VLT/XSHOOTER & Subaru/MOIRCS Spectroscopy of HUDF-YD3: No Evidence for Lyman-alpha Emission at z=8.55

0.26 MB·English

by  Andrew J. Bunker

Mon.Not.R.Astron.Soc.000,000–000 (0000) Printed21January2013 (MNLATEXstylefilev2.2)
VLT/XSHOOTER & Subaru/MOIRCS Spectroscopy of
α
HUDF-YD3: No Evidence for Lyman- Emission at
z = 8.55⋆
Andrew J. Bunker1†, Joseph Caruana1,2, Stephen M. Wilkins1, Elizabeth R. Stanway3,4,
Silvio Lorenzoni1, Mark Lacy5, Matt J. Jarvis1,6,7 & Samantha Hickey6
3
1 1Universityof Oxford, Department of Physics, DenysWilkinson Building, Keble Road, OX1 3RH, U.K.
0 2Leibniz-Institut fu¨rAstrophysik Potsdam, Ander Sternwarte 16, 14482 Potsdam, Germany
2 3H. H. WillsPhysics Laboratory, Tyndall Avenue, Bristol, BS8 1TL, U.K.
4Department of Physics, Universityof Warwick, GibbetHill Road, Coventry, CV4 7AL, U.K.
n 5NRAO, 520 Edgemont Road, Charlottesville, VA 22903, USA
a 6Centre for Astrophysics, Science & Technology Research Institute, University of Hertfordshire, Hatfield, HertsAL109AB, U.K .
J 7Physics Department, Universityof the Western Cape, Cape Town, 7535, South Africa
8
1

21January2013
]
O
C
ABSTRACT
.
h WepresentspectroscopicobservationswithVLT/XSHOOTERandSubaru/MOIRCS
p of a relatively bright Y-band drop-out galaxy in the Hubble Ultra Deep Field, first
- selected by Bunker et al. (2010), McLure et al. (2010) and Bouwens et al. (2010)
o to be a likely z ≈ 8 − 9 galaxy on the basis of its colours in the HST ACS and
r
t WFC3 images. This galaxy, HUDF.YD3 (also known as UDFy-38135539) has been
s
targettedfor VLT/SINFONI integralfield spectroscopyby Lehnertet al. (2010),who
a
[ published a candidate Lyman-α emission line at z = 8.55 from this source. In our

independent spectroscopy using two different infrared spectrographs (5 hours with
1
VLT/XSHOOTER and 11 hours with Subaru/MOIRCS) we are unable to reproduce
v
this line. We do not detect any emission line at the spectral and spatial location
7
reportedinLehnertetal.(2010),despitetheexpectedsignalinourcombinedMOIRCS
7
4 & XSHOOTER data being 5σ. The line emission also seems to be ruled out by the
4 faintness of this object in recently extremely deep F105W (Y-band) HST/WFC3
. imaging from HUDF12; the line would fall within this filter and such a galaxy should
1
have been detected at YAB = 28.6mag (∼ 20σ) rather than the marginal YAB ≈
0
30magobservedintheY-bandimage,>3timesfainterthanwouldbeexpectedifthe
3
emissionliewasreal.HenceitappearshighlyunlikelythatthereportedLyman-αline
1
: emissionatz >8isreal,meaningthatthehighest-redshiftsourcesforwhichLyman-α
v emission has been seen are at z =6.9−7.2. It is conceivable that Lyman-α does not
i
X escape galaxies at higher redshifts, where the Gunn-Peterson absorption renders the
Universe optically thick to this line. However,deeper spectroscopyona largersample
r
a of candidate z >7 galaxies will be needed to test this.
Key words: galaxies:evolution–galaxies:formation–galaxies:starburst–galaxies:
high-redshift – ultraviolet: galaxies
1 INTRODUCTION
Candidate galaxies within the first billion years, at red-
shifts z >6, are now being routinely identified through the
⋆ Basedonobservations collectedattheEuropeanOrganisation
Lyman-breaktechnique(e.g.,Stanwayetal.2003;Bunkeret
for Astronomical Research in the Southern Hemisphere, Chile,
al.2004;Bouwensetal.2006;Hickeyetal.2010; McLureet
aspartofprogramme086.A-0968(B), andbasedinpartondata
al. 2010). Other methods, such as gamma-ray burst follow-
collectedatSubaruTelescope,whichisoperatedbytheNational
AstronomicalObservatoryofJapan. up, have also yielded high-redshift galaxies, including one
† E-mail:[email protected] probably at z = 8.2 whose spectrum shows a continuum
(cid:13)c 0000RAS
2 Andrew J. Bunker, et al.
breakconsistentwithLyman-α(Tanviretal.2009).Forthis beingLyman-αemissionatz=8.55,closetothephotomet-
growing population of objects with a spectral break consis- ric redshift of z = 8.45 from McLure et al. (2010). If real,
tentwithz>6,properspectroscopicconfirmationisimpor- the emergence of Lyman-α emission well within the Gunn-
tant,ratherthanbroad-bandphotometryorputativebreaks Peterson epoch would have significant implications for the
in low S/N spectroscopy. size of H II regions around galaxies, and would mean that
ThemainfeaturewhichmightbedetectableisLyman- Lyman-α might still be a useful redshift indicator for very
αemission,resultingfromphotoionizationofHIIregionsby distant galaxies even at a time when most of the Universe
starformation.However,thediscoveryoftheGunn-Peterson is optically thick to this line. However, previous claims of
complete absorption trough below Lyman-α (Gunn & Pe- Lyman-α emission at similarly large redshift (e.g. Pello´ et
terson 1965, Scheuer 1965) in SDSS and UKIDSS QSOs at al. 2004; Chen, Lanzetta & Pascarelle 1999) have not sur-
redshiftsbeyondz 6.2(Beckeretal.2001;Fanetal.2001, vived critical re-analysis (e.g. Bremer et al. 2004; Weather-
≈
2006;Mortlocketal.2011)showsthattheUniverseisonav- ley, Warren & Babbedge 2004; Stern et al. 2000). In this
erage optically thick to this line at earlier times. This sug- paper we re-observe the galaxy HUDF.YD3 from Bunker
geststhatz 6liesattheendoftheEpochofReionization, etal.(2010)withVLT/XSHOOTERandSubaru/MOIRCS
≈
whosemid-pointmayhaveoccurredatz 11,accordingto spectroscopytoseeifwecanrepeatthedetectionofLyman-
≈
resultsfromWMAP(Dunkleyetal.2009).Ithasbeenspec- α at z =8.55 made byLehnert et al. (2010).
ulatedthatalargeenoughHIIbubblearoundagalaxymight The structure of this paper is as follows. We describe
render this line non-resonant when it encounters the neu- ourspectroscopicobservationsinSection2,andpresentthe
tral IGM, so Lyman-αmight possibly emerge after all even results of the spectroscopy and constraints from the HST
duringtheGunn-Petersonera.Whilespectroscopyhascon- imaginginSection3.OurconclusionsaregiveninSection4,
firmedi′ dropLyman-breakgalaxies at z 6(e.g. Bunker andthroughoutweadoptastandardΛCDMcosmologywith
et al.200−3; Stanwayet al.2004; Stanwayet≈al. 2007), spec- ΩM = 0.3, ΩΛ = 0.7 and H0 = 70kms−1Mpc−1. All mag-
troscopic follow-up of z >6 sources has had mixed success. nitudesare on theAB system.
Vanzellaetal.(2011)showingconvincinglineemissionfrom
two Lyman-break galaxies at z = 7.0 7.1, and with one
−
marginal z > 6.4 emission line out of 17 targets reported
by Stark et al. (2010) and another marginal detection (out 2 OBSERVATIONS AND DATA REDUCTION
of 7 targets) from Fontana et al. (2010; see also Penter-
2.1 Observations with VLT/XSHOOTER
icci et al. 2011). More recently, Schenker et al. (2012) tar-
getted 19 Lyman-break galaxies with photometric redshifts WeobservedtheY-banddrop-outhigh-redshiftgalaxycan-
6.3<z <8.8,butfoundonlyoneobjectatz>7(z=7.045 didate HUDF-YD3 using the XSHOOTER spectrograph
withanothermoremarginalcandidateatz =6.905).Asim- (D’Odorico et al. 2006) on the ESO VLT-UT2 (Kueyen)
ilarsurveybyCaruanaetal.(2012)failedtodetectLyman- as part of programme 086.A-0968(B) (PI: A. Bunker).
α from any z-band or Y-band drop-outs at z >7. Another XSHOOTER is an echelle spectrograph, with UV, visible
way to isolate emission lines directly is narrow-band imag- and near-infrared channels obtaining near-continuous spec-
ing,andSuprime-CamontheSubarutelescopehasrevealed troscopy from 0.3µm to 2.48µm. We will focus here on the
a Lyman-alpha emitter spectroscopically confirmed to be near-infrared spectroscopy around 1.12µm, at the location
at z = 6.96 (Iye et al. 2006) with another three possible oftheemissionlineclaimedbyLehnertetal.(2010)intheir
candidates(Otaetal.2010),Thez=6.96sourcewassubse- ESO/VLTSINFONIspectroscopy.
quently observed by Ono et al. (2012), who also confirmed The main target, HUDF.YD3, has a position
twoz-dropgalaxiesaz =6.844andz=7.213withLyman- RA=03:32:38.135, Dec.= 27:45:54.03 (J2000), with co-
αemission.Anothernarrow-band-selectedLyman-αemitter ordinates from Lorenzoni −et al. (2011). We set the posi-
has recently been confirmed at z = 7.215 (Shibuya et al. tion angle of the 11′′-long XSHOOTER slit to 54.0degrees
2012). East of North. We set the central coordinates to be
There has only been one recent claim of line emis- RA=03:32:38.086, Dec.= 27:45:54.71 (J2000), such that
sion beyond z 7.2, despite the large number of Lyman- HUDF.YD3lay1′′awayal−ongtheslitlongaxis.Wedithered
≈
break candidates now known at these redshifts. Lehnert et theobservationsinanABBAsequenceatpositions+3′′ and
al. (2010) presented a VLT/SINFONI spectrum of one of 3′′ from the central coordinates along the slit long axis
the brightest Y-drops in the WFC3 imaging of the Hub- (−i.e. a ‘chop’ size of 6′′), so that the expected position of
ble Ultra Deep Field, which had previously been indepen- HUDF.YD3 should be +4′′ above the slit centre in the ‘A’
dently selected on the basis of its broad-band ACS and position,and 2′′ inthe‘B’position.Toacquirethetarget,
WFC3photometrybythreeindependentgroups(thegalaxy we first peake−d-up on a bright star 76.1′′ East and 10.6′′
HUDF-YD3 in the catalogue of Bunker et al. 2010, object South of the desired central pointing, then did a blind off-
1721 in McLure et al. 20101, and galaxy UDFy-38135539 set. ESO guarantee an accuracy of < 0′.′1 for an offset of
in Bouwens et al. 20102). The Lehnert et al. (2010) spec- this size, providing the guide star remains the same (which
trum shows a 6σ line at 11616˚A which is consistent with wasthecase),meaningthatthepositionaluncertaintyisless
′′
than10percentoftheslit widthused(1.2)–wenotethat
ourblindoffset of1.′3islessthanthat of1.′5usedbyLehn-
1 The naming of this galaxy changes to HUDF2003 in McLure ert et al. (2010). The XSHOOTER slit width is also much
′′
greater than the limit of < 0.4 set on any positional offset
etal.(2011).
2 Wenotethatinasubsequentpaper(Bouwensetal.2011),this betweenthecontinuumpositionandthatoftheclaimedline
galaxyhasadifferentidentification number,UDFy-38125539. emission from Lehnert et al. (2010).
(cid:13)c 0000RAS,MNRAS000,000–000
Spectroscopy of HUDF-YD3 3
TheXSHOOTERobservationswereconductedin6ob- ing several exposures the noise (normalized to unit time)
servingblocks,eachof1hourduration(49minofwhichwas decreased as √time as expected. The wavelength and spa-
on-source) and consisting of a single ABBA sequence with tial position of each pixel in the two-dimensional spectrum
threeexposuresofthenear-IRarmofduration245sateach wasdeterminedfromtheskylinesintheactualdataandthe
A orB position. Theobservations were taken on thenights arc line calibration spectra taken through a pinhole mask.
of UT 2010 December 27, 29, 30 & 31, with two observ- The spectrograph setup seemed very stable between differ-
ing blocks taken on the nights of UT 2010 December 29 ent nights of observation, with shifts of only 0.2 pixels
≈
& 30 and single observing blocks on the other nights. Ob- between nights. Residual skyline emission was removed us-
servingconditionswerereportedtobeclear, andtheseeing ingthebackground task inIRAF.Theexpectedposition of
conditions were typically 0′.′5 0′.′6 FWHM (from DIMM Lyman-α at z = 8.55 appears at the red end of order 23
−
measurements taken at the time, and we checked this was (and at the blue end of order 22, but the throughput here
consistentwithobservationsofstandardstarstakenclosein islower). Thepipelineoptimally combinestheordersofthe
time to our observations). One of the two observing blocks echelle spectrum, but in our independent reduction we in-
takenonUT2010 December30hadsignificantly worsesee- spected both echelle orders separately. The depths quoted
′′
ing of 1.2 FWHM, and we reduced the full dataset twice, in Section 3 come from thedeepest spectrum, order 22.
withandwithoutthisbad-seeingblock.Thisdidnotappear We obtained a flux calibration from observations of
to have a significant impact on the final results. Our total spectrophotometric standard stars taken over UT 2010 De-
on-sourceexposuretimefor HUDF.YD3with XSHOOTER cember26–31,aroundthedateswhenourHUDF-YD3spec-
was4.9hours,with4.1hourstakeningoodseeingconditions trawere obtained. Webaseourfluxcalibration on observa-
′′ ′′
of 0.5 0.6. tions of the standard star LTT3218 on UT 2010 December
− ′′
From unblended spectral lines in the calibration arc 28takeningoodseeingof0.6,whichisaclosematchtothe
lamp spectra and in the sky spectra we measured a spec- seeinginourspectroscopy ofHUDF.YD3.Wehavechecked
tral resolution of R = λ/∆λFWHM = 5000. We note that theshapeofthespectralresponseissimilar onothernights
our arc and sky lines fill the slit, so for compact sources wherethefluxstandardsLTT3218andFeige110weretaken
which do not fill the slit in the good seeing the resolution inworseseeing.Wenotethatalthoughtheregionofinterest
will be better than this (we expect this to be the case for around11616˚Aisclose toatmospheric absorption features,
HUDF.YD3). thedepthoftheabsorptionatthiswavelengthwasnotgreat
We reduced the XSHOOTER spectroscopy in two dif- andwasstablenight-to-night.Aroundourwavelengthofin-
ferentways.Weinitially usedtheESOpipeline(Modigliani terest,1countinasingle245sintegration correspondstoa
et al. 2010), which used the two-dimensional arc spectra line fluxof 3.4 10−19ergcm−2s−1.
×
through a pinhole mask to rectify the spectra both spa-
tially and spectrally (the echelle spectra exhibited signifi-
2.2 Observations with Subaru/MOIRCS
cant spatial curvature and a non-linear wavelength scale),
mapping on to a final output spectral scale of 1˚Apix−1 We observed the HUDF with slitmask spectroscopy in the
(from an original scale of about 0.5˚Apix−1 at wavelengths near-infrared using the MOIRCS instrument (Suzuki et al.
close to 11616˚A),and a spatial scale of 0′.′21 (from an orig- 2008; Ichikawa et al. 2006) on Subaru. MOIRCS was used
′′
inal scale of 0.24). The pipeline applied a flat-field, identi- inslitmaskmode,whichusestwodetectorswithacombined
fied and masked cosmic ray strikes using the algorithm of field of view of 7′ 4′, although there is vignetting beyond
van Dokkum(2001), differenced thetwodither positions to a diameter of 6′ f×rom the field centre. Unfortunately a fil-
remove the sky to first order, and combined the different ter wheel issue meant that one of the two detectors was
echelle orders together into a continuous spectrum (taking unusable, so we ensured that all our priority HUDF tar-
into account the different throughputs in different overlap- gets were placed in theother half of theslitmask. Accurate
pingechelleorders)beforespatiallyregisteringandcombin- alignment of theslitmask was achieved bycentering 5stars
ingthedatatakenatthetwoditherpositions,andremoving within 3′.′5-wide boxes to an accuracy of 0′.′1. One of the
any residual sky background. slits was used to target HUDF.YD3, and≈this slit was 4′.′5
We note that the ESO pipeline interpolates the data in length, with the long axis of the slit (the Position Angle
onto a uniform grid, which has the effect of correlating the ofthemask) set to+57degrees East ofNorth.Weobserved
noise (making the measured noise an underestimate of the themask with individual integrations of 1200s, moving the
′′ ′′
truenoise), and also potentially spreading theeffect of cos- telescopealongtheslitaxisbyasmalldithersizeof2.0–2.5
mic ray strikes and hot pixels around neighbouring pixels. inanABABABsequencetoenablebackgroundsubtraction.
Hence, we also did our own independent reduction of the We observed the HUDF mask on U.T. 2010 October 21 &
′′
XSHOOTERspectroscopy,wherewedidnotinterpolatethe 22, with a slit width of 1.0, and using the zJ500 grism.
data,keepingeachpixelstatistically independent.Thedata This instrument set-up has a spatial scale of 0′.′117pix−1
wereflat-fieldedusinghalogen lamp spectra(thathad been andaspectralscaleof5.57˚Apix−1.Theresolvingpowerfor
normalized by division by the spectral shape of the lamp), objects which fill the slit is R=λ/∆λFWHM =300 (deter-
and multiple exposures at each dither position were aver- minedfromThorium-Argonarclines),butthetypicalseeing
aged usingtheIRAFtaskimcombine, usingaPoisson noise was 0′.′5 FWHM so for unresolved sources (such at most of
modeltoreject cosmic ray strikes.Thetwoditherpositions the high-redshift galaxies targetted) the resolving power is
were then combined, with known hot pixels masked. The R=500.OnU.T.2010October21weobtained8exposures
′′
measured noise in the reduced two-dimensional spectrum of 1200s, with a dither step of 2.5 (i.e. placing the target
was close to the expected Poisson noise from the sky back- at +1′.′25 and 1′.′25 above and below the slit centre). On
− ′′
ground, dark current and readout noise, and when combin- U.T. 2010 October 22 we reduced the dither step to 2.0,
(cid:13)c 0000RAS,MNRAS000,000–000
4 Andrew J. Bunker, et al.
giventhegoodseeing,andobtainedanother12exposuresof are unresampled, we measure the flux total flux in 30 inde-
1200sforatotalintegration timeof400min(6.67hours)in pendent pixels, and from the pipeline data (which involves
October2010.Weobservedthesameslitmask targetsagain interpolation) the fluxis measured over 20 pixels.
with Subaru/MOIRCS on U.T. 2010 December 07, obtain- Wedetectnosignificantlineemission–wemeasurethe
ing 12 exposures of 1200s (a total of 4hours) with a dither fluxinouraperturetobe( 0.45 1.2) 10−18ergscm−2s−1,
′′ − ± ×
size of 2.0. To take full advantage of the good seeing at wheretheerroristhemeasured1σ noise.Wealsomovethe
Subaru (which again was 0′.′5 for the December2010 obser- aperture by 2pixels in x and y in a 3 3 grid to bracket
± ×
vations) we used a new mask design with the same objects the maximum uncertainty in the position of the Lehnert
′′ ′′
targetted but with the slit width reduced to 0.7, instead et al. (2010) Lyman-α emission (< 0.4), and we have no
of 1′.′0 used in October 2010, achieving a resolving power detection of line emission at any of these locations. Our
R = 500. The narrower slits reduced the sky background, measured noise is consistent with the online ESO Expo-
while still capturing most of the flux from the unresolved sure Time Calculator for XSHOOTER. We note that the
galaxies, significantly improving our sensitivity at the ex- Lehnert et al. (2010) line flux would be detected at the
pectedLyman-αwavelength,11616˚A(whichisclosetoOH 5σ level if all the line emission fell within our aperture. In
sky lines). order to quantify the expected flux, corrected for aperture
We reduced the MOIRCS data using standard tech- losses, we created artificial emission lines to add in at this
niques in IRAF, treating the October 2010 and December spatial and spectral location, as shown in Figure 1. From
2010separatelyduetothedifferentslitwidths.Theaverage theHST/WFC3imaging,HUDF.YD3shouldbeunresolved
′′ ′′
ofmanydarkcurrentswassubtractedfromeachframe,and in our 0.5-0.6 FWHM seeing. While it is conceivable that
a flat field applied (obtained from dome flats, normalized resonantly-scatteredLyman-αlineemissionmaycomefrom
bythespectrumofthelamp).Wethencombinedseparately a larger halo than the stellar UV continuum (e.g., Bunker,
all the data frames in the A position of the dither, using Moustakas & Davis 2000; Steidel et al. 2011), the emis-
ccdclip in imcombine to reject cosmic rays given the pa- sion line reported in HUDF.YD3 by Lehnert et al. (2010)
rametersofthedetector(gainof3.3e−count−1 andreadout iscompact spatially (unresolved in their0′.′6seeing). Hence
noise of 29e−pix−1). The same was done for the B posi- we adopt a Gaussian profile for the spatial extent with a
tions, and the combined B frame was subtracted from the FWHM of 0′.′6. For the spectral direction, we also adopt a
combined A frame to remove the sky background to first Gaussian profile for the fake sources, and consider two sce-
order. This resulted in a frame where we expect a positive nariosforthevelocitywidth.Wenotethattheemissionline
signal from a source at position A, and a negative signal at inLehnertetal.(2010)isunresolvedormarginally resolved
positionB(offsetalongtheslitbytheditherstep).Wethen (withaFWHMof 9.2˚A,only1σ larger thantheresolution
shifted and combined these signals, and residual sky emis- ofSINFONIwhichhasR=1580).Ourfirstscenariohasthe
sion was subtracted through polynomial fits along the slit source spectrally unresolved by XSHOOTER, which has a
length. higherresolving powerof R=5000 (so∆λFWHM =2.3˚A).
Flux calibration was achieved through observation of Inthiscase,ourphotometricaperturewouldcapture87per
theA0star HIP116886, and checkedagainst thefluxof the cent of the line flux, and we would expect a line with the
alignment stars of known J-band magnitude seen through sametotalfluxasintheLehnertetal.(2010)tobedetected
the 3′.′5-wide alignment boxes in the data frames. Around at4.5σ.Thesecondscenariotakesthereported(marginally-
11616˚A (the wavelength of interest), 1 count in an indi- resolved)spectralwidthof9.2˚Afrom Lehnertetal.(2010),
vidual 1200s exposure corresponds to a line flux of 5.6 deconvolves this with the SINFONI resolution to obtain
10−20ergcm−2s−1. × an intrinsic line width of 5.5˚A FWHM (140kms−1), then
convolve this with our spectral resolution for XSHOOTER
to obtain an observed line width of 6˚A FWHM. For this
broaderline,ourphotometricaperturecaptures66percent
3 DISCUSSION of the line flux, and we would expect a line with the same
total flux as in the Lehnert et al. (2010) to be detected at
3.1 Upper Limits on the Lyman-α Flux at z =8.55
3.5σ. Our XSHOOTER spectroscopy appears to rule out
from VLT/XSHOOTER
the existence of the Lyman-α line reported by Lehnert et
WemeasuretheobservedfluxatthelocationofHUDF.YD3 al.(2010) atthe3.5 4.5σ level,dependingonthevelocity
−
in the XSHOOTER long-slit spectrum, 1arcsec above width of theline.
(North-East of) the slit centre, and at the expected wave-
length of Lyman-α from Lehnert et al. (2010), λvac =
3.2 Upper Limits on the Lyman-α Flux at z=8.55
11615.6˚A(λair =11612.4˚A).Wedetectnosignofanemis-
from Subaru/MOIRCS
sion at this location. We perform spectrophotometry using
a square aperture, of extent 5˚A (10 pixels across in the For theSubaru/MOIRCSdata, we used an aperture of size
wavelength domain for our own reduction of the data, and 5 5pixels centred on the expected position of Lyman-α,
5pixelsinthepipelinereduction),whichismorethantwice co×rresponding to 0′.′6 28˚A, which is slightly larger than
×
aslargeasthewidthofaspectrallyunresolvedline.Forthe a resolution element. The 1σ noise within this aperture
spatial extent of our aperture, we adopt 3 pixels (0′.′72) for was measured to be 2.1 10−18ergcm−2s−1 for the De-
our reduction, and 4 pixels (0′.′84) for the pipeline reduc- cember 2010 observations×, and 2.4 10−18ergcm−2s−1 for
×
tion (theXSHOOTERpipelineresamples theoriginal pixel theOctober 2010 observations (which had higher noise due
scale slightly), which is marginally larger than the size of to the wider slit used and hence more sky emission). For
′′
the seeing disk. Hence in our reduction, where the pixels the 0.5 seeing and a spectrally unresolved line (where the
(cid:13)c 0000RAS,MNRAS000,000–000
Spectroscopy of HUDF-YD3 5
(the maximum positional uncertainty for the Lyman-α line
given by Lehnert et al. 2010). Both our VLT/XSHOOTER
andSubaru/MOIRCSspectroscopy yield consistent results:
we see no emission line at λvac = 11615.6˚A at the posi-
tion of HUDF.YD3, whereas if the flux reported by Lehn-
ert et al. (2010) is accurate we should have seen a signal
at 3.5 4.5σ with XSHOOTER and 2.7σ with MOIRCS.
−
Combiningtheresultsfromtwodifferentspectrographswith
inverse-varianceweighting,theLehnertetal.(2010)lineflux
is ruled out at the5σ level.
3.3 HST Photometry
OurVLT/XSHOOTERandSubaru/MOIRCSspectroscopy
ofHUDF.YD3strongly suggests thatthereisnolineat the
wavelength and line flux claimed by Lehnert et al. (2010)
on the basis of their VLT/SINFONI spectroscopy. We now
briefly consider whether the Lehnert et al. (2010) emis-
sion line would have been consistent with the HST/WFC3
broad-band photometry of this object reported by several
groups (Bunker et al. 2010; Bouwens et al. 2010; McLure
et al. 2010; Lorenzoni et al. 2011). The first WFC3 obser-
Figure 1. The pipeline-calibrated XSHOOTER spectrum, with vations of the HUDF taken as part of the programme GO-
thelocationofHUDF.YD3(1arcsecabovethecentreofthelong
11563 (HUDF09, PI: G. Illingworth) used the F105W (“Y-
slit) and the expected wavelength of the Lyman-α emission re-
band”),F125W(“J-band”)andF160W(“H-band”)filters.
portedbyLehnertetal.(2010)markedwithawhitecircle.Wave-
length increases from left to right, and we show the 50˚A either Anemissionlineat11615.6˚AwouldlieentirelywithintheY-
sideof 11616˚A, and the vertical axis is the 4.4arcsec covered in band(andalsowithinthewideJ-band),intheareaofpeak
bothnodpositionsoftheXSHOOTERslit.Fromtoptobottom: transmission of the sharp-sided Y-filter. If we take the line
(a)thepipeline-reduceddata;(b)thepipeline-reduceddatacon- fluxof 6.1 10−18ergcm−2s−1, then thiswould beequiva-
volvedwithaGaussianofσ=1pixel(1˚A,0′.′21).(c)afakesource lent to an o×bserved broad-band magnitude of YAB =28.89.
withthesamelineflux(6×10−18ergcm−2s−1)andwavelength ThereshouldalsobeacontributionfromtheUV-continuum
astheLehnertetal.(2010)lineaddedintotheframe.Weassume photons long-ward of Lyman-α (assuming near-total ab-
a spatially and spectrally unresolved source, with FWHM=0′.′6 sorption by the Lyman-α forest at shorter wavelengths).
spatiallyandFWHM=2.3˚Aspectrally. Theresultingframehas
Only 20 per cent of the Y-band filter transmission would
beensmoothedwithaGaussianwithσ=1pixel.(d)afakesource
lie at wavelengths above Lyman-α at the claimed redshift
withthesamelinefluxandwavelengthastheLehnertetal.(2010)
lineaddedintotheframe,withabroaderFWHM=5˚Aandagain of z = 8.55 (Lehnert et al. 2010), which would imply a
broad-band magnitude from the claimed line and contin-
unresolved spatially. The three vertical lines of higher noise are
duetonightskyemissionlines. uum of YAB = 28.57. In calculating the UV flux density
we use the measured HST/WFC3 broad-band magnitudes
ofJAB =28.18 0.13andHAB =28.10 0.13(Lorenzoniet
resolution is 600kms−1), such an aperture encloses 68 per al.2011),anda±doptarest-UVspectral±slopeoffλ λ−2.0,
cent of the total flux. Hence we would expect an emis- consistentwiththeHST/WFC3coloursafterwecor∝rectthe
sion line of the flux and wavelength reported by Lehn- J-band for the small fraction of flux within this filter that
ert et al. (2010) to be present at the 2.7σ level in our would fall below Lyman-α (a correction of 0.15mag, com-
total Subaru/MOIRCS spectrum, with most of the sensi- parable to the measurement error on the magnitudes). We
tivity coming from the December 2010 data using a nar- notethat HUDF.YD3has amagnitude fainter than the2σ
′′
rower slit (where such a line should be present at the 2.0σ limiting magnitude of YAB(2σ) = 29.65 in a 0.6-diameter
level). However, in both sets of MOIRCS observations this aperture for the HUDF09 data, and is formally undetected
line is undetected, with a total aperture-corrected flux of intheHUDF09WFC3imaging(Bunkeretal.2010;McLure
(1.6 3.1) 10−18ergcm−2s−1 for the deeper December et al. 2010; Bouwens et al. 2010; Lorenzoni et al. 2011).
2010±, and a×total flux of ( 0.1 2.3) 10−18ergcm−2s−1 The first WFC3 imaging with the F105W filter was 14
− ± ×
whencombiningalltheMOIRCSobservationstogetherfrom orbits (with another 4 orbits compromised by cosmic ray
allthreenights(usinginverse-varianceweighting).Although persistence), and since then this field has been extensively
the MOIRCS spectrum is less deep than our XSHOOTER targetted for further imaging with this filter as part of the
spectrum (on account of the lower spectral resolution of HUDF12programme(Ellisetal.2013)increasingthedepth
MOIRCS), the MOIRCS spectrum still is useful because to 100 orbits in total. In this deeper data, McLure et al.
′′
we are very confident of slit position, at the 0.1 level, due (2013)andSchenkeretal.(2013)reportafaintdetectionof
to the number of alignment stars used to position the slit- a corresponding object (labelled UDF12-3813-5540 in their
mask.FortheXSHOOTERspectrum(andindeedtheLehn- catalogues)ofYAB =30.1 0.2, closetothe5σ limit(using
′′ ±
ert et al. SINFONI spectrum) a blind offset was performed anapertureof0.4diameter,althoughapparentlytheyhave
from a nearby star, which does introduce some uncertainty notappliedanaperturecorrectiontothe 70percentofflux
′′ ≈
– although the tolerance is supposed to be less than 0.4 enclosed, sothetotal magnitudewill be 0.3mag brighter,
≈
(cid:13)c 0000RAS,MNRAS000,000–000
6 Andrew J. Bunker, et al.
YAB = 29.8). This is a factor of > 3 times fainter than eree(MasanoriIye)forusefulcommentsonthismanuscript.
theexpectedmagnitude of YAB =28.57 iftheemission line We are indebted to Ichi Tanaka of the Subaru Observa-
flux reported by Lehnert et al. (2010) was real and due to tory, NAOJ, for his invaluable assistance in designing the
Lyman-αfrom a Lyman-break galaxy at z=8.55. MOIRCS slitmask and during the observations. Based on
Hencethebroad-bandphotometryintheY-bandisin- observationsmadewiththeNASA/ESAHubbleSpaceTele-
consistent with the Lehnert et al. (2010) line flux and red- scope associated with programmes #GO-11563 & #GO-
shiftbeingreal–ifthelinewasreal,thenthedeepHUDF12 12498, obtained from the Data Archive at the Space Tele-
HST/WFC3 Y-band should have obtained a clear 15σ de- scope Science Institute, which is operated by the Associa-
tection, whereas the actual result was close to the 5σ lim- tion of Universities for Research in Astronomy, Inc., under
iting magnitude. The broad-band photometry alone seems NASA contract NAS 5-26555. MJJ acknowledges the sup-
to rule out the claimed line flux from Lehnert et al. (2010) portofaRCUKfellowship.JCandSLaresupportedbythe
at high significance. Consistency with theLehnertet al. re- Marie Curie Initial Training Network ELIXIR of the Euro-
sult would require both that the broadband flux is greatly pean Commission undercontract PITN-GA-2008-214227.
underestimated due to noise and that the line flux is over-
estimated, a coincidence which is statistically unlikely.
REFERENCES
4 CONCLUSIONS BeckerR.H.etal.,2001,AJ,122,2850
BouwensR.J.IllingworthG.D.BlakesleeJ.P.;FranxM.,2006,
We have presented spectroscopic observations with
ApJ,653,53
VLT/XSHOOTER and Subaru/MOIRCS of a relatively
Bouwens,R.J.,etal.2010,ApJ,709,L133
bright Y-band drop-out galaxy in the Hubble Ultra Deep Bouwens,R.J.,etal.2011,ApJ,737,90
Field, first selected by Bunker et al. (2010), McLure et al. Bremer,M.N.,Jensen,J.B.,Lehnert,M.D.,Schreiber,N.M.F.,
(2010)andOeschetal.(2010)tobealikelyz 8 9galaxy &Douglas,L.2004,ApJ,615,L1
on the basis of its colours in the HST ACS≈an−d WFC3 Bunker,A.J.,Moustakas,L.A.,&Davis,M.2000,ApJ,531,95
images. This galaxy, HUDF.YD3 (from the catalogues of Bunker A. J., Stanway E. R., Ellis R. S., McMahon R. G., Mc-
Bunker et al. 2010 and Lorenzoni et al. 2011) has been Carthy,P.J.2003, MNRAS,342,L47
BunkerA.J.,StanwayE.R.,EllisR.S.,McMahonR.G.,2004,
targetted for VLT/SINFONI integral field spectroscopy
MNRAS,355,374
by Lehnert et al. (2010), who published a candidate
Bunker,A.J.,etal.2010,MNRAS,409,855
Lyman-α emission line at z = 8.55 from this source. In
Caruana,J.,Bunker,A.J.,Wilkins,S.M,Stanway, E.R.,Lacy,
our independent spectroscopy using two different infrared
M., Jarvis, M. J., Lorenzoni, S., Hickey, S. 2012, MNRAS,
spectrographs we are unable to reproduce this line. In 427,3055
our 5hour spectrum with XSHOOTER with a moder- Chen, H.-W., Lanzetta, K. M., & Pascarelle, S. 1999, Nat, 398,
ately high resolving power of R = 5000, the line flux of 586
6.1 10−18ergcm−2s−1 reported by Lehnert et al. (2010) D’Odorico,S.,etal.2006,Proc.SPIE,6269,
×
should have resulted in a detection at the 3.5 4.5σ level Dunkley,J.,etal.2009,ApJS,180,306
(depending on the velocity width of the line),−and in our Ellis,R.etal.,2013, arXiv:1211:6804
Fan,X.,etal.2001, AJ,122,2833
low-resolution (R=500) 10.7hour MOIRCS spectrum this
Fan,X.,etal.2006, AJ,132,117
line flux would correspond to a 2.7σ signal. We do not
Fontana, A.,etal.2010,ApJ,725,L205
detectanyemission lineatthespectralandspatiallocation
Gunn,J.E.,&Peterson,B.A.1965,ApJ,142,1633
reported in Lehnert et al. (2010), despite the expected
Hickey, S., Bunker, A., Jarvis, M. J., Chiu, K., & Bonfield, D.
signal in our combined MOIRCS & XSHOOTER data
2010,MNRAS,404,212
being 5σ. The line emission also seems to be ruled out Ichikawa,T.,etal.2006, Proc.ofSPIE,6269,38
by the faintness of this object in the very deep Y-band Iye,M.,etal.2006,Nat,443,186
HST/WFC3 image (YAB = 30.1); the line would fall Lehnert,M.D.,etal.2010,Nat,467,940
within this filter, and the corresponding magnitude of Lorenzoni, S., Bunker, A. J., Wilkins, S. M., Stanway, E. R.,
YAB = 28.57 should have been detected at 20σ rather Jarvis,M.J.,&Caruana,J.2011,MNRAS,414,1455
than the marginal 5σ observed. Hence it a≈ppears highly McLure, R. J.,Dunlop, J. S.,Cirasuolo, M.,Koekemoer, A.M.,
Sabbi, E., Stark, D. P., Targett, T. A., & Ellis, R. S. 2010,
unlikely that the reported Lyman-α line emission at z > 8
MNRAS,403,960
is real, meaning that the highest-redshift sources for which
McLure,R.J.,etal.2011, MNRAS,418,2074
Lyman-α emission has been seen are at z = 6.96 7.2.
− McLure,R.J.,etal.2013, arXiv:1212:5222
It is conceivable that Lyman-α does not escape galaxies
Modigliani,A.,etal.2010,Proc.SPIE,7737,
at higher redshifts, where the Gunn-Peterson absorption Mortlock,D.J.,etal.2011, Nat,474,616
renders the Universe optically thick to this line. However, Ono,Y.,etal.2012,ApJ,744,83
deeper spectroscopy on a larger sample of candidate z > 7 Ota,K.,etal.2010, ApJ,722,803
galaxies will be needed to test this. Pello´, R., Schaerer, D., Richard, J., Le Borgne, J.-F., & Kneib,
J.-P.2004,A&A,416,L35
Pentericci,L.,etal.2011,ApJ,743,132
Acknowledgements Schenker, M. A., Stark, D. P., Ellis, R. S., Robertson, B. E.,
Dunlsop,J.S.,McLure,R.J.,Kneib,J.-P.,Richard,J.2012,
We gratefully acknowledge Chris Willott, Daniel Stark and ApJ,744,179
the NIRSpec Instrument Science Team for useful discus- Schenker,M.A.,etal.2013,arXiv:1212:4819
sion, and thank Richard Ellis, Matt Lehnert and the ref- Scheuer,P.A.G.,1965, Nature,207,963
(cid:13)c 0000RAS,MNRAS000,000–000
Spectroscopy of HUDF-YD3 7
Shibuya, T., Kashikawa, N., Ota, K., Iye, M., Ouchi, M., Furu-
sawa,H.,Shimasaku,K.,Hattori,T.2012, ApJ,752,114
Stanway E. R., Bunker A. J., McMahon R. G. 2003, MNRAS,
342,439
Stanway,E.R.,Bunker,A.J.,McMahon,R.G.,Ellis,R.S.,Treu,
T.,McCarthy,P.J.2004, ApJ,607,704
Stanway, E.R.,etal.2007,MNRAS,376,727
Stark,D.P.,Ellis,R.S.,Chiu,K.,Ouchi,M.,&Bunker,A.2010,
MNRAS,408,1628
Steidel,C.C.,Bogosavljevi´c,M.,Shapley,A.E.,Kollmeier,J.A.,
Reddy,N.A.,Erb,D.K.,&Pettini,M.2011,ApJ,736,160
Stern, D.,Eisenhardt, P.,Spinrad, H.,Dawson, S.,van Breugel,
W.,Dey,A.,deVries,W.,&Stanford, S.A.2000, Nat,408,
560
Suzuki,R.,etal.2008,PASJ,60,1347
Tanvir,N.R.,etal.2009,Nat,461,1254
Vanzella,E.,etal.2011,ApJ,730,L35
vanDokkum,P.G.2001,PASP,113,1420
Weatherley,S.J.,Warren,S.J.,&Babbedge,T.S.R.2004,A&A,
428,L29
ThispaperhasbeentypesetfromaTEX/LATEXfileprepared
by theauthor.
(cid:13)c 0000RAS,MNRAS000,000–000