increases for the worst seeing con-
It is worth noting that the majority of the sources an- ditions. The fact that the LyC emission could be intrin-
alyzed here are at z <3.8 (15 out of 19). Moreover,two sically offset from the main galaxy (as reported in N11
examples at relatively high redshift have recently been and I09) further complicates the interpretation, if the
reported: a LBG at z =3.8 (Ion1, also discussed in this redshift of these potential emitters is not known. This
work)andaAGNatz =4.0withi =26.09,forwhich is stilltrue if highspatialresolutionimagesareavailable
775
a LyC emission at λ <830˚A and λ <800˚A is de- andthe sourcesarewellseparated,i.e.,inabsenceofthe
rest rest
tected at S/N of 5.2 and 3.3 by our deep U-band imag- redshift information an intrinsically offset LyC emission
ing, respectively (V10b). This suggest that despite the is fully compatible with the emission of a lower redshift
low average IGM transmission at z & 3.8, its stochas- object.
tic behavior makes the variance to be relatively large. On the one hand, the effects of this foreground con-
Indeed, at fixed redshift, the distribution of the IGM tamination can be corrected statistically. On the other
transmissions (U-band convolved) is asymmetric with hand, it is worthinvestigatingcarefully eachLyC candi-
an extended tail toward high values (Inoue & Iwata date,sinceanyconsiderationaboutthemechanismsthat
2008). The reason is that the Lyman continuum ab- allowtheLyCphotonstoescapeprimarilydependonthe
sorption by the IGM is very stochastic because it is re- reliability of the LyC detection. Therefore we now turn
lated to the presence of relatively rare Lyman limit sys- tothe discussionofobservationalevidence thatwill help
tems (LLS) or damped Lyman α system (DLA), having us to constrain the redshift, and hence the nature, of
NHI > 1017 cm−2. With a Lyman limit system, the these sources in our sample (Sect. 4).
transmission is suddenly cut down at the corresponding
4. CONSTRAININGTHEREDSHIFTOFTHECANDIDATES
wavelength. Conversely,withoutaLLS(orDLAsystem)
LYCEMITTERSINTHEGOODS-SSAMPLE
nearthesource,we canexpectasignificanttransmission
4.1. The Escape Fractions
evenfarbelowthesourceLymanlimit(seeInoue&Iwata
2008). In this section we calculate the absolute and relative
Having this in mind, we decided to keep the whole Lymancontinuumescapefractions(definedbelow)ofour
sample up to z ∼ 4.4 in the analysis performed in the potential LyC emitters assuming that they are at the
present work. If we could establish that the U–band same redshift of the LBG. Since f and (f ) have
esc esc,rel
selectedsourcesobservedinproximityoftheirLBGwere to obey to clear limits, this in turn puts constraints on
at the same redshift of their companions then we would other quantities, in particular the redshift. In order to
safely conclude that they are LyC emitters. Before to do that, we have to first define the relation among var-
investigate the nature of their redshift, in the following ious quantities and calculate the escape fraction where
sectionwebrieflyrecalltheissueofthecontaminationby spatially offset LyC and nLyC emissions are present.
lower redshift sources randomly placed at small angular
separations from the higher redshift LBG. 4.1.1. Escape Fraction From a Morphologically Resolved
Lyman Continuum
3. THEOCCURRENCEOFFOREGROUND
CONTAMINATION Following Siana et al. (2007) the observed flux ratio
The likelihood that a foreground interloper located in betweenthe1500˚AandtheLymancontinuumisaffected
the vicinity of a LBG is responsible for the detection in by several factors and is expressed as :
the U band increases with redshift of the LBG, with the
angular separation and the quality and depth of the ob- f1500 L1500
servations. Siana et al. (2007), I09 and N11 calculated = ×10−0.4(A1500−ALyC)×
(cid:18) f (cid:19) (cid:18)L (cid:19)
analyticallytheprobabilitythatcandidateLyCemission LyC OBS LyC INT
from a LBG is due to an interloper, giventhe character- ×eτHI,IGM(LyC)×eτHI,ISM(LyC), (1)
istics of the data (quality and depth). V10a performed
where LyC is the wavelength at which the Lyman con-
the same calculations and ran Monte-Carlo simulations
tinuumisobserved,(L1500/L ) istheintrinsiclu-
to quantify this effect. LyC INT
minosity density ratio, (A −A ) is the differential
The displacement of the U–band light relative to that 1500 LyC
dust attenuation (in magnitudes), τ (LyC) is the
at redder wavelengths, as measured from the light cen- HI,IGM
line-of-sight opacity of the IGM for LyC photons (and
troid in the U–band and ACS images, for the sample 19
galaxies discussed here, is in the range 0.4 < ∆θ < 1.9 transmission is TLIyGCM = e−τLIGyCM), and τHI,ISM(LyC)
arcsec; in all cases there always is a counterpart to the is the optical depth of the Lyman continuum absorp-
U–bandsourceinthe HST/ACSandWFC3 images. We tion from Hi within the observed galaxy’s interstellar
calculatedthatthenumberofsourcesobservedinthetwo medium, ISM (whose transmission is defined as THI =
annular bins with radii 0′′-1′′ and 1′′-2′′ from the LBG e−τHI,ISM(LyC)). ISM
centroid in the ACS z –band images is equal, within
850 The relative fraction of escaping LyC photons relative
the errors,to the expectations forforegroundgalaxiesat to the fractionofescaping nLyC (1500˚A) photons is ob-
increasingseparations. Thatis,giventhenumbercounts
tained rearranging the above equation :
and assuming a uniform distribution, the fraction of in-
tercepted foreground sources increases with the area of (L1500/L )
f = LyC int exp(τIGM), (2)
the annulus considered. esc,rel (F1500/F ) LyC
LyC obs
OneoftheparametersadoptedinV10awasthe seeing
(thePSFoftheimages),strictlyrelatedtothepossibility itcomparestheobservedfluxdensityratio(correctedfor
todeblendclosesourcesandrelatedtotheprobabilityof the IGMopacity)withmodels ofthe ultravioletspectral
4 Vanzella et al.
energy distribution of star-forming galaxies. If the dust one (see Siana et al. 2007, their Figure 2). Therefore
attenuation A is known, f can be converted to also the relative escape fraction has to be less than one.
1500 esc,rel
fesc as fesc = 10−0.4A1500fesc,rel (e.g., Inoue et al. 2005; 4.1.2. Anomalous Escape Fractions In Our Sample
Sianaetal. 2007). Again,fromtheaboveequations,f
esc
Following the discussion of the previous section, we
can be written as:
have estimated f for our sample in the correctway,
esc,rel
i.e.,insamespatialregionswheretheU–bandemissionis
fesc =exp[−τHI,ISM(LyC)]×10−0.4(ALyC), (3) observed, assuming that the U–band sources are at the
same redshift as the LBG, i.e., they are LyC emitters.
thetwofactorsontherightsidehavevaluesintherange
Following Eq. 2 we calculate the f by assuming an
[0–1]. Clearly, their product, i.e., f , cannotbe greater esc,rel
esc intrinsicratio(L1500/L ) =3(Shapleyetal. 2006)
than 1. LyC int
and the maximum IGM transmission at the redshift of
It has been recently argued by N11 that due to the
the LBG.1 The intrinsic ratiois also justified by the fact
details of the radiativetransfer and sourcesof the corre-
that the sources show an UV spectral slope (from their
sponding photons, any Lyα emission and escaping LyC
(i−z) color) similar or redder than their LBGs. There-
flux will not be necessarily co-spatial with either each
fore the adopted ratio is a conservative assumption.
otherorwiththebulkoftherest-frameUVfluxinagiven
Thef valuesarereportedinthelastcolumnofTa-
galaxy. This is true when considering Lyα emission, esc,rel
ble 1. All but one U–band LyC candidates have f
whichtypically arises frombackscatteringofmoving hy- esc,rel
larger than 100%, that would implies a less dust atten-
drogengasthatcanbe spatiallydecoupledfromthe ion-
uation for shorter wavelengths, i.e., A < A , in
izing sources. The observed light centroids (barycenter) LyC nLyC
contrast to any dust extinction law. This is even more
of LyC and nLyC can be displaced in the case in which
evident if we adopt the ratio (L1500/L ) =7 (e.g.,
the LyC arises from a sub-region of a larger area, how- LyC int
Siana et al. 2007; V10b), that increases the f
ever,thelocalemissionisnotspatiallyshifted,i.e.,ifthe esc,rel
values of Table 1 of a factor 7/3. The one case with
ionizationradiationismeasuredinsomesub-region,then
f <100%hasinconsistentphotometricredshiftand
the nLyC radiation is expected to be detected too (typ- esc,rel
UV colors as described below.
ically with brighter magnitude). The crucial point here
is thatthetwoquantities (F1500)obs and(FLyC)obs must 4.2. The UV Colors
be measured in the same spatial (i.e., physical) region,
The B−V color of galaxies at the redshift considered
where the ionizing and non-ionizing radiation arise.
here largely depends on the cosmic opacity of the IGM
Conversely, recent works have performed measures
(see Madau 1995), specifically due to the effects of the
with the aim to include both the fluxes (F1500) and
obs Lyα forest. Given the spatial correlation scale of the
(F ) byenlargingtheapertures(I09)orbyderiving
LyC obs IGM, if the U-band companions are at the same red-
total fluxes from the images regardless of the misalign-
shiftastheirassociatedLBGs,theirB−V colorsshould
ment (e.g., SExtractor MAG-AUTO in U and R bands,
therefore be similar.
as in N11), resulting in measures within different aper-
Figure 4 shows the difference of the observed (B−V)
tures size and/or shape. In this way, the measured ob-
color ∆(B − V) between the LBG and their compan-
servedratio(F1500/F ) ,andconsequentlythef
LyC obs esc ions LyC candidates (empty circles, ∆(B −V) = (B −
quantity, are strongly biased. In particular the resulting
V) −(B−V) )asafunctionofthetransversesep-
f would be severely underestimated, because the cor- LBG Uem
esc aration(physical) calculated at the mean redshift of the
rect (F1500) value would be smaller. Figure 3 shows
obs Lyαforest. Thesameisshownforthefour4.0<z <4.5
an illustrative example in the GOODS-S field (also dis-
galaxies for which the ∆(V −i) has been adopted, more
cussed below and reported in Tab. 1) in which the ob-
suitablethanthe(B−V)colorinprobingtheIGMdecre-
servedratiohasbeencalculatedwithandwithouttaking
ment. For all, the typical error in the color is .0.1 mag,
into accountof the offset emission, top and bottom pan-
becausethesesourcesaredetectedatmorethan10-sigma
els, respectively. Under the same assumptions (intrin-
in each band. As the figure illustrates, the LyC candi-
sic luminosity ratio and IGM transmission), the former
dates are much bluer than their companion LBGs, im-
method (top panel) produces an f that is ∼ 10 times
esc plyingeitheraveryhighfractionofescapingionizingra-
lower than the latter (bottom panel).
diationand/oranhightransmissionoftheIGM,ormore
The consequence is that if the estimated f exceeds
esc likelyas we are aboutto argue,that their redshift is sig-
the value of 100% then some other quantity has to be
nificantlylowerthanthatoftheLBG,inagreementwith
revised. It is evenmore significantif this happens under
the previous result and the photometric redshift analy-
conservative assumptions of (L1500/L ) , dust and
LyC int sis (reported in the next section). This is seen through
IGM attenuation.
a Monte Carlo simulations in which we have simulated
If, from one hand, the f has to be less than 100%,
esc f = 1 for the LyC candidates in the UV rest–frame
the constraints can be even more stringent if f has esc
esc,rel continuumprobedbytheB -bandandwehaveconser-
to be less than 100%. Indeed from Eq. 1 and 2 it turns 435
vatively assumed that the IGM transmission at λ<912
out that :
1 TIGM hasbeencalculatedfollowingtheInoue&Iwata(2008)
LyC
10−0.4(A1500−ALyC)×f =exp[−τ (LyC)] prescriptionconvolvedwiththeVIMOS/Ubandfilter,asdescribed
esc,rel HI,ISM in V10b; here we include in the calculations the recent statistics
(4)
of the LLS provided by Prochaska et al. (2010) and Songaila &
InordertohavethetransmissionoftheISMcorrectlyin Cowie(2010),thatincreaseslightlyTIGM (seeInoueetal. 2011).
LyC
therange[0–1](rightside),thef hastobelessthan
esc,rel TransmissionsattheredshiftoftheLBGhave beencalculatedon
100.4(A1500−ALyC), which in turns is a quantity less than 10000randomlineofsights.
Observations of LyC emission from z∼3–4 star-forming galaxies 5
˚A is the same as that at 912<λ<1215 ˚A (for the sim- 4.3. Photometric Redshift and SED Fitting
ulations we used the IGM transmission by Inoue et al.
To further investigate the nature of the U–band
(2008, 2011) and the SB99 UV templates with age 100
sources, namely whether regions of the host LBG from
Myr, continuous star formation and metallicity z=0.004
where LyC ionizing radiation is escaping or sources lo-
and Salpeter IMF). We can then correct the observed
cated at different, very likely lower, redshifts, we have
(B −V) color of the LyC emitters and recalculate the
fit the spatially–resolved CANDELS multi–band HST
∆(B−V), which we plot in the Figure 4 as filled circles
(BVizJYH),ground–based(UandK)andSpitzer/IRAC
(case in which the f = 1). The corrected color differ-
esc photometry to templates and spectral population syn-
ences still show that the LyC candidates are much bluer
thesis models to derive their photometric redshifts and
than their putative LBG companions. An even much
the parametersof their stellar populations (stellar mass,
more conservative correction can be made by assuming
star–formationrate, dust obscuration and age).
unitary IGM transmission blueward of the Lyα line, in
The images in the above bands have different angu-
allcasesandrecomputingthecolordifferences(filledtri-
lar resolution, and to robustly measure the photometry
angles), which still shows that the candidates would be
of our sources we have used the TFIT software package,
much bluer than the LBG.
developed by the GOODS and CANDELS teams (Lai-
The observed median (B −V) color and its 68% per-
dler et al. 2007), which uses positional priors and PSF
centile interval of the LyC candidates is 0.24−+00..4202, sig- informationtomeasureapparentmagnitudesinmatched
nificantly bluer than that of the LBG sample, which is apertures. For each source, TFIT uses the spatial posi-
1.85+−00..9558, in contrast with what would be expected if tion and morphology from the HST high–resolution im-
theirredshiftwerethesameasthatoftheLBGandtheir ages (we used both the ACS z–band and the WFC3/IR
stellar populations were similar (e.g., Meiksin 2006). H band) to construct a template image, which includes
We also explored a more extreme possibility by cal- nearby sources. While these are fully resolved at the
culating at a given redshift and fesc = 1 the expected HST angular resolution, they might be marginally or
colors arising from population III stars (Pop III). The evenheavilyblended in the other images. This template
PopIII stars of Inoue (2011)template (with zero metal- is then fit to the images of the object in all other low-
licity), convolved with the U-band filter and the IGM resolution bands after convolution with the appropriate
transmissions (1000 lines of sight) are partially able to PSF. During the fitting procedure, the fluxes of the ob-
reproduce the observed (B −V) colors only in the case ject in low-resolution bands are left as free parameters.
ofhightransparencyoftheIGM.Moreover,thepresence Thebest-fit fluxesareconsideredasthe fluxes ofthe ob-
of Pop III stars produces a very steep UV spectral slope ject in low-resolution bands. These procedures can be
(with a color (i−z)=−0.2), that is in contrastto what simultaneously done for several objects which are close
observed in our sample, i.e., < (i−z) >= 0.1±0.2 2. enough to each other in the sky. Experiments on both
Therefore we exclude the presence of Pop III stars. simulated and real images show that TFIT is able to
Regardlessoftheoreticalexpectations,inallcasesthat measureaccurateisophotalphotometryofobjectsto the
we considered the U–band companions have (B − V) limiting sensitivity of the image (Laidler et al. 2007).
color that are bluer by more than one magnitude (see We derive the photometric redshift and the physical
∆(B −V) in Tab. 1) than their corresponding LBG. If properties of the LyC emitter candidates by fitting the
placedatthe sameredshift,thelineofsighttoeachLyC observed spectral energy distributions (SEDs) to stellar
candidate/LBGpairwouldbeprobingtheIGMattrans- population synthesis models. Models used to measure
verse separations smaller than 20 kpc (physical) at the photo-zs are extracted from the library of PEGASE 2.0
meanredshiftofthe Lymanforest. This is muchsmaller (Fioc & Rocca-Volmerange 1997). Instead of using the
than the transverse cross-correlation function observed redshift with the minimum χ2, we integrate the proba-
in QSO pairs separated of several arcminutes, for which bility distribution function of redshift (zPDF) and de-
the coherence length in the IGM is at least of the or- rive the likelihood-weighted average redshift. When the
der of 500h−1 kpc proper, (e.g., Fang et al. (1996); zPDFhastwoormorepeaks,weonlyintegratethemain
D’Odorico et al. (1998); Rauch et al. (2005); Cappetta peak that has the largest power.
et al. (2010)). This implies that the IGM attenuation Since the U–band selected LyC candidates have coun-
due to the forest should be the same for the LyC candi- terpartsintheHSTz andHbandsineachcase,wehave
dates and their associatedLBGif they were at the same generatedtwo sets of fits, one based on positional priors
redshift. Since the (i−z) colors of the Uem sources are fromtheACSz–bandimages,whichsampletherestfar–
similar or even redder than those of the LBG, imply- UVSEDofthe galaxiesatλ.1800˚A,andtheotheron
ing comparable or even redder ultraviolet spectral slope priorsfromtheWFC3H–bandimages(λ.3600˚A).The
(similar stellar populations and obscuration properties),
first(GOODS)setusesphotometryintheUandtheACS
their(B−V)colorsshouldalsobesimilar,sincetheIGM
BVizbands,intheVLT/ISAACJHKimagesofGOODS-
attenuation would be in practise the same.
South described by Giavalisco et al. (2004) and Retzlaff
Thus, we interpret these simulations as evidence that
et al. (2010), and in the four GOODS Spitzer/IRAC
the LyC candidates are not regions of the LBG where
bands. The second (CANDELS) set uses the WFC3/IR
ionizing radiation is escaping from but rather relatively
YJH near–IR photometry in place of the ground–based
unobscured star–forming galaxies at significantly lower
one. The z–band positional priors might better reflect
redshift than the LBGs.
thespatiallocationoftheLyCemittingregions,sincees-
sentiallythe same starsthat powerit alsopowermostof
2 The bluest sources inour sample(two with −0.2<(i−z)<
−0.1) have a (B −V) color that is bluer than the bluest value the lightin the nLyC far UV spectrum. The z–bandim-
amongthe1000realizationscalculated forPopIIIstars. agesalsohavebetterangularresolutionthantheH–band
6 Vanzella et al.
ones. The second set offers an independent set of mea- data to the derivationof the photometric redshift makes
sures,whichtakesadvantageofthegenerallymoresensi- no difference. In the other two, neglecting the U–band
tiveandhigher–resolutionWFC3data;itlacks,however, data results in very good agreement between the spec-
the K–band data. troscopicandphotometricdata,whileincludingitlowers
Since the potentialemissionin the LyC spectral range the photometric redshift solution.
is not taken into account in the models, and thus flux The LyC emitter candidate with coordinates
in the U–band could skew the photometric redshift cal- RA=03:32:36.85, DEC=-27:45:57.6, associated with
culation towards lower values, for both photometric sets galaxyHUDF-J033236.83-274558.0is discussed in detail
we have run the fit with and without the U–band pho- in the Appendix. It is the fainter among the cases
tometry. Figure 5 shows the redshift probability func- discussed in this work with an U mag of 28.6. A new
tionderivedfromthe GOODSphotometric set; Figure 6 observed feature in the J-band (at 10σ significance) is
the one from the CANDELS set. In all cases the solid consistent with relatively strong line emission of a low
curve refers to the calculation made using the U–band redshift galaxy, lower than its companion LBG.
photometry, the dashed curve without it. In all cases A general characteristic of the fits is that including or
the photometric redshifts are calculated as the weighted nottheU–bandphotometryinthederivationofthepho-
average of zPDF as: tometricredshiftoftheLyCemittercandidatesmakesno
substantialdifferenceintheresultsandthatthereisvery
∞
zP(z)dz good quantitative agreement between the GOODS and
z = 0 . (5)
phot R ∞P(z)dz CANDELSphotometricredshift, pointingto the robust-
0 ness of our analysis.
R
A blank panel in Figure 6 means that the source is out- Finally, a further check has been performed by cal-
sideoftheregioncoveredbytheCANDELSobservations. culating the photometric redshifts without the inclusion
The generalresultfromouranalysisis thatthe photo- of the U and B bands, that makes the test indepen-
435
metric redshifts of the LyC emitter candidates are sys- dent from the single B −V color analysis described in
tematically much smaller than the spectroscopic red- Sect. 4.2. Even though the efficiency of the photomet-
shifts of their host LBGs. Exceptions include the ric redshift prediction is slightly reduced, the resulting
cases of the LyC candidates GDS J033204.91-274451.0, z best-fit values still favor low redshift solutions for
phot
GDS J033220.95-275021.8, GDS J033222.95-274727.8, allsources,exceptforGDSJ033223.32-275155.9thatin-
and GDS J033223.32-275155.9 (see Figures 5 and 6), creasesfromz =0.07±0.02to3.60±0.2. Inthiscase,
phot
where the GOODS photometric redshift is much lower however, the (B −V) = 0.65 color, the relative escape
thanthespectroscopicredshiftoftheLBG,buttheCAN- fractionandthe∆(B−V)measurementsareincontrast
DELS photometric redshift is not. These are cases in with this high redshift solution.
which the separation between the LBG and the Uem is
.1′′ andarenotindividually detectedin the CANDELS 4.4. Conclusion From fesc, UV Colors and Photometric
Redshift Analysis
H-band based catalog (they are markedwith red crosses
in Figure 6). In summary, the combination of photometric redshift,
• In the firstcase,GDS J033204.91-274451.0,the LyC relative UV colors (that trace the IGM decrement) and
emitter candidate is resolved as a different source from the relative escape fraction estimated for the LyC can-
the LBG in the ACS z–band image but not in the H– didates all argue against them being ionizing radiation
band one. The photometric redshift from the GOODS escaping form the nearby LBG. Rather, they are most
photometry yields a significantly lower redshift than the likely low–redshift interlopers (as we had originally ar-
LBG’s spectroscopic one, with some dependence of the gued in V10b). The more extreme case among the
redshift probability function on the exclusion of the U– sources here analyzed (HUDF J033236.83-274558.0) is
bandfromthe fit, but the resultingphotometric redshift discussed in detail in the Appendix, in which we report
isinanycasemuchlowerthantheLBG’s. SincetheLyC new observational evidence supporting a possible con-
candidate and the LBG are not resolved in the H band, tamination from a superimposed object.
theCANDELSphotometricredshiftisessentiallythatof Finally, we conclude by noting that deep UV imaging
theLBGitself. IftheU–bandphotometryisnotincluded withHST/WFC3willbeabletoclarifythenatureofthe
inthe fit, the spectroscopicandphotometricredshiftare U–bandoffsetLyCcandidates,sincetheprobabilitythat
in very good agreement. Including, the U–band data, LyC radiationat λ≪912˚Aescapes the host galaxy and
however, yields a photometric redshift somewhat lower eludes the IGM is extremely low (for an estimate of this
thanthe spectroscopicone (zspec =3.404,zphot=3.351), probabilityatλ<700˚AseeInoue&Iwata2008;Meiksin
which is not surprising, since this is the LBG with the
2006).
lowestredshiftinoursampleandthussomefluxdetected
inthebluestbandmighteasilyskewthefittowardlower 5. ACRITICALANALYSISOFTHECURRENT
redshift values. MEASUREMENTS
• The other three sources (GDS J033222.95-274727.8, In this section we review all known (to us) galaxies at
GDS J033220.95-275021.8, GDS J033223.32-275155.9) high redshift (z &3) for which emission of LyC ionizing
are also cases in which the LyC candidate and the LBG radiation has been reported or suspected.
areclassifiedasindividualsourcesinthe z–bandbutnot • Direct detection of LyC ionizing radiation has been
in the H band. In both cases, the GOODS photometric reported by Shapley et al. (2006, S06) in deep spectra
redshiftismuchlowerthantheLBG’sspectroscopicone. of two LBGs at z ∼ 3 (U–band dropouts) in the SSA22
In the former, since the LBG redshift is at the high end field,dubbedD3andC49intheirnomenclature,outofa
ofour sample (z =4.440),adding or notthe U–band totalof 14 sourcesobservedwith comparablesensitivity.
spec
Observations of LyC emission from z∼3–4 star-forming galaxies 7
It is worth noting that I09 and N11 found a highly sig- extreme stellar populations to justify the flux ratios
nificant null detection for the object SSA22a-D3, which observed, (B) the LBG reported in S06 (SSA22-C49)
is the brighter of the two objects for which S06 claimed that shows an offset (0.5′′) U-band emission whose red-
an ionizing flux detection. shift is not conclusively known, and finally (C) the one
• Iwata et al. (2009) used deep narrow–band (NB) we present in the next section and discovered in the
imaging to image the rest–frame SED blueward of the GOODS-S field (Ion1).
912˚ALymanlimit ofmembers ofa well–knownoverden- 6. ALYCEMITTERFOUNDINTHEGOODSSAMPLE:
sityofgalaxiesatz ∼3.1,alsointheSSA22field(Steidel ION1
et al. 1998). They reported 17 detections, 7 from galax-
The direct detection of LyC emitters at high redshift
ies selected as LBGs and 10 as LAEs.
(ashighaspossible),allowusto characterizetheir prop-
• Nestor et al. (2011) used the same technique to ob-
erties at λ>1215.7˚A and try to identify similar sources
servegalaxiesin the sameSSA22 field, but extended the
sensitivity of previous observations by ≈0.6 mag. They and/or look if they are more common during the re–
ionizationepoch. Forthis reasonitis importanttoiden-
reported34detections,outof156sources,whichinclude
tify secure LyC emissions. We recall one of the most
6 LBGs and 28 LAEs.
• In the paper V10b, we used a technique similar, but promising stellar ionizers at redshift 3.795we have iden-
tified in the GOODS-S field and currently the highest
notidentical,totheonebyI09andN11. Insteadofusing
redshift known so far (GDS J033216.64-274253.3, Ion1)
anarrow–band,wecarriedoutultra–deepU–bandimag-
and is useful to revisit it in the context of this work by
ing to searchfor candidate LyC emitters amonggalaxies
at 3.4 . z . 4.5, namely such that the redshifted 912˚A comparing its appearance with the offset Uem sources
discussed above.
ionization edge of the targeted sources is to the red and
completely outside of the filter’s bandpass. We found 1 6.1. Observed Photometric and Spectroscopic Properties
candidate out of 102 LBGs (described in Sect. 6). of Ion1
FormostoftheNB–selectedcandidatesandLBGsthe
This source was initially reported in V10b, and here
region where the LyC ionization radiation originates is
we briefly summarize and report on new observational
foundto be spatiallyoffsetfromthat ofthe non-ionizing
constraints, and compare it with the offset Uem sources
far–UV light, namely the main body of the galaxy. The
discussed above.
displacement is generally less than one arcsec, but in
some cases values as high as 2′′ have been reported. SED:ThesourcehasnotbeendetectedintheF225W,
F275W and F336W channels of HST/WFC3 observa-
We performed a visual inspection of the images of the
tions in the GOODS-S (down to 26.3, 26.4, 26.1 at 5σ
galaxiesby I09 and N11,and consulted Table 4 and 5 of
forpoint-likeobjects,Windhorstetal. (2011)). Itliesin
N11, and found that & 70% of the presently identified
the B-band dropout selection scheme, with (B-V)=1.70
LyC emitter candidates exhibit a spatial offset between
and(V-z)=0.40(V09),andhasablueUVspectralslope,
the LyC emission and the “non-ionizing” far–UV image
having an (i−z)=−0.015, and a β =−2.09±0.16 de-
of the galaxies. Since an offset emission is also consis-
rived from Castellano et al. (2011). It is worth noting
tent with a lower redshift interloper, a key step is to
that this source has the bluest (B−V) color among the
measure the statistical occurrence of these cases (as we
available 17 LBGs with 3.7 < z < 3.9 and compara-
did in V10a). In particular with availability of large–
ble UV slope. This is consistent with the fact that the
area HST UV and near–IR surveys with WFC3, such as
B –bandis probingthe rest-frameintervalλ.1020˚A,
CANDELS, the prospect for substantial progress in this 435
in which the LyC is contributing to the observed flux
regard is good. The other important point is that if the
(a S/N of 10 is measured in this band) and reduces the
offset LyC emission is real, then care must be exercised
(B −V) color. The probability that zphot > 3.4 is ≃
when calculating the fraction of escaping ionizing radia-
100%. Ithasnotbeendetectedinthenew4MsChandra
tion f in each case, i.e., whether it is from the galaxy
esc
X-ray observations,for which we set a 1σ upper limit of
asa whole or onlyfroma close companion,to avoidbias
L < 3×1042erg/s, nor in the 24µm SPITZER/MIPS
in the derivation of average properties. For example it X
observations (Santini et al. 2009). The SED is fully
is worth mentioning that the offset possible LyC detec-
comparable to that of a star-forming galaxy, in particu-
tion in S06, SSA22-C49, has a reported observed ratio
of (F1500/F ) = 16.4±6.1 (N11), in which the larfromthe SEDfitting we obtain: SFR≃50M⊙yr−1,
flux F1500 uLsyeCd OinBSthe calculation is that of the main stellar mass of 2.3×109M⊙, E(B −V) . 0.1 (see Fig-
ure 7).
galaxy. Presumably, a calculation restricted to the local
Optical spectrum: Optical spectra from
region where the putative Lyman continuum emission is
Keck/DEIMOS and VLT/VIMOS have been ob-
seen would produces a much smaller value, i.e., a much
tained, although the latter is too shallow and the
higher f .
esc,rel
S/N prevent us to add information. From the Keck
Finally,itmustbekeptinmindthatpartofthecandi-
spectrum it is clear that the Lyα line is not in emission,
dates LyC emitters reported above may have the contri-
the continuum–break is evident together with the
bution to f from an AGN component that is difficult
esc [Siiv]1393.8–1402.8 and [Civ]1548.2–1550.8 absorption
to identify without multi-wavelength and spectroscopic
lines, at redshift 3.795. Also [Cii]1335.1 in absorption
surveys.
seems to be present. The other UV absorption lines like
Therefore the amount of stellar LyC emission from
[Oi]1302.2,[Siii]1260.4and[Siii]1526.7arenotdetected
high-z star-forming galaxies is still uncertain. Very few
(see Figure 7).
sources have spectroscopic redshift confirmation: (A)
UV and B-band rest-frame morphology: At
those (five) reported by Inoue et al. (2011) that need
1700˚A/1900˚A and B–band rest-frame wavelengths the
8 Vanzella et al.
galaxyshowsaresolvedandcompactshape,withanhalf- ing radiation and at the same time a high Lyα equiva-
lightradiusof0.9kpc(physicalsize)inboththez and lent width in emission; it can be realized with a sort of
850
H bands (the latter from CANDELS), respectively (see “unipolar” outflow behind the source that backscatters
Figure 7). the Lyα photons to the observer (as is usually seen),
The hypothesis that it is a faint AGN (L < 3 × while along the line of sight (the front of the galaxy),
X
1042erg/s) is still open. However, the spectral features the ISM andCGM are free fromdust or gasattenuation
in the optical spectrum and the shape of the SED from and the LyC radiation can escapes. However, situations
theU–bandtothefarinfrared(Spitzer/MIPS24µm)are like this have never been clearly observed until now and
fully compatible with a star-forming and relatively low the interplaybetweenthe escapefractionofionizingand
mass galaxy. In the following we assume that it is a Lyα photons is still not known.
star-forming galaxy. It is worth noting that the absence of Lyα emission
TheLyCandthenLyCemissions(namelyUandi775– fromanobjectwithhighfesc,likeIon1,makesextremely
bands)arespatiallyalignedwithintheerrors,∆θ <0.1′′, difficultthespectroscopicconfirmationofsimilarsources
and the LyC detection at λ < 830˚A implies a rela- at z > 7, i.e. during the reionization epoch. Indeed,
tively transparent line of sight free from LLSs. From if place at z = 7 it would be one magnitude fainter,
Eq. 2 we derive a minimum f = 82% assuming Y ≃ 25.9, and the continuum break hard to measure
esc,rel
(L1500/L ) = 7 and a IGM transmission of 0.575 spectroscopically (Vanzella et al. 2011; Fontana et al.
LyC int
(maximum value for our U-band filter of the 10000 re- 2010;Pentericci et al. 2011).
alizations at z=3.8). Given its E(B − V) . 0.1 and 7. CONCLUSIONS
adoptingthe Calzettiextinctionlawwehavef >56%
esc
(Calzetti et al. 2000). Assuming a intrinsic value for The observation of the LyC in distant galaxies is a
(L1500/L ) = 3 the escape fractions are f > difficult task because several attenuations occur: in the
LyC int esc,rel
35% and f >24%. galaxy itself and circumgalactic medium (by dust and
esc
nutral hydrogen gas) and along the intergalactic travel
6.2. How Does Ion1 Compare With the Rest of the LBG (Lyman alpha forest, LLSs and DLAs). A further com-
Population ? plicationisrelatedtothepossiblepresenceofforeground
(lower-z) superimposed sources that can mimic the LyC
It has been firmly established that as the Lyα equiva-
emissionofthebackgroundgalaxy. Itisthereforecrucial
lentwidthincreasesfromabsorptiontoemission,theUV
to establish whether they are genuinely associated with
spectralslope becomes bluerandthe strengthofthe UV
the redshift of the main LBG.
absorption lines (stellar and interstellar) drastically de-
In this work we have discussed three diagnostics that
creases(see,e.g.,Shapleyetal. 2003;Korneietal. 2010;
can be used to constrain the redshift of these offset and
Balestra et al. 2010; Vanzella et al. 2009). Moreover,
faint sources. To this end, the deep and high angular
on average,high Lyα equivalent widths tend to be asso-
resolution multi-wavelength images available from the
ciated to UV compact morphologies; conversely sources
GOODS and CANDELS surveys have been exploited.
withLyαinabsorptionappearslesscompact(seeLawet
In particular, starting from the sample of U-band emit-
al. 2007; V09; Pentericci et al. 2010). The Ion1 galaxy
ters identified in V10b, we (1) calculated accurate pho-
shows :
tometric redshifts, (2) considered the IGM radial and
transverse absorption (simulating UV colors and com-
1. a blue UV spectral slope β = −2.09±0.16, from
paringthoseofLBGsandtheirU-bandemitters)and(3)
the photometric fitting of Castellano et al. (2011).
provided new constraints on their nature by calculating
2. relatively weak UV absorption lines, from the the fesc quantity under conservative assumption. These
Keck/DEIMOS spectrum (see Figure 7). threeanalysissuggestthatnoneofoffsetU-bandsources
are actual sources of LyC emission from the LBGs at
3. a compact morphology, 0.9 kpc physical half light z > 3.4, but are instead foreground interlopers. This
radius in the 1600˚A and 3300˚A rest-frame wave- strengthens the results of V10b, i.e., the median fesc
lengths (from HST/ACS and WFC3). quantity is very small for L > L∗ galaxies, or assum-
ing a bimodaldistribution for the f , the highervalues
esc
Theseareallfeaturesthatpositivelycorrelatewiththe are rare.
equivalentwidth ofthe Lyα in emission, nonetheless the At present, the offset candidates reported in litera-
Lyα is absent. In this respect it deviates from the aver- ture by S06, I09 and N11 need spectroscopic redshift
age LBG population, particularly if we consider its LyC confirmation and are difficult to access if deep multi-
emission. The fact that the Lyα line is not in emission wavelength high angular resolution imaging is not avail-
suggests that it can be absorbed and/or it is intrinsi- able.
cally weak because the ionizing radiationis escaping the Moreover, we note that in order to avoid biased mea-
galaxy. The detection of LyC emission would suggest sures, the calculation of the f quantity in the case
esc
low gas attenuation along the line of sight and the blue of spatially offset emission must be performed carefully
UV slope low dust absorption. In the extreme case in consideringquantities arisingfromthe same physicalre-
whichallthe ionizingradiationisescaping(f is100%, gion. We argue that the measurements reported by S06,
esc
or is close to), nebular emission lines like Lyα , Oxygen I09andN11areaffectedbythisproblemandneedfuther
or the Balmer transitions are no longer pumped, and investigations.
therefore drop to intrinsecally small equivalent widths The currentsituationis far fromclear. The number of
(Schaerer 2003). We can image a geometrical configura- bona fide LyC detections at z > 3 is very small and not
tion in which there is a high fraction of escaping ioniz- statistically significant.
Observations of LyC emission from z∼3–4 star-forming galaxies 9
We have discoveredone good candidate (named Ion1) of the f or due to other effects like the dust attenua-
esc
that is currently the highest redshift galaxy known tion, that however, would be small, giving the blue UV
with direct LyC detection. Its LyC emission at λ < slope.
830˚A (probed by the U-band) is aligned with the source Its relatively low stellar mass (2.3×109M⊙) may also
detected in the UV nLyC, i.e., no offset is observed beasignaturethatfeedbackprocessesweremoreefficient
with this resolution. Apart from the three cases re- to clean the line of sight in this type of galaxy, favoring
portedby Inoue et al. (2011)with alignedLyC emission the escape of ionizing radiation. Precisely in this regard
that need Population-III stars to explain their observed a measure of the wind velocity is important (Heckman
f1500/f fluxratios(evensmallerthanone),theIon1 et al. 2011; Overzier et al. 2010). We note that Ion1
LyC
emission is easily explainable with standard stellar pop- is not a LAE, therefore it is important to complement
ulation and an f > 25%. If the AGN component is the LAE surveys with sources like this. Clearly sources
esc
negligible, as seems to be the case from the X-ray and like Ion1 placed at z>7 would be extremely difficult to
spectroscopic data, it would be the most promising stel- detectspectroscopicallywithpresentfacilities. However,
lar ionizer that may resemble those responsible for HI since at z > 7 the direct measure of LyC is not feasi-
reionization. ble, sources like Ion1 are the only viable ways we have
It is worthnoting that a high value ofthe escape frac- to investigate the mechanism that allow the ionization
tion would correspond to faint nebular emission (e.g., ratiationtoescape,i.e.,toaddresstheinterplaybetween
Robertson et al. 2010). In this case the Lyα line is the fesc quantity and the non-ionizing UV features like
not in emission, despite the fact that the source shows a the strength of the nubular emission lines (e.g., Oxigen,
blue UV spectral slope (β = −2.09), relatively weak in- Balmer and Lyman lines), UV slopes, nature of the stel-
terstellar absorption lines and compact morphology, all lar populations, geometry, winds, etc.
characteristicsthatpositivelycorrelatewiththestrength
ofthe Lyα emissionline (e.g., Shapley et al. 2003;V09;
Weacknowledgefinancialcontributionfromthe agree-
Pentericci et al. 2010; Balestra et al. 2010). It is not
ment“COFIS”ASI-INAF1/009/10/0. Wewouldliketo
cleariftheabsenceofLyα isconnectedtothehighvalue
thank A. K. Inoue for providing us the transmissions of
the IGM and the spectral template of Pop III stars.
REFERENCES
Balestra,I.,Mainieri,V.,Popesso,P.,Dickinson,M.,etal.,2010, Heckman,TimothyM.,Borthakur,Sanchayeeta, Overzier,
A&A,512,12 Roderik,Kauffmann,Guinevere, Basu-Zych,Antara,Leitherer,
Beckwith,S.V.W.,etal.2006,AJ,132,1729 Claus,Sembach, Ken,Martin,D.Chris,Rich,R.Michael,
Boutsia,K.,Grazian,A.,Giallongo,E.,Fontana, A.,Pentericci, Schiminovich,David,Seibert,Mark,2011,ApJ,730,5
L.,Castellano,M.,Zamorani,G.,Mignoli,M.,Vanzella,E., Hu,EstherM.,Cowie,LennoxL.,Kakazu,Yuko,Barger,AmyJ.,
Fiore,F.,Lilly,S.J.,Gallozzi,S.,Testa,V.,Paris,D.,Santini, 2009,ApJ,698,2014
P.,2011, ApJ,736,41 Inoue,A.K.,Iwata,I.,Deharveng, J.-M.,Buat,V.,Burgarella,
Calzetti,Daniela,Armus,Lee,Bohlin,RalphC.,Kinney,AnneL., D.,2005,A&A,435,471
Koornneef,Jan,Storchi-Bergmann,Thaisa,2000,ApJ,533,682 Inoue,AkioK.,Iwata, Ikuru,2008,MNRAS,387,1681
Cappetta, M.,D’Odorico,V.,Cristiani,S.,Saitta,F.,Viel,M., Inoue,A.K.,Kousai,K.,Iwata, I.,Matsuda,Y.,Nakamura,E.,
2010, MNRAS,407,1290 Horie,M.,Hayashino,T.,Tapken, C.,Akiyama,M.,Noll,S.,
Castellano,M.,Fontana, A.,Grazian,A.,Pentericci,L.,Santini, Yamada,T.,Burgarella,D.,Nakamura,Y.,2011, MNRAS,
P.,Koekemoer,A.,Cristiani,S.,Galametz, A.,Gallerani,S., 411,2336
Vanzella,E.,Boutsia,K.,Gallozzi,S.,Maiolino,R.,Menci,N., Inoue,A.K.,2011,MNRAS,415,2920
Paris,D.,2011, arXiv/1109.1757 Iwata,I.,Inoue, A.K.,Matsuda,Y.,etal.,2009,ApJ,692,1287
D’Odorico,V.,Cristiani,S.,D’Odorico,S.,Fontana, A., Izotov, YuriI.,Guseva,NataliaG.,Thuan,TrinhX.,2009,ApJ,
Giallongo,E.,Shaver,P.,1998,A&A,339,678 728,161
Fanidakis,N.,Baugh,C.M.,Benson,A.J.,Bower,R.G.,Cole, Laidler,VictoriaG.,Papovich,Casey,Grogin,NormanA.,Idzi,
S.,Done,C.,Frenk,C.S.,Hickox,R.C.,Lacey, C.,DelP. Rafal,Dickinson,Mark,Ferguson,HenryC.,Hilbert,Bryan,
Lagos,C.,MNRAS,tmp.2006 Clubb,Kelsey,Ravindranath,Swara,2007,PASP,119,1325
Fang,Yihu,Duncan,RobertC.,Crotts,ArlinP.S.,Bechtold, Law,DavidR.,Steidel,CharlesC.,Erb,DawnK.,Pettini,Max,
Jill,1996,ApJ,462,77 Reddy,NaveenA.,Shapley,AliceE.,Adelberger,KurtL.,
Faucher-Gigu`ere,Claude-Andr´e,Lidz,Adam,Zaldarriaga, Simenc,DavidJ.,2007,ApJ,656,1
Matias,Hernquist,Lars,2009,ApJ,703,1416 Madau,P.,1995,ApJ,441,18
Fiore,F.,Puccetti,S.,Grazian,A.,Menci,N.,Shankar,F., Meiksin,Avery,2006,MNRAS,365,807
Santini,P.,Piconcelli,E.,Koekemoer,A.M.,Fontana, A., Oke,J.B.,1974,ApJS,27,21
Boutsia,K.,Castellano,M.,Lamastra,A.,Malacaria,C., Kakazu,Yuko,Cowie,LennoxL.,Hu,EstherM.,2007, ApJ,668,
Feruglio,C.,Mathur,S.,Miller,N.,Pannella,M.,2011, 853
arXiv/1109.2888 Koekemoer,A.M.etal.2011,ApJS,197,36
Fontana, A.,Vanzella,E.,Pentericci,L.,Castellano,M., Kornei,KatherineA.,Shapley, AliceE.,Erb,DawnK.,Steidel,
Giavalisco,M.,Grazian,A.,Boutsia,K.,Cristiani,S., CharlesC.,Reddy,NaveenA.,Pettini,Max,Bogosavljevi´c,
Dickinson,M.,Giallongo,E.,Maiolino,R.,Moorwood,A., Milan,ApJ,711,693
Santini,P.,ApJ,725,205 Nestor,D.B.,Shapley,A.E.,Steidel,C.C.,Siana,B.,ApJ,in
GiallongoE.,CristianiS.,D’OdoricoS.,etal.,2002,ApJ,568,9 press,(arXiv/1102.0286), N11
Giavalisco,M.,Ferguson,H.C.,Koekemoer,A.M.,Dickinson, Nonino,M.,Dickinson,M.,Rosati,P.,etal.,2009,ApJ,183,244
M.,Alexander,D.M.,Bauer,F.E.,Bergeron,J.,etal.2004, Overzier,RoderikA.,Heckman,TimothyM.,Tremonti,Christy,
ApJ,600,L93 Armus,Lee,Basu-Zych,Antara,Gonalves,Thiago,Rich,R.
Grogin,N.A.etal.2011,ApJS,197,35 Michael,Martin,D.Christopher,Ptak,Andy,2009. ApJ,706,
FrancescoHaardt,PieroMadau,2011, ApJ, 203
submitted,(arxiv/1105.2039) Pentericci,L.,Grazian,A.,Scarlata,C.,Fontana, A.,Castellano,
M.,Giallongo,E.,Vanzella,E.,2010, A&A,514,64
10 Vanzella et al.
Pentericci,L.,Fontana, A.,Vanzella,E.,Castellano,M.,Grazian, Steidel,C.C.,Pettini,M.,&Adelberger,K.L.2001, ApJ,546,
A.,Dijkstra,M.,Boutsia,K.,Cristiani,S.,Dickinson,M., 665
Giallongo,E.,Giavalisco,M.,Maiolino,R.,Moorwood,A., Vanzella,E.,Cristiani,S.,Dickinson,M.,Giavalisco,M.,etal.,
Paris,D.,Santini,P.,2011,ApJ,743,132 2008,A&A,478,83
Popesso,P.,Dickinson,M.,Nonino,M.,Vanzella,E.,Daddi,E., Vanzella,E.,Giavalisco,M.,Dickinson,M.,etal.,2009,ApJ,695,
etal.,2009, A&A,494,443P 1163
Prochaska,J.Xavier,O’Meara,JohnM.,Worseck,Gabor,ApJ, Vanzella,E.,Siana,B.,Cristiani,S.,Nonino,M.,2010a,MNRAS,
2010, 718,392 404,1672
Rauch,Michael,Becker,GeorgeD.,Viel,Matteo, Sargent, Vanzella,E.,Giavalisco,M.,Inoue, A.K.,Nonino,M.,Fontanot,
WallaceL.W.,Smette, Alain,Simcoe,RobertA.,Barlow, F.,Cristiani,S.,Grazian,A.,Dickinson,M.,Stern,D.,Tozzi,
ThomasA.,Haehnelt,MartinG.,ApJ,632,58 P.,Giallongo,E.,Ferguson,H.,Spinrad,H.,Boutsia,K.,
Retzlaff,J.,Rosati,P.,Dickinson,M.,Vandame,B.,Rit´e,C., Fontana, A.,Rosati,P.,Pentericci,L.,2010b,ApJ,725,1011
Nonino,M.,Cesarsky,C.,GOODSTeam,2010,A&A,511,50 Vanzella,E.,Pentericci,L.,Fontana, A.,Grazian,A.,Castellano,
Robertson,BrantE.,Ellis,RichardS.,Dunlop,James S., M.,Boutsia,K.,Cristiani,S.,Dickinson,M.,Gallozzi,S.,
McLure,RossJ.,Stark,DanielP.,2010,Nature,468,49 Giallongo,E.,Giavalisco,M.,Maiolino,R.,Moorwood,A.,
Santini,P.,Fontana, A.,Grazian,A.,Salimbeni,S.,Fiore,F., Paris,D.,Santini,P.,2011,ApJ,730,35
Fontanot, F.,etal.,2009,A&A,504,751 Windhorst,RogierA.,Cohen,SethH.,Hathi,NimishP.,
Schaerer,D.,2003, A&A,397,527 McCarthy,PatrickJ.,Ryan,RussellE.,Jr.,Yan,Haojing,
Shapley,A.E.,Steidel,C.C.,Pettini,M.,Adelberger,K.L.,2003, Baldry,IvanK.,Driver,SimonP.,Frogel,JayA.,Hill,David
ApJ,588,65 T.,Kelvin,LeeS.,Koekemoer,AntonM.,Mechtley,Matt,et
Shapley,A.E.,Steidel,C.C.,Pettini,M.,Adelberger,K.L.,& al.,ApJ,193,27
Erb,D.K.2006,ApJ,651,688 Xue,Y.Q.,Luo,B.,Brandt,W.N.,Bauer,F.E.,Lehmer,B.D.,
Siana,B.,Teplitz,H.I.,Colbert,J.,etal.,2007,ApJ,668,62S Broos,P.S.,Schneider,D.P.,Alexander,D.M.,Brusa,M.,
Siana,Brian,Polletta,MariadelCarmen,Smith,HardingE., Comastri,A.,Fabian,A.C.,Gilli,R.,Hasinger,G.,
Lonsdale,CarolJ.,etal.,2008, ApJ,675,49 Hornschemeier,A.E.,Koekemoer,A.,Liu,T.,Mainieri,V.,
Songaila,Antoinette, Cowie,LennoxL.,2010,ApJ,7211448 Paolillo,M.,Rafferty,D.A.,Rosati,P.,Shemmer,O.,
Steidel,CharlesC.,Adelberger,KurtL.,Dickinson,Mark, Silverman,J.D.,Smail,I.,Tozzi,P.,Vignali,C.,2011,ApJ,
Giavalisco,Mauro,Pettini,Max,Kellogg,Melinda,1998,ApJ, 195,10
492,428