Stellar populations in high redshift Ly-alpha emitters

L. PentericciA.Grazian, A. Fontana

(INAF-ROME)

( paper will appear on astroph in the next few days)

HEIDELBERG 08/10/2008

A meaningful comparison to understand the connection between the two populations should include the physical properties of galaxies (total stellar mass, age, SFR and SF histories …)

For LBGs many studies exist from low to high redshift:

z ≈2: 6x 1010 M⊙ (Erb et al. 06);

z ≈3: u-dropouts 3-400 Mys , 1010 M⊙(Shapley et al 01, Papovich et al. 01, Iwata et al. 05, Rigopoulou et al. 06);

z ≈ 4: B-dropouts 2-300 Myrs 109 -1010 M ⊙ ( Pentericci et al. 07);

z ≈ 5: V-dropouts: 100 Myrs , few x109 M⊙ (Verma et al. 07);

z ≈6: i-dropouts (Yan et al. 06, Eyles et al. 05, Egami et al. 05);

……and many others…………

LAEs and LBGs: same galaxies or diffrent populations ??

For narrow band selected LAEs fewer stellar population

studies exist (Lai et al. 2006, Gawiser et al. 06, Pirzkal et al

2007 Finkelstein et al. 07,08)

GENERAL PICTURE : Ages : 10-100 Myrs Masses typically ≈108 M⊙ but some as low as 107 M⊙ No extinction

YOUNGER, LESS MASSIVE, LESS DUSTY THAN LBGs at similar redshift

However!!! Few LAEs have been observed in the mid-IR which is essential to constrain

the masses (e.g. Fontana et al. 06) Few samples have good near-IR data which are required to break SED model

degeneracy (e.g. young dusty population vs old dust-free population) Most studies rely on stacked analysis i.e. derive only average ages

and masses of the entire sample.

To overcome these limitations in the analysis of LAE stellar populations, we study Lyα emitting galaxies that have been selected from their continuum properties as LBGs

We take advantage of the large area, depth and excellent multivawelength spectral coverage of the GOODS survey (Dickinson et al. 04, Giavalisco et al. 04)

Sample of LBGs with Lyα in emission selected from GOODS-South field

Galaxies were selected from the GOODS-MUSIC z-detected sample (Grazian et al. 06) as B,V and i dropouts (color selection as in Giavalisco et al. 04)

We then selected galaxies with published spectroscopic confirmation mostly from GOODS/FORS (Vanzella et al. 05,06,08) or other papers

Only objects with Lyα line in emission were retained

No AGNs were included (no X emis.)

Sample of LBGs with Lyα in emission selected from GOODS-South field

Galaxies were selected from the GOODS-MUSIC z-detected sample (Grazian et al. 06) as B,V and i dropouts (color selection as in Giavalisco et al. 04)

We then selected galaxies with published spectroscopic confirmation mostly from GOODS/FORS (Vanzella et al. 06,07) or other papers

Only objects with Lyα line in emission were retained

No AGNs were included (no X emis.)

Sample of LBGs with Lyα in emission selected from GOODS-South field

Galaxies were selected from the GOODS-MUSIC z-detected sample (Grazian et al. 06) as B,V and i dropouts (color selection as in Giavalisco et al. 04)

We then selected galaxies with published spectroscopic confirmation mostly from GOODS/FORS (Vanzella et al. 06,07) or other papers

Only objects with Lyα line in emission were retained

No AGNs were included (no X emis.)

Redshift and EW distribution Redshift distribution

showing the B , V and i dropouts: ≈70 galaxies

Restframe EW distribution:

• > 50% has EW0 larger than 20Å so they would be selected as Lyα emitters in narrow band surveys

• For comparison we plot the LBGs sample by Tapken et al. 07 (continuum selected) at z=3-5 (green histogram) and the 22 Lyα emitters from Finkelstein et al. 07 at z=4.5 which have detections also in at least 2 continuum bands ( red histogram)

GOODS – MUSIC catalogue The full multi-wavelength information consisting of 14

bands photometry (VIMOS /2.2-WFI U, HST/ACS BVIz , VLT/ISAAC JHK, IRAC 3.6, 4.5,5.8,8 μ) was retrieved from the GOODS-MUSIC catalog (Grazian et al. 06) which uses

the ConvPhot algorithm (DeSantis et al. 07) to match PSFs from the different instruments

Most of the galaxies are detected in all bands (above the Lyman break) and in particular most have near-IR and/or IRAC detections

This implies that we can determine physical properties of individual galaxies without stacking the sample

Stellar population models BRUZUAL & CHARLOT (2003)

CHARLOT & BRUZUAL (2007)

MODEL PARAMETERS: Salpeter IMF between 0.1 and 65 M⊙

SFR: exponentially declining with e-folding time

τ =0.1,0.3,0.6,1,2,3,5,9,15 Dust extinction: Calzetti or SMC law

E(B-V)=0,0.03,0.06 then 0.1-1.1 (in bins of 0.05) Metallicity : 0.02-0.2,1,2.5 solar Age : <age of universe at that redshift

OUTPUT Total stellar masses (with uncertainties reduced thanks to the inclusion of the

4 IRAC bands e.g. Fontana et al. 06) continuum based SFR best fit stellar ages, E(B-V), metallicity , tau…

The fits are well constrained for most of the galaxies

For galaxies with EW> 20 Å, a second fit was performed excluding the band that contains the Lyα line: these fits give totally comparable results. This implies that1. The fits are robust and do not depend on the single band2. The properties we derive do not depend on the presence of

line emission

NB: The results obtained with the new Charlot & Bruzual (2007) models are similar: • the masses are lower by a 25% factor • the ages are younger by 20% Differences are systematic all main results and correlation are also valid

Increasing Stellar Mass

Increasing Age

Increasing Stellar Mass

Increasing Age

Redshift 3.79Mass=6.3 x 108 M⊙Age= 10 Myrs

Increasing Stellar Mass

Increasing Age

Redshift 4.2Mass=3 x 1010 M⊙

Age=510 Myrs

Ages: old Lyα emitters?

Age distribution is shown: best fit values (black) and the sum of the Probability Distribution function of each galaxy (red)

Subsample with EW0 > 20 A is shown in green Most galaxies are modeled with very young stellar populations (T=

10-200 Mys) as expected and comparable to what found by NB selected LAEs

13/68 galaxies have best fit ages > 0.5 Gyrs & Agemin > 300 Myrs so they are most certainly not primeval galaxies.

OLD

Examples of old Lyα emitters

These objects clearly show a prominent Balmer break: SED modelling gives large best fit ages (1.1 Gyr & 0.7 Gyr) Considerably younger models can be ruled out with high confidenceBoth of them are bright Lyα emitters with EW >> 20 Å

Old Lyα emittersThe Age distribution of the EW>20 subsample is consistent with

that of the entire sample

Median Age =300 Myr

Median Age=250 Myr for

sub-sample with EW>20 Å

(with some galaxies having ages ≈ 1 Gyr

some evolved galaxies should be observed

also in NB selected samples : where are they???

Perhaps extreme properties are not found in large samples because of stacked photometry method: the few old galaxies are “diluted” in the more numerous young population

Also most NB surveys lack good data coverage in the region around 4000 Å restframe which is crucial to reduce model degeneracies

In the most recent work of Finkelstein et al. (2008) found 2 out of 15 LAEs with ages around 700 Mys

Q. How can we observe bright Lyα emission from evolved stellar populations ?

- A clumpy dusty ISM can actually enhance Lyα EW by selectively suppressing the continuum more than the Lyα photons (see talk by Finkelstein )

- In (some) older galaxies dust could have been destroyed or swept away by gas outflows so the Lyα photons can again escape freely (e.g. Shapley et al. 03)

- Type II LAEs (see talk by Shimizu)- Other….

Total stellar masses

Masses are in the range 108.5 -10 11 M⊙

Median stellar mass≈ 6 x 109 M⊙

4 x 109 M⊙ for EW>20 Å subsample

higher than most other estimates for LAEs at redshift 3-6 (Pirzkal et al. 07 Finkelstein et al. 07 Gawiser et al. 06)

Significant lack of massive galaxies with large EW

This could explain the lack of massive objects in samples of NB selected Lyα emitters: this technique tends to pick objects with very high EW— much larger than the nominal cut of 20 Å

Q. Why do massive galaxies lack bright Lyα emission?

- Not an observational bias - Massive galaxies could be more dusty (e.g. Schaerer

& Verhamme) not observed in our data - Massive galaxies reside in most massive DM halos

and therefore could be contain larger amounts of neutral gas (Steidel et al.)

- Massive galaxies contain outflowing neutral gas that has larger velocity dispersion and can depress partially the Lyα line, also resulting in more asymmetric profiles (e.g. Tapken et al. 2007,Ahn et al. 01)

Dust content: The presence of dust (although in small amounts) is required by

the SED fits of many galaxies E(B-V) ≠ 0 for half of the galaxies. There is a significant correlation between E(B-V) determined from the continuum fit and the Lyα EW from the spectra.

In agreement with Shapley et al.(03) for LBGs at z=3 and with Pentericci et al. (07) for LBGs at z=4

who find that LBGs without

line emission are dustier than

LBGs with line emission

A scenario where all LBGs are intrinsic Lyα emitters and dust suppress the emission has been recently proposed by Schaerer & Verhamme (07) and could explain some of the observed trends

In summary….Most Lyα emitters are extremely young galaxies but a non

negligible fraction contains an evolved stellar population:The derived masses are larger than those of NB-selected

galaxiesThere is a lack of massive galaxies with large EWThere is a strong correlation between line EW and dust

extinction

also (results not shown in this talk)Masses and SFRs are in broad accordance with model

predictions SFR estimated from UV continuum > than SFR from line

emission

There is a continuous range of physical properties between LAEs and LBGs

The smaller inferred masses of NB selected samples is due to the observed trend mass-EW

The lack of old galaxies in NB selected samples could be due to lack of proper near/mid-IR observations to constrain models and/or to “dilution” of the small fraction of old objects in the more numerous young galaxies

Unification between the two population is not simply LAEs=young LBG=older galaxies but probably requires other variables, including amount and distribution of dust amount and kinematics of neutral gas, viewing angle etc etc