The Ages and Masses of Lyman Alpha Galaxies at Redshift $z\sim 4.5$

0.2 MB·English

by  Steven L. Finkelstein

Draftversion February5,2008
PreprinttypesetusingLATEXstyleemulateapjv.12/14/05
THE AGES AND MASSES OF LYMAN ALPHA GALAXIES AT Z ∼ 4.51
Steven L. Finkelstein2, James E. Rhoads, Sangeeta Malhotra3, Norbert Pirzkal4 & Junxian Wang5
1 ObservationsreportedherewereobtainedattheMMTObservatory,ajointfacilityoftheUniversityofArizonaandtheSmithsonian
Institution
2 DepartmentofPhysics,ArizonaState University,Tempe,AZ85287
3 SchoolofEarthandSpaceExploration,ArizonaStateUniversity,Tempe,AZ85287
4 SpaceTelescopeScienceInstitute, 3700SanMartinDrive,Baltimore,MD21218and
5 UniversityofScienceandTechnology ofChina,Hefei,Anhui230026, China
Draft versionFebruary 5, 2008
ABSTRACT
7 We examine the stellar populations of a sample of 98 z ∼ 4.5 Lyman alpha emitting galaxies using
0 their broadband colors derived from deep photometry at the MMT. These galaxies were selected by
0 narrowband excess from the Large Area Lyman Alpha survey. Twenty-two galaxies are detected in
2 two or more of our MMT filters (g’, r’, i’ and z’). By comparing broad and narrowband colors of
these galaxies to synthetic colors from stellar population models, we determine their ages and stellar
n
a masses. The highest equivalent width objects have an average age of 4 Myr, consistent with ongoing
J star formation. The lowestEW objects show an age of 40 - 200 Myr, consistent with the expectation
5 that larger numbers of stars are causing low EWs. We found masses ranging from 2×107 M⊙ for
the youngest objects in the sample to 2×109 M⊙ for the oldest. It is possible that dust effects could

2 produce large equivalent widths even in older populations by allowing the Lyman alpha photons to
v escape, even while the continuum is extinguished, and we present models for this scenario also.
1 Subject headings: galaxies: fundamental parameters– galaxies: high-redshift– galaxies: ISM – galax-
1
ies: evolution
5
2
1 1. INTRODUCTION formationactivityinthesegalaxies(Partridge & Peebles
6 1967), and it could mean that these are some of the
There are two popular techniques for locating galax-
0 youngest galaxies in the early universe.
ies at high redshift: the Lyman Break technique
/ There are a few possible scenarios that could be cre-
h (Steidel et al. 1996) and narrowbandselection. The sec-
p ondentailsobservingagalaxyinanarrowbandfiltercon- ating this large EW. Strong Lyα emission could be pro-
- taining a redshifted emission line. This method is useful duced via star formation if the stellar photospheres are
o hotterthannormal,whichcouldhappeninlowmetallic-
to select highredshift galaxieswith strong Lyman alpha
r ity galaxies. It could also mean that these galaxies have
t (Lyα) emission lines. While the line emission proper-
s theirstellarmassdistributedviaatop-heavyinitialmass
a ties of Lyman Break galaxies (LBGs) are known, little
function (IMF). Both of these scenarios are possible in
: is known about the continuum properties of Lyα galax-
v primitive galaxies, which are thought to contain young
ies because they are so faint. Studying the continuum
i stars and little dust.
X properties of these Lyα galaxies at high-redshift can tell
ActivegalacticnucleialsoproducealargeLyαEW.In
us about their age, stellar mass, dust content and star
r an optical spectroscopic survey of z ∼ 4.5 Lyα galaxies,
a formation history.
Dawsonetal. (2004)foundlargeLyαequivalentwidths,
Wehaveusedthenarrowbandselectiontechniquetolo-
but narrow physical widths (∆v < 500 km s−1). They
cateandstudyLymanalphaemittinggalaxies(LAEs)at
also placed tight upper limits on the flux of accompany-
z∼4.5. Similarstudieshavebeendonebothinthisfield
ing high-ionization state emission lines (e.g., NV λ1240,
andinotherfields(e.g.,Rhoadsetal2000,2004;Rhoads
SiIVλ1398,CIVλ1549andHeIIλ1640),suggestingthat
&Malhotra2001;Malhotra&Rhoads2002;Cowie&Hu
the large Lyα EW is powered by star formation rather
1998; Hu et al 1998, 2002, 2004; Kudritzki et al 2000;
than AGN. Additionally, Type I (broad-lined) AGN are
Fynbo, Moller, & Thomsen 2001; Pentericci et al 2000;
ruled out because the width of the Lyα line in such an
Stiavelli et al 2001; Ouchi et al 2001, 2003, 2004; Fu-
objectisgreaterthanthewidthofthenarrowbandfilters
jita et al 2003; Shimasaku et al 2003, 2006; Kodaira et
which were used (see Section 2.3). While Type II AGN
al 2003; Ajiki et al 2004; Taniguchi et al 2005; Vene-
lines are narrower than our filters, they remain broader
mans et al 2002, 2004). In many of these studies, the
than the typical line width of our Lyα sample as de-
strength of the Lyman alpha (Lyα) line has been found
termined by followup spectroscopy (e.g., Rhoads et al.
to be much greater than that of a normal stellar pop-
2003). Malhotra et al. (2003) and Wang et al. (2004)
ulation (Kudritzki et al. 2000; Dawson et al. 2004).
searched for a correlation between LAEs and Type II
Malhotraand Rhoads (2002;hereafter MR02)found nu-
AGN using deep Chandra X-Ray Observatory images.
merousLAEswithrestframeequivalentwidths(EWs)>
101 Lyα emitters were known to lie in the two fields
200˚A. Assumingaconstantstarformationrate,theEW
observedwith Chandra, andnone ofthem weredetected
of a normal stellar population will asymptote toward 80
˚A by 108 years, down from a maximum value of ∼ 240 at the 3σ level (L2−8keV = 2.8 × 1042 ergs s−1). The
sources remained undetected when they were stacked,
˚A. These galaxies are of interest because it is believed
andtheyconcludedthatlessthan4.8%ofthe Lyαemit-
that this strong Lyα emission is a sign of ongoing star
2 Finkelstein et al.
terstheystudiedcouldbe possibleAGNs. Therefore,we sure time in the g′, r′, i′ and z′ filters were 4.33, 3.50,
are confident that there are at most a few AGN in our 4.78 and 5.33 hours respectively. Given that the seeing
Lyα galaxy sample. duringthe runwasrarelybelow0′.′8,wechoseto bin the
Anotherpossible scenarioinvolvesenhancementofthe pixels 2×2 to reduce data volume, resulting in a final
Lyα line in a clumpy medium. In this scenario, the con- pixel scale of ∼0′.′16 per pixel in our individual images.
tinuum light is attenuated by dust whereas the Lyα line
isnot,resultinginagreatlyenhancedobservedLyαEW 2.2. Data Reduction
(Neufeld 1991; Hansen & Oh 2006). This can happen if
The data were reduced in IRAF1 (Tody 1986, 1993),
the dustisprimarilyincold,neutralclouds,whereasthe
using the MSCRED (Valdes & Tody 1998; Valdes 1998)
inter-cloud medium is hot and mainly ionized. Because
and MEGARED (McLeod et al. 2006) reduction pack-
Lyαphotonsareresonantlyscattered,theyarepreferen-
ages. The data were first processed using the standard
tially absorbed at the surface. Thus it is highly unlikely
CCDreductionstepsdonebythetaskccdproc(overscan
that a Lyα photon will penetrate deep into one of these
correction,trim, bias subtraction and flat fielding). Bad
clouds. Theywillbeabsorbedandre-emittedrightatthe
pixels were flagged using the Megacam bad pixel masks
surface,effectivelyscatteringoffofthecloudsandspend-
included with the MEGARED package.
ingthemajorityoftheirtimeintheinter-cloudmedium. Fringing was present in the i′ and z′ band data, and
However, continuum photons are not resonantly scat-
needed to be removed. For this, we made object masks
tered, and if the covering factor of clouds is large, they
using the task objmasks2. We then took a set of sci-
willingeneralpassthroughtheinteriorofone. Thusthe
ence framesineachfilter,andcombinedthem to makea
continuum photons will suffer greater attenuation than
skyflatusingsflatcombine(usingthe objectmaskstoex-
the Lyα photons, effectively enhancing the Lyα EW.
cludeanyobjectsfromtheresultantimage). Thisskyflat
These Lyα galaxies are so faint compared to other
was then median smoothed on a scale of 150×150pixels
high-z objects that have been studied, that not a lot is (i′)and200×200pixels (z′). The smoothedimageswere
known about their continuum properties. Our primary
subtracted from the unsmoothed images, and the result
goalis to better constrainthe ages and stellar masses of
was a fringe frame for eachband. The fringing was then
these objects fromtheir continuum lightin orderto bet-
removedfromthescienceimagesusingthetaskrmfringe,
ter understand the driving force behind the strength of
which scales and subtracts the fringe frame. To make
the Lyα line. We have obtained deep broadband images
the illumination correction, skyflats were made, median
in the g′, r′, i′ and z′ bands for this purpose. Using the
smoothed 5×5 pixels, normalizedand divided out of the
colorsofthese galaxies,wecanstudy continuumproper-
science images.
ties of individual Lyα emitting galaxies at this redshift
The amplifiers in the reduced images were merged
forthefirsttime,allowingustoestimatetheirage,mass,
(from 72 to 36 extensions) using the MEGARED task
star formation rate and dust content. The next phase of
megamerge. The WCS written in the image header at
our project will be to look into the likelihood of dust
the telescope is systematically off by 0.5 degrees in rota-
causing the large Lyα EW, which we will begin to do
tion,soweusedthetaskfixmosaictoadjusttherotation
in this paper, as the colors of these galaxies might dis-
angle so that the WCS solution works better. The WCS
tinguish the cause of the large EW. Blue colors would
was then determined using megawcs, and the WCS dis-
indicate young stars with hot photospheres, while red
tortionterms were installed in the header using the task
colors would indicate dust quenching of the continuum,
zpn.
enhancing the Lyα EW. Observations, data reduction
The 36 chips were combined into a single extension
technique and sample selection are reported in Section
using the program SWarp3. We ran SWarp on each in-
2; results are presented in Section 3, and they are dis-
putimage,usingacorrespondingweightmapwhichgave
cussedinSection4. ConclusionsarepresentedinSection
zero weight to known bad pixels and cosmic ray hits.
5.
We located the cosmic rays using the method of Rhoads
2. DATAHANDLING (2000). We specified the center of the output images to
2.1. Observations bethe averagecenterofthe fiveditherpositionsweused
(02h04s56m, -05o01′01′′ J2000), which corresponded to
TheLargeAreaLymanAlpha(LALA)surveybeganin
an image size of 11988×10881pixels and a pixel scale of
1998usingthe4mMayalltelescopeattheKittPeakNa- 0′.′1587 per pixel. During this process, each image was
tionalObservatory(KPNO).Thefinalareaofthissurvey
resampled (using the Lanczos-3 6-tap filter) and inter-
was 0.72 deg2 in two fields, Boo¨tes and Cetus, centered
polated onto the new pixel grid. While this was done,
at 14h25m57s, +35o32′ (J2000) and 02h05m20s, -04o55′
SWarpdeterminedandsubtractedthe backgroundvalue
(J2000) respectively (Rhoads et al. 2000). The LALA
in each extension.
survey found large samples of Lyα emitting galaxies at
z = 4.5, 5.7 and 6.5, with a spectroscopic success rate 1IRAFisdistributedbytheNationalOpticalAstronomyObser-
of up to 70%. We observed the LALA Cetus Field for vatories(NOAO),whichisoperatedbytheAssociationofUniver-
three full nights in 2005 November, and again for four sities forResearch inAstronomy, Inc. (AURA) under cooperative
agreementwiththeNationalScienceFoundation.
1/4 nights in 2006 January, using the Megacam instru-
2 In order to prevent fringing from making it onto the object
ment (McLeod et al. 1998) at the Multiple Mirror Tele-
masks,wefirstmadea skyflat withoutobject masks, anddivided
scope (MMT). Megacam is a large mosaic CCD camera thisskyflatoutofthesciencedata. Wethengottheobjectmasks
with a 24’x24’ field of view, made up of 36 CCDs with fromthesequasi-fringeremovedimages. Therestoftheprocedure
2048x4608pixels. Each pixel is 0′.′08 on the sky. We ac- usedtheuntouched scienceframes.
3SWarpisaprogramthatresamplesandco-addstogetherFITS
quired deep broadbandimages in four Sloan Digital Sky
images,authoredbyEmmanuelBertin. ItisdistributedbyTerapix
Survey (SDSS) filters: g′, r′, i′ and z′. The total expo- at: http://terapix.iap.fr/soft/swarp.
Lyα Galaxies at z ∼ 4.5 3
An initial stack for each filter was made using SWarp Rhoads & Malhotra 2001,were as follows: (1) 5σ signif-
to average together the individual images. To make the icance detection in the narrowband: This is calculated
finalversionofthe stack,we neededto removethe satel- using an SExtractor aperture flux with the associated
lite trails which littered our images. To do this, we cre- flux error; flux/error ≥ 5. (2) 4σ significant excess of
ated a difference image between each individual input narrowbandflux: This was calculatedby taking narrow-
image and the initial stack. Using a thresholding tech- band flux - broadband flux, both calibrated in physi-
nique, we flagged the satellite trails in this image, and cal units, and the associated error (sqrt(error2 +
broad
combined those flags with the existing weight-maps. We error2 )) and demanding that flux difference / er-
narrow
ran SWarp a final time with this as the input weight ror ≥ 4. (3) Factor of ≥ 2 ratio between broad and
map, creating the final stack. We did this process for narrowbandfluxes,calibratedtothe sameunits. (4)No
each band, giving us four images, one final stack in each morethan2σsignificantfluxinthebluestfilterobserved
of the g′, r′, i′ and z′ bands. In an attempt to increase (B ).
w
the signal of our objects, we tried co-adding R and I
band images from the NOAO Deep Wide-Field Survey 2.4. Stellar Population Models
(NDWFS)(Jannuzi & Dey1999)toourr′ andi′ images, In order to study the properties of the galaxies in
however the additional signal did not make a significant our sample we compare them to stellar population mod-
difference in the results. els, using the stellar population modeling software by
To calculate the photometric zeropoint for our obser- Bruzual & Charlot (2003). With these models, we were
vations we took images of the standard stars SA95 190 able to choose a range of ages,metallicities and star for-
and SA95 193 during our run, where the zeropoint is mation rates for comparison with our sample. We chose
defined as the magnitude of an object with one count ages ranging from 106 - 109 years, metallicities from .02
in the image (for an integration time of 400s). These Z⊙ -Z⊙ andexponentiallydecayingstarformationrates
standardstarsarefromLandolt(1992),andweusedthe withacharacteristictime-scaleofτ =103 -2×109 years.
transformsfromJohnson-Morgan-CousinstoSDSSmag- We included dust via the Calzetti dust extinction law
nitudes from Fukugita et al. (1996). Averaging the re- (Calzetti et al. 1994), which is applicable to starburst
sultsfromthetwostarsforeachband,weobtainedthese galaxies,intherange: A =0-2. Lastly,weincluded
zeropoints: g′=33.13, r′=33.00, i′=32.58 and z′=31.43. intergalacticmedium(IG1M200)absorptionviatheprescrip-
Using these zeropoints with our data, all of our results tion of Madau (1995).
are in AB magnitudes4. TheBruzual&Charlotcode(BC03)outputthefluxof
agivenstellarpopulationinunitsofL⊙ ˚A−1. Inorderto
2.3. Lyman α Galaxy Selection go from flux in these units to bandpass averaged fluxes,
To extract the objects from each stack, we used the we used the method outlined by Papovich et al. (2001).
SExtractor package (Bertin & Arnouts 1996). For SEx- In short, we took the output from BC03, and converted
tractor to accurately estimate the errors in the flux, it it from fλ into fν (units of erg s−1 cm−2 Hz−1). This
needed to know the gain in the input images. To calcu- fluxwasthenmultipliedbythetransmissionfunctionfor
late this, we needed the mean and standarddeviation of agivenbandpass,andintegratedoverallfrequency. The
the background of the images. Because the background bandpass averaged flux hfνi is this result normalized to
was subtracted out in SWarp, we obtained the average the integral of the transmission function. AB magni-
background value in each band by averaging the values tudes (Oke & Gunn 1983) for the candidates were then
obtained from the pre-SWarpedimages, and adding this computed.
valueontothefinalstacks. WethenranSExtractorinthe In order to calculate colors which we could directly
two-imagemodeusinganinepixelaperture(2.′′32)with compare to our observations, it was necessary to add in
thenarrowbandimagesfromtheLALAsurvey,givingus emission line flux from the Lyα line, which appears in
catalogs of objects which were detected in both images. the r′ filter at z ∼ 4.5. While the BC03 software does
The method we have used to locate Lyα galaxies in- not calculate emission line strengths, we were able to
volves taking an image using a narrowband filter con- use one ofits many other output quantities: the number
taining the wavelengthfor Lyα ata certainredshift. We ofionizingphotons. WecalculatedtheLyαemissionline
usedthe catalogfromMR02,whichwasbasedonLALA strength(inunitsofL⊙ ˚A−1 tomatchthemodeloutput)
narrowband images and broad band B , R, and I data by using:
w
from the NDWFS (Jannuzi & Dey 1999). The narrow- hc 2
banddataconsistsoffiveoverlappingnarrowbandfilters, Lyα Linestrength= × ×nion (1)
eachwithaFWHM∼80˚A. Thecentralwavelengthsare λLyαL⊙∆λ 3
λλ6559(H0),6611(H4),6650(H8),6692(H12)and6730 where n is the number of ionizing photons, ∆λ is the
ion
(H16), giving a total redshift coverage 4.37 < z < 4.57. bin size of the wavelength array (1˚A), and the factor
Toensurethattherewasnooverlap,weonlyusedobjects of 2/3 represents the fraction of ionizing photons which
selected from the H0, H8 and H16 filters in our analy- willproduceLyαphotonswheninteractingwiththelocal
sis. Inorderto comparethe twodatasets,theMegacam interstellarmedium(ISM)underCaseBrecombination5.
data were registered and remapped onto the same scale Inordertomodeltheclumpydustscenario,weneededto
as the narrowband data using the IRAF tasks wcsmap ensurethat the Lyα line did notsuffer dustattenuation.
and geotran. To do this, the continuum flux was multiplied by the
Selection criteria for the MR02 catalog, following Calzettidust lawbefore we addedin the Lyα flux to the
4 For more details on our reduction and analysis process, see: 5 Thiscalculationfollowsthesimpleassumptionsthatnoioniz-
http://stevenf.asu.edu/Reduction.html ingphotons escape, andthatallLyαphotons escape.
4 Finkelstein et al.
spectrumatthecorrectwavelengthbin. Inthiscase,the withEWthishigharepossiblefromnormalstellarpopu-
continuum suffers dust attenuation while the Lyα line lations,theratioofhigh-lowEWsismuchgreaterthanit
does not. ought to be when we account for the 76 objects without
MMTdetections. Thisimpliesthatsomethingiscausing
3. RESULTS the EW to be higher than normal. One of our goals in
3.1. Lyα Galaxy Candidates this study is to investigate whether the cause is massive
starscreatingmoreLyαphotons,ordustsuppressingthe
Using the catalog described above, we have identified
continuum and thus enhancing the Lyα EW.
98 objects as Lyα galaxy candidates within the 24′×24′
Megacam field of view. We have classified 22 of these
objects as continuum detections on the basis that they
haveatleast2σ detectionsintwoofther′,i′ orz′ bands
(the other76objectswereundetectedatthis levelinthe
broadbanddata,butwestackedtheirfluxesforanalysis).
WehaveinourpossessionIMACSspectraofobjectsin
the LALA Cetus Field (Wang et al. 2007). These spec-
trahavebeenusedtoidentifytheredshiftoftheLyαline
in these galaxies. Out of the 98 total Lyα galaxy candi-
dates,28havebeenspectroscopicallyconfirmedasbeing
Lyα emitters at z ∼ 4.5. Seven of these confirmations
are among the 22 Megacam detections. Another seven
ofthe MegacamdetectionshaveIMACSspectrabuthad
nostrongLyαlineidentified,howeversomeofthesemay
show a line with deeper spectra.
3.2. Equivalent Width Distribution Fig. 1.— The distribution of rest-frame Lyα equivalent widths
fromour22detections. TheEWsrangefrom5–800˚A. Thesingle
We calculated the rest-frame equivalent widths of our objectwithEW< 20˚A islikelyaninterloper that has managed
22 detections using the ratio of the line flux to the con- tosatisfyourselectioncriteria. Anormalstellarpopulationhasan
tinuum flux via: EW that asymptotes toward ∼ 80 ˚A as its age goes to 108 years,
so something is causing the Lyα EW of many of these objects to
1−η behigher..
EW = ×(1+z)−1 (2)
∆ληBB − ∆λ1NB!
We notethatfourobjects amongourMMT detections
whereη isthe ratioofnarrowbandtobroadbandflux(in show observed equivalent width in excess of 230˚A. Such
units of f ), and the ∆λs are the filter widths (80˚A and highequivalentwidthsmightbeexplainedbyacombina-
ν
1220˚Aforthenarrowandbroadfiltersrespectively). The tion of extreme youth and/or low metallicity (e.g. Mal-
narrowbandfluxesaremeasuredfromtheLALAnarrow- hotra & Rhoads 2004), or might be produced by radia-
banddatadiscussedabove,andthebroadbandfluxesare tive transfer effects (Neufeld 1991, Hansen & Oh 2006).
measured from our MMT r′ data. However, two of these objects are consistent with 200˚A
Figure 1 shows the calculated rest-frame EWs of our equivalentwidthatthe1σlevel,andtheothertwoatthe
22continuumdetectedgalaxies. TherangeofEWsis5– 2σ level,so suchmechanisms are not strictly requiredto
800˚A6. Whilethe rangeofourEWdistributionpeaksat explain these objects. Stronger evidence for such effects
the low end of previous studies (i.e. most of our objects may be found among the 76 sources whose continuum
have rest-frame EW . 200˚A), we recognize that the 76 emission was too faint for our MMT data.
undetected objects would all lie at the high EW end of
this plot (in fact, they peak at EWwell over200˚A). Be- 3.3. Stacking Analysis
cause they are undetected, this means that their broad-
Our goal has been to be able to compare our objects
band fluxes were too faint to be detected in our survey.
tothemodelstoobtainestimatesofphysicalparameters
However, they still have very bright narrowband fluxes,
such as age and stellar mass, along with dust content.
indicating strong Lyα emission. This strong emission,
We divided our objects into six groups which we then
coupledwithfaintbroadbandr′ magnitudes,resultsina
stacked (in order to damp down the errors) to get com-
high Lyα EW.
positefluxeswhichwecouldcomparetothemodels. The
Frommodelsofnormalstellarpopulations(those with
groups are: non-detections, which consists of the 76 se-
a near-Solar metallicity and a Salpeter IMF), it was
lected objects which did not have a 2 σ detection in at
found that the maximum EW a galaxy could have is least two bands (out of r′, i′ and z′); low EW, which
∼ 240˚A at t ∼ 106 yr, with the EW approaching a
consists of the seven detected objects with EW of 20–
steady value of 80˚A at t ∼ 108 yr (MR02; Charlot & 40˚A; middle EW, which consists of the seven detected
Fall 1993). Our data show many objects with a rest- objectswithEWof45–100˚A;highEW,whichconsistsof
frame EW greater than 100˚A. While individual galaxies the sevendetectedobjects withEW≥110˚A; detections,
which consists of the 21 objects which were detected in
6 However,theobjectwithanEWof5˚Aislikelyaninterloper. the Megacam data; spectroscopically confirmed, which
Ithasan∼10σdetectioning′,and>30σdetectionsinr′,i′and
z′. Wewillexcludethisobjectfromtherestofouranalysis,giving consists of the 28 objects (7 detected) which were con-
us21detections. firmed to be Lyα emitters at z ∼ 4.5. These stacks are
Lyα Galaxies at z ∼ 4.5 5
shown in Figure 2. (r′, i′ and z′), itwould be hardto fit eachstack to mod-
els with a full range of every parameter. However, given
3.4. Age and Mass Estimates whatwe havelearnedfromstudying these objects in the
In order to study these galaxies, we have created nu- color-colorplane,itis relativelystraightforwardtofitan
merous synthetic stellar population models which we age to each of these stacks (assuming a star formation
compare to our observed galaxies. The most illuminat- history and no dust). In order to fit the age, we con-
ing way in which we can study the observed vs. model sidered 20 different ages ranging from 1 Myr - 200 Myr
galaxies are in a color-color plot (Figures 2 and 3). We (including the five ages shown in Figure 2). By finding
have plotted the r′ - i′ color vs. the r′ - nb color of our the model point (we used the A = 0 point for each
1200
objects, and then overplotted many model curves. Be- model;seediscussionfordetails)closesttoeachstack,we
cause changing the metallicity did not much change the haveeffectivelyfoundtheaverageageofanobjectinthe
position of the model curves in the plane we are study- stack. Because we didn’t want to limit ourselves to one
ing, we have elected to only use models with Z = .02 star formation history, we have found two ages for each
Z⊙ . The differing line styles represent the two different object; one from the closest model with a constant star
star formation rates we used, with the solid lines repre- formation rate (SFR), and one from the closest model
senting continuous star formation, and the dashed lines withanexponentialSFR.TheseagesarereportedinTa-
representing exponentially decaying star formation with bles1and2forconstantandexponentiallydecayingSFR
a decay time of τ = 107 years. respectively. For a constant SFR, the ages ranged from
We ran the models at 24 different ages ranging from 1 - 200 Myr, while for an exponentially decaying SFR,
1 Myr - 2 Gyr7, and in the figure we show the five ages they ranged from 1 - 40 Myr.
which best surround the data, represented by different In order to find the stellar mass, we wanted to only
colors. We included dust in the models, ranging from use the best-fit model for each stack. We took that to
A = 0 - 2. The extent of the model tracks repre- be the model with the best-fit age found above, and we
1200
sents the different dust optical depths. As we discussed calculated a mass for each SFR. The mass was found by
above,thisdustdoesnotaffectthestrengthofthemodel a simple weighted ratio of object flux to model flux. We
Lyα line. We have connected the zero-dust end of the derivedtheweightingbyminimizingtheratioofthemass
model trackswith the dotted black line in orderto show error to the mass. The mass error was calculated via a
where galaxies with no dust should lie. monte carlo simulation using the flux errors from each
band for each stack. These mass estimates are reported
in Tables 1 and 2. The masses of the objects from both
typesofstarformationratesrangefrom∼2×107-2×109
M⊙ . Thesemassesareindicativeofanaverageobjectof
the stack, and likewise the reported error is the error in
the mean of the object mass in each stack, rather than
the error in the measurement.
Fig. 2.— Color-color plot showing the six stacked points along
with stellar population model tracks. The vertical axis is the r′ -
narrowband(H0)color,andthehorizontalaxisisthe r′ -i′ color.
The model tracks shown represent the colors of models in this
plane. Allmodelsarefor.02Z⊙ . Thesolidtracksdenotecontin-
uousstarformation,whilethedashedtracksdenoteexponentially
decayingstarformation,withadecaytimeofτ =107 years. The
differentcolorsofthemodelcurvesrepresenttheagesofthemod-
els. Thelengthofthemodelcurvesrepresentthemodelcolorswith
Fig. 3.— Color-color plot showing the location of the 21 indi-
differing amounts of dust. Models were run with A1200 = 0 – 2,
withtheA1200pointlyingatthesmallestr′-nbcolor. Thedotted vidual detections (along with the outlier), along with stellar pop-
ulation model tracks. The individual galaxies are plotted with
blacklineconnectsthebottomofthemodelcurves,andassuchis
different colored symbols to denote whether they have been spec-
the“zero-dust”line. Ageswerefoundforeachstackbyminimizing
troscopicallyconfirmedtobeatz∼4.5. Theaxesandmodeltracks
thedistance fromthe stacktothenearest A1200 =0modelpoint
are the same as in Figure 2. In general, objects above the zero-
(i.e. nearest point on the zero-dust line) for each star formation
dustlinewouldhavetheirEWenhancedduetodusteffects,while
rate(20differentpossibleages wereused).
objectsbelowthislinewouldhaveintrinsicallyhighEWs,orhave
theirwholespectrum attenuated byageometricallyhomogeneous
dustdistribution(forthelowEWsubsample).
With significant flux in only three broadband points
7 At1Myr,thetwodifferentstarformationrateshaveidentical In Figure 3, we have plotted the 21 individually de-
colors, so their model tracks overlap, that is why it appears as if
tected galaxies (plus the one outlier), along with their
thereisnocurveforanexponentially decayinggalaxyat1Myr.
6 Finkelstein et al.
TABLE 1 TABLE 2
Best-Fit AgesandMasses fora ConstantSFR Best-Fit AgesandMasses foran ExponentiallyDecaying
SFR
Stack r’mag. Age χ2 Mass(108M⊙ )
ofstack (Myr) perobject Stack Age χ2 Mass(108M⊙ )
(Myr) perobject
7LowEW 23.27±0.10 200 18.48 23.54±1.66
7MidEW 23.49±0.12 80 0.93 8.52±0.74 7LowEWObjects 40 4.05 16.15±1.14
7HighEW 24.08±0.21 4 0.94 0.95±0.15 7MidEWObjects 20 1.05 4.24±0.37
21Det. 22.39±0.09 40 0.62 4.60±0.29 7HighEWObjects 4 0.97 0.94±0.14
76Undet. 23.31±0.47 1 6.59 0.19±0.07 21Det. Objects 10 3.43 1.97±0.13
28Spec. Conf. 22.96±0.18 3 0.05 0.62±0.09 76Undet. Objects 1 6.59 0.19±0.07
28Spec. Conf. Objects 2 0.09 0.68±0.10
Note. — Each model has a metallicity of .02 Z⊙ . The
low EW bin is from 20–40 ˚A; the middle bin is from 45–100 Note. — See Table 1 notes.
˚A; the highest bin has objects with EW ≥ 110 ˚A. The bin
with 21 detected objects consists of the 21 galaxies which M⊙iscomparabletootherstudiesofsimilarobjects. Due
were detected at the 2σ level in at least two of the r′, i′ or to the fact that our observed magnitudes are brighter
z′ bands. The last bin consists of the 28 objects which were than the majority of surveys which we can compare to,
spectroscopically confirmed to have a Lyα line at z ∼ 4.5. we point to the trend between apparent magnitude and
Theχ2 isdefinedasthesquareofthedistanceinsigmaunits mass. That is, that the fainter an object is, the less
to thebest-fit age point (SeeFigure 2). The r′ magnitudeis massiveitoughttobe (assumingtheyareatsimilarred-
for the stack as a whole, while themass is per object. shifts). Ellis et al. (2001) discovered a very faint (I ∼
30)Lyαemittinggalaxyatz=5.576thatwasabletobe
1σ error bars. We have distinguished between the indi- detected because it was lensed by a foreground galaxy.
vidualgalaxiesinordertoshowthosethatwerespectro- Due to its faint magnitude, this object was determined
scopically confirmed. By their position in this plot, we tohaveamassofaround106M⊙ ,implyingaveryyoung
hopetofindoutwhethertheyhaveanintrinsicallystrong age(∼2Myr). Thisobjectisconsistentwithourresults,
Lyα line, or one that is enhanced due to dust effects. inthatitisbothmuchfainterandmuchlessmassivethan
the objects we are studying in this paper. Gawiseret al.
4. DISCUSSION
(2006) studied numerous LAEs which they detected at
4.1. Age and Mass Estimates z∼ 3.1, with a median R magnitude of ∼ 27. In order
Wefittedtheagesofourstackedpointsusingdust-free to take advantage of the full spectral range available to
models. Objects in the stack of the 76 non-detections them, they stacked their sources before performing SED
have the lowest age by far of all of the stacks (1 Myr), fitting. Whenthiswasdone,theyfoundanaveragemass
and thus the lowest mass (0.19 × 108 M⊙ ). This is not perobjectof∼5×108M⊙. Allofourstacks(exceptthe
surprising, however, because these objects are selected stack of undetected objects) have an average individual
based on their Lyα line strength. In these galaxies, the objectr′ magnitudeoflessthan27,soevengiventhedif-
lineisstrongenoughtobeseenclearlybutthecontinuum ferences in photometric systems, they are likely brighter
is not, so they are intrinsically faint galaxies,implying a (much brighter in fact, due to their increased distance).
low stellar mass. Their colors are best fit by a low age, While (depending on the SFR) not all of these stacks
which also points to a low mass; the galaxy has simply have masses higher than 5 × 108 M⊙ , they are of the
not had that much time to form stars. same order.
The three equivalent width bins show a logical trend Perhapsthebestcomparisontoourresultscomesfrom
in age and mass. The lowest EW bin has the highest the study by Pirzkal et al.(2007), where they are study-
age at 200 (40) Myr, and thus the highest mass at 23.54 ing the properties of z ∼ 5 LAEs detected in the GRism
(16.15)×108 M⊙ foracontinuous(exponentiallydecay- ACS Programfor ExtragalacticScience (GRAPES) sur-
ing) SFR. This age is consistent with the fact that the vey(Pirzkal et al.2004). Usingmodelswithanexponen-
EWs in this bin were not higher than normal, and thus tially decayingSFR, they founda massrangeof3 × 106
this EW bin is best fit by an older stellar population. –3×108 M⊙ inobjectswithani′ rangeof25.52–29.35
The middle EW bin has a younger age and a smaller (the i′ range of an average object in our stacks is from
mass, and the highest EW bin has the youngest age out 25.12 – 29.90). While the majority of our objects have
ofthe three bins at4 (4) Myr,with the smallestmassat i′ brighter than 26 (all except objects from the unde-
j0e.c9t5s(w0.i9th4)th×e1h0i8ghMes⊙t.EWThiasrewotruuldlyinthdeicmatoestthyaotutnhgeaonbd- taellctoefdtshteacGkRanAdPtEhSe sopbejcetcrtossceoxpceicpatllyonceonhfiarvmeeid′ sfatainctke)r,
primitive galaxies in our sample (that were detected). than 26. This explains well the fact that our derived
Theresultsforthestackofall21MMT-detectedobjects masses are in general greater than those derived for the
areintermediateamongtheresultsgroupedbyequivalent GRAPES LAEs. We are studying intrinsically brighter,
width, as expected. The ages for this stack were 40 (10) and thus more massive objects. While our wavelength
Myr,andmasseswere4.6(2.0)×108 M⊙ . The stackof baseline is not very large, we are able to estimate the
spectroscopicallyconfirmedobjectsshouldliesomewhere mass of stars producing UV and Lyαlight.
between the detected and non-detected stacks, and in-
4.2. Dusty Scenario
deed this is the case, with agesof 3 (2) Myr, andmasses
of 0.62 (0.68) × 108 M⊙ . In studying Figure 3, we have identified three distinct
The overall mass range we found of ∼ 2×107 – 2×109 regions in the color-colorplane: one region which would
Lyα Galaxies at z ∼ 4.5 7
houseintrinsicallyblueobjects,oneregionofredobjects jects in our sample. There still may be dust present in
with high EW, and one of red objects with low EW. these objects, but it would be in a uniform distribution,
The region that lies below the zero-dust line and blue- quenching both the continuum and the Lyα line.
ward of r′ - i′ = 0 would appear to house objects that
have moderate-to-high EWs with blue colors, indicating
5. CONCLUSION
that they are intrinsically blue, containing young stars.
ThelowerEW/redderr′ -i′ colorsub-areaofthisregion WehaveusedMMT/Megacambroadbandphotometry
would contain objects may have some dust, but it is af- in order to study the continuum properties of Lyα emit-
fecting both the continuum and Lyα line, lowering the ting galaxies at z∼ 4.5. By dividing our objects into
EW and reddening the color. six differentcatergoriesand stackingthem, we wereable
The second regionconsists of the area above the zero- to derive age and stellar mass estimates for an average
dust line with an r′ - nb color & 2. This region would galaxy in each stack (ignoring dust effects). In all cases,
hold objects that lie in the dusty regime of the models, the best fit ages were young, with the oldest age com-
yet they would still have “red” r′ - nb colors, indicating ing from objects with the lowestequivalent width. Even
a large EW. Objects such as these could be explained those objects had anage of only 200Myr (40 Myr)for a
by dust quenching of the continuum (causing the red continuous(exponentially decaying)star formationrate.
r′ - i′ color), while the Lyα line would not be affected As would be expected, the bin with the highestEW had
by the dust, making it appear to be strong (causing the the youngest age out of the continuum detected objects
red r′ - nb color). The last region consists of the area at 4 Myr (from both SFRs). The youngest age from
above the zero dust line, but with a smaller r′ - nb color bothSFRs camefromobjectswhosecontinuumflux was
(r′ - i′ ∼ 1, r′ - nb = 1 – 1.5). Objects in this area undetected, mostly because these were detected in the
would have a redder r′ - i′ color, with a smaller r′ - narrowbandimage, and not in the continuum images.
nb color (smaller EW), indicating that they are perhaps The derived stellar masses ranged from ∼ 2×107 M⊙
older galaxies with some dust effects present. for the undetected objects to ∼ 2×109 for the lowest
In the above paragraph,we have outlined different re- EW objects. Our derived masses are consistent with
gionsin our color-colorplane whichwould house objects other mass estimates of similar objects. In the major-
from the different scenarios we are examining. We plan ity of cases, the masses of our objects were greater than
to usethis to determine the likelihoodofdustquenching those from other studies (Ellis et al. 2001; Gawiser et
ofthe continuumresulting ina largeLyαEW.However, al. 2006; Pirzkal et al. 2007). However, the magni-
the current error bars on individual objects are large tudes of our individual objects were brighter than those
compared to their distance from the zero-dust line. In of the comparisonstudies, implying that our objects are
ordertoquantifythis,wecalculatedthe distanceofeach larger, more massive galaxies than those from the other
object from the zero-dust line in units of σ. The mean studies. In conclusion, while we have not yet been able
distance of the 21 objects we are studying is 1.10 ± 0.48 to definitively determine the likelihood of dust enhance-
σ. While it is possible that the dust enhancement of the ment of the Lyα line, we have been able to shed some
Lyα line is present in our sample, due to the large error light on the physical properties of these high-z objects.
bars we cannot rule out the possibility that all of these The Lyα galaxies in this survey, especially those with
objects lie on the zero-dust line at a level of ∼ 1 σ. the largestLyαEW,aresomeofthe youngestandleast-
Performing the same exercise with the six stacked massive objects in the early universe known to date.
points with their smaller error bars, the mean distance
from the zero-dust line is 1.17 ± 0.81 σ. However, the
smaller error bars do allow us to make more of a char- This work was supported by the Arizona State Uni-
acterization of a few of the stacks. We can most likely versity/NASA Space Grant, the ASU Department of
rule out clumpy dust enhancement of the Lyα line for Physics and the ASU School of Earth and Space Explo-
the low and middle EW stacks, mainly because they lie ration. This workmadeuseofimagesand/ordataprod-
below the zero-dust line, and far from the region in our ucts provided by the NOAO Deep Wide-Field Survey
color-color plane where we would expect to find the en- (Jannuzi and Dey 1999), which is supported by the Na-
hanced line. Also, in the line enhancement scenario, one tional OpticalAstronomy Observatory(NOAO). NOAO
would expect to find extremely high EW Lyα lines, and is operated by AURA, Inc., under a cooperative agree-
these twostacksdo notcontainthe highest33%EWob- ment with the National Science Foundation.
REFERENCES
Ajiki,M.etal.2004,PASJ56,597 Fukugita, M., Ichikawa, T., Gunn, J. E., Doi, M., Shimasaku, K.
Bertin,E.&Arnouts,S.1996, A&AS,117,393 &Schneider,D.P.1996,AJ,111,1748
Bruzual,G.&Charlot,S.2003,MNRAS,344,1000 Fynbo,J.U.,Mo¨ller,P.&Thomsen,B.2001,A&A,374,443
Calzetti,D.,Kinney,A.L.&Storchi-Bergmann,T.1994,ApJ,429, Gawiser,E.etal.2006,/apj,642,L13
582 Hansen,M.&Oh,S.P.2006, MNRAS,367,979
Charlot,S.&Fall,S.M.1993,ApJ,415,580 Hu,E.M.,Cowie,L.L.&McMahon,R.G.1998,ApJ,502,L99
Cowie,L.L.,&Hu,E.M.1998,AJ,115,1319 Hu,E.M.,Cowie,L.L.,McMahon,R.G.,Capak,P.,Iwamuro,F.,
Dawson, S., Rhoads, J. E., Malhotra, S., Stern, D., Dey, A., Kneib,J.-P.,Maihara,T.&Motohara,K.2002,ApJ,568,L75
Spinrad, H., Jannuzi, B. T., Wang, J. X. & Landes, E. 2004, Hu,E.M.,Cowie,L.L.,Capak, P.,McMahon,R.G.,Hayashino,
ApJ,617,707 T.&Komiyama,Y.2004,AJ,127,563
Ellis,R.,Santos,M.R.,Kneib,J.-P.&Kuijken,K.2001,ApJ,560, Jannuzi,B.T.,Dey,A.1999,inASPConf.Ser.191, Photometric
L119 RedshiftsandHighRedshiftGalaxies,ed.R.J.Weymann,L.J.
Fujita,S.S.etal.2003,ApJ,586,L115 Storrie-Lombardi,M.Sawicki, & R. J. Brunner (San Francisco:
ASP),111
8 Finkelstein et al.
Kodaira,K.etal.2003,PASJ,55,L17 Rhoads,J.E.&Malhotra,S.2001,ApJ563,L5
Kudritzki,R.-P.etal.2000,ApJ,536,19 Rhoads, J. E., Dey, A., Malhotra, S., Stern, D., Spinrad, H.,
Landolt,A.U.1992,AJ,104,340 Jannuzi, B.T.,Dawson,S.,Brown,M.J.I.&Landes,E.2003,
Madau,P.1995,ApJ,441,18 AJ,125,1006
Malhotra,S.&Rhoads,J.E.2002, ApJ,565,L71 Rhoads, J. E., Xu, C., Dawson, S., Dey, A., Malhotra, S., Wang,
Malhotra, S., Wang, J.X., Rhoads, J. E., Heckman, T. M. & J. X.,Jannuzi, B. T., Spinrad, H.,& Stern, D.2004, ApJ, 611,
Norman,C.A.2003,ApJ,585,L25 59
Malhotra,S.&Rhoads,J.E.2004, ApJ,617,L5 Shimasaku,K.etal.2003,ApJ,586,L111
McLeod, B. M., Caldwell, N., Williams, G. & Shimasaku,K.etal.2006,PASJ,58,313
Conroy, M. 2006, Megacam Observers Manual, Steidel, C. C., Giavalisco, M., Pettini, M., Dickinson, M. &
http://grb.mmto.arizona.edu/∼ggwilli/mmt/docs/megacam/megacam mAadneulbale.rhgtemr,lK.L.1996,ApJ,462,L17
McLeod, B. A., Gauron, T. M., Geary, J. C., Ordway, M. P. & Stiavelli, M., Scarlata, C., Panagia, N., Treu, T., Bertin, G. &
Roll,J.B.1998, SPIE,3355, 477 Bertola,F.2001,ApJ,561,L37
Neufeld,D.A.1991,ApJ,370,L85 Taniguchi,Y.etal.2005,PASJ,57,165
Oke,J.B.&Gunn,J.E.1983,ApJ,266,713 Tody,D.1993,ASPC,52,173
Ouchi,M.etal.2001, ApJ,558,83 Tody,D.1986,SPIX,627,733
Ouchi,M.etal.2003, ApJ,582,60 Valdes,F.G.1998,ASPC,145,53
Ouchi,M.etal.2004, ApJ,611,685 Valdes,F.G.&Tody,D.1998,SPIE,3355,497
Papovich,C.,Dickinson,M.&Ferguson,H.C.2001,ApJ,559,620 Venemans,B.P.etal.2002, ApJ,569,L11
Partridge,R.B.&Peebles,P.J.E.1967, ApJ,147,868 Venemans,B.P.etal.2004, A&A,424,L17
Pentericci,L.etal.2000, A&A,361,25 Wang, J. X., Rhoads, J. E., Malhotra, S., Dawson, S., Stern, D.,
Pirzkal,N.etal.2004,ApJS,154,501 Dey, A., Heckman, T. M., Norman, C. A. & Spinrad, H. 2004,
Pirzkal,N.etal.2007,inpreparation ApJ,608,L21
Rhoads,J.E.2000,PASP,112,771,703 Wang,J.X.etal.2007,inpreparation
Rhoads, J. E., Malhotra, S., Dey, A., Stern, D., Spinrad, H. &
Jannuzi, B.T.2000,ApJ,545,L85