Starformation Chemical Enrichment and Feedback at z2 3
Star-formation, Chemical Enrichment, and Feedback at z~2 -3 Alice Shapley (Princeton) Collaborators: Chuck Steidel, Max Pettini, Dawn Erb, Naveen Reddy, Kurt Adelberger
Overview and Motivation • UV-color-selected galaxies at z~1. 5 -3. 0 (9 -11 Gyr) • Lots of progress in understanding their global and detailed properties, their effects on the IGM • New results: • The origin of the ionizing background at z~3, direct spectroscopic measurement of ionizing radiation from galaxies at z~3, constraints on fesc, Jn , G • Robust measurement of mass-metallicity relation at z~2. Plus: new info on physical conditions in starforming regions at z~2, implications for inferred chemical abundances, nature of star-formation
z>2 color-selection • Adjust z~3 UGR crit. for z~2 (Adelberger et al. 2004) • Spectroscopic follow-up with optimized UV-sensitive setup (Keck I/LRIS-B) (Steidel et al. 2004)
Redshift Distributions LBG: z~3 (940) SLBG=1. 7/arcmin 2 n. LBG=1. 4 x 10 -3 Mpc-3 BX: z=2 -2. 5 (816) SBX=5. 2/arcmin 2 n. BX=2. 0 x 10 -3 Mpc-3 Unsmoothed BM: z=1. 5 -2. 0 (118) SBM=3. 8/arcmin 2 n. BM=1. 7 x 10 -3 Mpc-3 750 gals at z=1. 4 -2. 5 BX/BM/LBG with R<=25. 5 (Steidel et al. 2004; Adelberger et al. 2004)
Contribution of LBGs to the Ionizing Background at z~3
Redshift Distributions LBG: z~3 (940) SLBG=1. 7/arcmin 2 n. LBG=1. 4 x 10 -3 Mpc-3 BX: z=2 -2. 5 (816) SBX=5. 2/arcmin 2 n. BX=2. 0 x 10 -3 Mpc-3 Unsmoothed BM: z=1. 5 -2. 0 (118) SBM=3. 8/arcmin 2 n. BM=1. 7 x 10 -3 Mpc-3 750 gals at z=1. 4 -2. 5 BX/BM/LBG with R<=25. 5 (Steidel et al. 2004; Adelberger et al. 2004)
Origin of the Ionizing Background • Major question: how did the universe transform from having a neutral to highly ionized IGM? • Many new observational results (SDSS QSOs, WMAP), plus theoretical work (i. e. predictions for future 21 cm telescopes) • Nature of the ionizing background important for understanding reionization and inferring physical properties of Lya forest • Measured with QSO “proximity effect”: current estimates J 912~10 -21 erg/s/cm 2/Hz/sr , G~10 -12 s-1 (Scott et al. 2000) • Questions: What are relative contribution of QSOs & gals vs. z? (drop-off in QSO number density at high redshift) What is fesc for galaxies? ? ?
Definitions of fesc • What is fesc? fesc= Llyc, out/Llyc, in= exp(- ISM, lyc) • What is fesc, rel and why? fesc, rel= fesc x (LUV, out/LUV, in)-1 = fesc x exp( ISM, UV) Useful for deriving global quantities, such as J 912, gal, based on LBG UV luminosity function • fesc, rel can be re-written: fesc, rel= (LUV/Llyc)in/(FUV/Flyc)obs x exp( IGM, lyc) Stellar pop. models: 1. 5 -5. 5 We measure this Simulations of z~3 IGM opacity in Lyc range
Search for Lyman-Continuum Emission Lyman-Continuum Observations of Galaxies • z~0: HUT spectra at z~0, fesc<0. 01 -0. 15 (Leitherer et al. ) • z=1. 1 -1. 7: HST/STIS UV imaging, fesc <0. 01 -0. 06 (Malkan et al. ) • z~3: LBG composite spectrum, fesc, rel>0. 5, fesc~0. 10 (Steidel et al. ) • Controversy at z~3!! • HST/WFPC 2 UBVi colors, fesc <0. 04 (Fernandez-Soto et al. ) • VLT spectra of 2 gals, fesc<0. 02 -0. 03 (Giallongo et al. 2002) • NB imaging of 2 z>3 galaxies, fesc, rel <0. 72 (Inoue et al. 2005)
Detection of Ly-C Emission? Lyman Cont. Leakage • 29 gals at <z>=3. 4+/-0. 09 • Significant Ly-cont flux in composite spectrum 5 times more ionizing flux than QSOs at z~3 • Bluest quartile in (G-R)0, strong Lya emission, IS abs lines weaker than c. B 58: is it representative? (Steidel et al. 2001) • fesc different from local SB (Leitherer et al. 1995)
Search for Lyman-Continuum Emission Lyman-Continuum Observations of Galaxies • z~0: HUT spectra at z~0, fesc<0. 01 -0. 15 (Leitherer et al. ) • z=1. 1 -1. 7: HST/STIS UV imaging, fesc <0. 01 -0. 06 (Malkan et al. ) • z~3: LBG composite spectrum, fesc, rel>0. 5, fesc~0. 10 (Steidel et al. ) • Controversy at z~3!! • HST/WFPC 2 UBVi colors, fesc <0. 04 (Fernandez-Soto et al. ) • VLT spectra of 2 gals, fesc<0. 02 -0. 03 (Giallongo et al. 2002) • NB imaging of 2 z>3 galaxies, fesc, rel <0. 72 (Inoue et al. 2005)
New Ly-C Observations • Deep Keck/LRIS spectra for a sample of 14 objects • Red side: detailed obs of interstellar lines (17 hr, 2 x resolution) • Blue side: sensitive observations of Lyman Continuum region (22 hr, 8 hr with New Blue Camera, 1. 3 x resolution) • Observations approach the quality of c. B 58, but larger sample • Deep Sample (bright LBGs): <z>=3. 06, <RAB>=23. 92, <E(B-V)>=0. 11, WLya=2Å (em/abs) • Steidel et al. Ly-cont sample: <z>=3. 40, <RAB>=24. 33, <E(B-V)>=0. 07, WLya=15Å • Composite Sample of Shapley et al. (811 galaxies w/ spectra): <z>=2. 96, <RAB>=24. 60, <E(B-V)>=0. 13, WLya=15Å
Examples of Deep LBG Spectra • LRIS-R spectra (LRIS-B covers Lya and Lyman continuum), 1 -1. 5Å res. • Use these spectra to measure physical properties of abs. lines; connect with UV-color, Lya, Lyman continuum
New Ly-C Observations Main observational results: • Ionizing flux is detected for 2 out of 14 objects. Why? • F 900/F 1500 (i. e. Flyc/FUV) quite large for these two galaxies --> significant galaxy-to-galaxy variation in fesc? • Ratio of F 900/F 1500 in average spectrum is four times lower than that in Steidel et al. (2001)
Detections: 2 D Spectra Lyman limit, 912 AA D 3, z=3. 07 Rs=23. 37 3350 AA 3950 AA Lyman limit, 912 AA C 49, z=3. 15 Rs=23. 85 3350 AA 3950 AA
Detections: 1 D Spectra • D 3, double, z=3. 07, Rs=23. 37 F 1500/F 900=5. 1 (7. 1 if you add F 1500 from both components), 15% uncertainty • Yet, IS lines are black! • Reddest galaxy in sample • IGM correction x 3 (starts to strain stellar pop. models) Steidel et al. (2001) found <F 1500/F 900>=17. 7, then corrected by factor of 3. 8 for IGM opacity.
Detections: SSA 22 a-D 3 HST/NIC F 160 W D 3 (z=3. 07, Rs=23. 37) is double, and upper brighter component appears to have significant flux below 912 Å Lots of additional info: near-IR imaging &spectra, HST, Lya blob Lyman limit, 912 AA 3710 AA Lya 4950 AA
Detections: SSA 22 a-D 3 U NB-excess BV R 2” Matsuda et al. 2004 P 200 Rs HST/NICMOS F 160 W
Detections: 1 D Spectra • C 49, z=3. 15, Rs=23. 85 F 1500/F 900=12. 7, 15% uncertainty • One of the bluest galaxies in the sample, with weakest IS lines (non-unity covering fraction), but others with even less, and no Lyman Cont. emission • IGM correction x 3 Steidel et al. (2001) found <F 1500/F 900>=17. 7, then corrected by factor of 3. 8 for IGM opacity.
Detections: 1 D Spectra • D 3 has very young inferred age (<10 Myr), while age of C 49 is unknown. Clear direction to estimate ages for entire sample (near-IR and IRAC photometry), as F 900/F 1500 is very age-dependent Steidel et al. (2001) found <F 1500/F 900>=17. 7, then corrected by factor of 3. 8 for IGM opacity.
Non-Detections: 1 D Spectra C 32, z=3. 29, Rs=23. 68 Weak IS lines (covering fraction smaller than in C 49) C 35, z=3. 10, Rs=24. 18
The Average Spectrum Stack 15 deep spectra, normalized by F 1500 WLya=2Å (plus abs), WIS=2. 2Å abs F 1500/F 900=65 (25% uncertainty), four times larger than S 01 Systematics of sky subtraction VERY IMPORTANT, comparable to avg detection RMS 2. 5 times lower than S 01 (red curve)
The Average Spectrum Benefits of Averaging. Recover information bluewards of Lya, such as Lyman series (independent estimate of typical N(HI) ) Possibly molecular H Other metal lines
Implications • S 01 measured <F 1500/F 900>=17. 7, and applied a factor of 3. 8 S 01 result correction for z=3. 4 IGM opacity -->F 1500/F 900, corr=4. 6 • We now measure <F 1500/F 900>=65. 5, and z=3 IGM correction factor is more like ~3 --> F 1500/F 900=22. So, we measure fesc, rel that is 4 -5 times lower (~10%). • If F 1500/F 900 is representative (which we still need to establish), we can use it to convert LBG 1500 AA luminosity density into 912 AA luminosity function, and then from e(912) to J 912 (and then to G) From Bolton et al. • If we scale S 01 results for J 912, we get 3 x 10 -22 erg/s/cm 2/Hz/sr, and G~8 x 10 -13 s-1, which is more consistent with recent estimates (Bolton et al. 2005) of G(z=3) based on Lya forest optical depth
The Future • Understand ISM/outflow properties of sample (based on redside data of interstellar absorption lines), try to connect to emergent Lyman continuum radiation • Understand spectroscopic sky-subtraction systematics • Compare with AGN contribution -- new value of J 912 is comparable to that from QSOs (Hunt et al. 2004), but depends on fesc in faint QSOs • While we have a larger sample of deep z~3 spectra now, we are also trying to detect Lyman continuum in an independent manner: NB imaging just below the Lyman Limit (Inoue et al. 2005) • Special opportunity because of large structure of galaxies at z=3. 09, unparalleled U-band sensitivity of LRIS-B
Metallicities & Gas Fractions
Redshift Distributions LBG: z~3 (940) SLBG=1. 7/arcmin 2 n. LBG=1. 4 x 10 -3 Mpc-3 BX: z=2 -2. 5 (816) SBX=5. 2/arcmin 2 n. BX=2. 0 x 10 -3 Mpc-3 Unsmoothed BM: z=1. 5 -2. 0 (118) SBM=3. 8/arcmin 2 n. BM=1. 7 x 10 -3 Mpc-3 750 gals at z=1. 4 -2. 5 BX/BM/LBG with R<=25. 5 (Steidel et al. 2004; Adelberger et al. 2004)
Redshift Distributions Other surveys & space densities: LBG: z~3 (940) • K 20 (z=1. 4 -2. 5): ~10 -4/Mpc 3 SLBG=1. 7/arcmin 2 • GDDS (z=1. 6 -2. 0): ~10 -4/Mpc 3 n. LBG=1. 4 x 10 -3 Mpc-3 • SMG (z~2. 5): ~10 -5 /Mpc 3 • FIRES(z=2 -3. 5): ~10 -4/Mpc 3 BX: z=2 -2. 5 (816) SBX=5. 2/arcmin 2 n. BX=2. 0 x 10 -3 Mpc-3 Unsmoothed BM: z=1. 5 -2. 0 (118) SBM=3. 8/arcmin 2 n. BM=1. 7 x 10 -3 Mpc-3 750 gals at z=1. 4 -2. 5 BX/BM/LBG with R<=25. 5 (Steidel et al. 2004; Adelberger et al. 2004)
Other redshift desert surveys Other surveys with galaxies at z~1. 4 -2. 5 • K 20/Bz. K (Cimatti et al. ) (K<20 selection) (~40) • Gemini Deep (Abraham et al. ) (K<20. 6, photo-z) (34) • Radio-selected SMG (Chapman et al. ) (73) (+18 OFRG) • FIRES (Franx, van Dokkum et al. ) (J-K selection) (~10) Now that there are several groups using different selection techniques to find galaxies at z~2, we need to understand how the samples relate to each other Reddy et al. (2005) considered the overlap among different samples, and contribution of each to the sfr density at z~2 -2. 5
Evolution of Galaxy Metallicities • Gas phase oxygen abundance in star-forming galaxies • Fundamental metric of galaxy formation process, reflects gas reprocessed by stars, metals returned to the ISM by SNe explosions (HII regions in sf-galaxies, stars in early-type). • Galaxies display universal correlations between Luminosity (L), Stellar mass (M), and metallicity (Z) • Departures from closed-box expectations can reveal evidence for outflow/inflow • Closed box: Z = y x ln (1/ ) (Z=metallicity, y=yield, =gas fraction=Mgas/(Mgas+M*))
Evolution of Galaxy Metallicities • 10, 000 s of galaxies in the local universe with O/H • Now the challenge is to obtain these measurements at high redshift (evolution will give clues, compare metal census with inferred metal density from integrating the global starformation history) • Measuring gas fractions is very important to quantify how much material has been processed into stars -- currently very indirect • Physical conditions may be different in high-redshift HII regions: implications for inferred metallicities!!!
Abundance Measurements vs. z • Lots of local emission line measurements (10000’s, 2 d. F, Sloan) • At z=0. 5 -1, 4 groups (>=200 gals, CFRS, DGSS, CADIS, TKRS) • At z>2 there were < 10 measurements (mainly LBGs) (Tremonti et al. 2004) • DLAs provide metallicity information from abs lines, but hard to relate
Near-IR spectroscopy of z~2 gals • z~2 ideal for measuring several neb lines in JHK • evidence of M-Z relation at z~2, intriguing information about HII region physics
[NII]/Ha ratios: z~2 metallicities N 2=log([NII] 6584/Ha) 12+log(O/H)=8. 9+0. 57 x. N 2 s~0. 18, factor of 2. 5 in O/H • relationship between [NII]/Ha and O/H • N is mixture of primary and secondary origin • age, ionization, N/O effects, integ. spectra, DIG, AGN (Pettini & Pagel 2004)
Near-IR spectroscopy of z~2 gals M*=4 1011 M K=19. 3, J-K=2. 3 M*=5 109 M Ha spectra of 101 z~2 gals Keck. II/NIRSPEC • Kinematics: linewidths, Mdyn, some spatiallyresolved, tilted lines, compare with stellar masses • Line ratios: HII region metallicities, physical conditions • Ha fluxes: SFRs, compare with UV, models • Offsets between nebular, UV abs and Lya em redshifts -> outflows
Local M-Z Relationship • M-Z possibly more fundamental than L-Z relationship • closed box model relates gas fraction and metallicity, according to the yield • SDSS sample revealed lower effective yield in lower mass galaxies (Tremonti et al. 2004) • importance of feedback
Local M-Z Relationship • We have metallicity information for a large sample of (~100) galaxies at z~2 • For the majority of these, we have used broadband optical and IR photometry to estimate stellar masses (see next slides) (Tremonti et al. 2004) • Therefore we can construct a z~2 M-Z relationship
Opt/NIR SED: Age/Dust Degeneracy • Only optical photometry can’t distinguish between age/dust • Balmer break is sensitive to A/O stars age • K-band probes other side of Balmer break • More discriminating power, IRAC gives you even more (? ? ? ) (Shapley et al. 2001, Papovich et al. 2001)
Stellar Populations & Masses Near/Mid-IR Imaging Ks (2. 15 m) (Barmby et al. 2004, Steidel et al. 2005) • Deep J, K imaging with WIRC, Palomar 5 -m, to Ks~22. 5, J~23. 8 • 4 fields, ~420 galaxies with zsp> 1. 4 • Spitzer IRAC data in Q 1700 field, 3. 6, 4. 5, 5. 4, 8 m • Combine optical and IR SED to model stellar populations, masses
Stellar Populations & Masses Near/Mid-IR Imaging IRAC (4. 5 m) (Barmby et al. 2004, Steidel et al. 2005) • Deep J, K imaging with WIRC, Palomar 5 -m, to Ks~22. 5, J~23. 8 • 4 fields, ~420 galaxies with zsp> 1. 4 • Spitzer IRAC data in Q 1700 field, 3. 6, 4. 5, 5. 4, 8 m • Combine optical and IR SED to model stellar populations, masses
z~2 M-Z Relationship • New sample of 87 star-forming galaxies at z~2 with both M* and [NII[/Ha (gas phase O/H) measurements • Divide into 6 bins in M* , which increases as you move down • clear increase in [NII]/Ha with increasing M* (Erb et al. 2005) • M-Z at z~2!!
z~2 M-Z Relationship • Evol. Comparison with SDSS, where Z is based on [NII]/Ha (for both samples) • Clear offset in relations --> at fixed stellar mass, z~2 galaxies significantly less metal rich than local gals (Erb et al. 2005) • Not evolutionary (z~2 probably red&dead by z~0)
z~2 M-Z Relationship • Important: measure change of Z with (gas fraction) • Must estimate Mgas, which we do from SHa, to Ssfr, to Sgas (assuming Schmidt law) • Very indirect! • Low stellar mass objects have much higher (Erb et al. 2005)
z~2 M-Z Relationship (Erb et al. 2005) • Different models for feedback, using different yields and different outflow rates • Data are best fit by model with super-solar yield and outflow rate greater than SFR (for all masses) • Rate of change of with metallicity gives evidence for feedback ; is very important (estimated indirectly)
z~2 M-Z Relationship (Erb et al. 2005) • Increase of Z with decreasing is shallower than expected for no outflow. Same thing goes for increase of Z with M* • Important: SDSS sample showed evidence for preferential loss of metals from lower mass galaxies (would be steeper) • While the range of baryonic masses in the z~2 sample is small, we see NO evidence for preferential metal loss for less massive galaxies • Remember: outflows generic feature of UV-selected gals, metals in cluster gas
HII Region Physical Conditions
z~2 Physical Conditions • Well-defined sequence in [OIII]/Hb vs. [NII]/Ha in local galaxies (SDSS) (star-formation vs. AGN) • small sample of z~2 starforming galaxies with [OIII]/Hb are offset from this locus (as is DRG) • ne, ionization parameter, ionizing spectrum (IMF, star-formation history) (Erb et al. 2005) • Implications for derived O/H
z~1 Physical Conditions • Well-defined sequence in [OIII]/Hb vs. [NII]/Ha in local galaxies (SDSS) (star-formation vs. AGN) • z~1. 4 star-forming galaxies are offset from this locus (as is DRG) • ne, ionization parameter, ionizing spectrum (IMF, star-formation history) (Shapley et al. 2005) • Implications for derived O/H
z~2 Physical Conditions • Among K<20 galaxies (brightest 10%), [SII] line ratio indicates high electron density • Inferred electron density is ~1000/cm 3, • This is higher than in local HII regions used to calibrate N 2 vs. O/H relationship (Pettini et al. 2005)
Beyond [NII]/Ha: All the lines O HN • Deriving Oxygen Abundance from [NII]/Ha relied on several assumptions • Physical conditions appear different. Observations of full set of lines important for constraining sfr in high redshift objects • GNIRS/Gemini (van Dokkum et al. 2005)
Summary • Lots of new results from UV-selected surveys, including • At z~3, we have detected leaking ionizing radiation from individual galaxies. Global contribution from LBGs may be lower than previous estimates. • At z~2, using IR imaging (J, K, IRAC) and spectroscopy, we have measured a strong correlation between stellar mass and gas phase metallicity. This arises from increase in metallicity as gas fraction decreases, modulated by metal loss at all masses (high mass too!). • Physical conditions may be different in high-redshift star-forming regions!
Q 1700+64: Distribution of M* 6/72 have M*>1011 M & contain 50% of the mass (Steidel et al. 2005) • Mstar, med=2. 0 x 1010 Msun • Distributions with and without IRAC data very similar (uncertainties smaller with IRAC data) • Stellar mass density of z~2 BX/MD galaxies is ~108 Msun. Mpc-3 16% of z=0 stellar mass density (Cole et al. 2001) • caveats: IMF, bursts • relate to z~3, evolution (? ? Weak constraints) • Better constraints for most massive gals (8%)
Q 1700+64 Field • QSO field (V~16, zem=2. 72), very sensitive obs. of IGM • Optical, near-IR imaging and spectra • Spitzer/IRAC photometry 3. 6 -8. 0 m (Barmby et al. 2004) • BIG ? ? ? : HOW MUCH DOES SPITZER HELP US? • 72 gals at z=1. 4 -2. 9 (most are at z<=2. 5) with optical-IRAC SEDs and redshift IDs--> stellar populations, masses • BC 2003 exp(-t/ ) and CSF models, Calzetti extinction • Serendipity: Large redshift spike at z=2. 299+/-0. 015 (20 galaxies) --> study properties vs. environment
Q 1700+64: Distribution 6/72 have M*>1011 M & contain 50% of the mass (Steidel et al. 2005) • Mstar, med=2. 0 x 1010 Msun • Distributions with and without IRAC data very similar (uncertainties smaller with IRAC data) • Stellar mass density of z~2 BX/MD galaxies is ~108 Msun. Mpc-3 16% of z=0 stellar mass density (Cole et al. 2001) • caveats: IMF, bursts • relate to z~3, evolution (? ? Weak constraints) • Better constraints for most massive gals (8%)
Q 1700+64: M/L (optical, near-IR) • At z~2, even mid-IR observations not direct tracers of stellar mass • Inferred stellar mass can differ by factor of ~10 at a given 4. 5 m luminosity, ~70 at Kband (Steidel, Shapley et al. 2005)
Q 1700+64: Weak Constraints 6/72 have M*>1011 M & contain 50% of the mass Low-mass Median High-mass • M/L related to relative amount of rest-UV to opt/IR • For low-mass (young) and typical galaxies, very weak constraints on sf-histories, and therefore ages, E(B-V), sfr • More robust constraints for most massive galaxies (Shapley et al. 2005)
Q 1700+64: Six M*>1011 Gals: IRAC (4. 5 m) (Steidel et al. 2005) • UV-selection DOES find massive galaxies • Most of sample, no constraints on , but these must have >=0. 2 Gyr • Best-fit implies t=2 -3 Gyr, already massive at z~3 • z=2. 2 -2. 4, n~2 x 10 -4 /Mpc 3, • Descendants of z~3 SMG? LBG? • Progenitors of old, red galaxies at <z>=1. 7 (Cimatti, Mc. Carthy)?
Q 1700+64: The z=2. 3 Spike • Only identified ~25% of galaxies in the field -> >80 galaxies in the spike • Galaxy redshift overdensity, d=6 • <Mstar> in spike is x 2 larger than outside • <Age> in spike is 1. 4 Gyr, whereas it is 0. 7 Gyr outside • zform earlier in spike?
Q 1700+64: Implications • Typical stellar mass of z~2 UV-selected galaxy, 2 x 1010 M , z~2 sample contains ~16% of z=0 stellar mass density • Weak constraints on SF-histories for most of population. Timescales are TOUGH. Masses are EASIER (assuming no past burst). • Typical galaxies could be hiding upper limit 2 -3 times stellar mass in old burst • Clustering strength indicates that UV-selected z~2 gals are hosted by the descendants of DM halos hosting z~3 LBGs (Adelberger et al. 2004) • Lack of constraints on timescales (and possible hidden mass from past bursts) makes it difficult to determine if z~2 galaxies are the direct descendants of LBGs, based on stellar populations. With CSF (longest timescale for given set of colors), >50% had not turned on by z=2. 7 (end of LBG era) may imply shorter, episodic timescale for SF.
Q 1700+64: Implications • UV-selection translates into a large range of inferred stellar masses and star-formation histories. High mass tail, 8% with M>1011 M , t~1 -3 Gyr, n=2 x 10 -4/Mpc 3, can’t hide much mass in such galaxies, already have the masses of large elliptical galaxies and are metal rich • Massive UV-selected galaxies are old, were already massive during z~3 LBG era, could have been SMGs, can become reddead gals by z=1. 7 (Cimatti, Mc. Carthy) (right # density) • Proto-cluster enables study of galaxy properties vs. environment
z~3 LBG Metallicities: R 23 • Measure [OIII] and Hb in the K-band with Keck. II/NIRSPEC • Measure [OII] in the Hband • Relative flux-calibration is difficult, dust extinction • Measurements for fewer than 10 gals at z>2 (Pettini et al. 2001; Kobulnicky & Koo 2000)
z~3 LBG Metallicities: R 23 LBG nebular O/H ~ 0. 1 -1. 0 Zsun, 2 -4 mags brighter than local L-Z relation (Pettini et al. 2001)
[NII]/Ha ratios: z~2 metallicities O 3 N 2=log([NII] 6584/Ha) 12+log(O/H)=8. 73 -0. 32 x. O 3 N 2 s~0. 14 • at O/H greater than 0. 25 solar O 3 N 2 index is useful • N 2 increases as O 3 decreases, so very sensitive to O/H (Pettini & Pagel 2004)
Schematic model of superwinds • Collective effect of multiple SNe: mechanical energy input causes shock-heated expanding superbubble with T~108 K, expands, entrains cold ISM (radiation pressure) • Superwind seen in local starbursts with S>0. 1 Msun/year/kpc 2 (Heckman et al. 1990, Heckman 2002)
Outflows: Local Starbursts M 82 • Nearby starburst galaxy • Red is Ha emission from ionized gas above the plane of the galaxy • outflow speed: >500 km/s • plenty of detailed spatial information (Subaru Image)
Importance of feedback/outflows • Enrichment and heating of IGM • Enrichment and entropy floor in ICM • Problems in models of galaxy formation: overcooling and angular momentum • Feedback often invoked as solution, but crudely modeled in both semi-analytic and numerical codes • Observationally important to constrain mass, energy, metals escaping, and how these depend on galaxy properties
Typical L* LBG Spectrum Unsmoothed Smoothed by 5 pix RAB=24 -25. 5 1. 5 hr exposure S/N ~ 2/pix, 9 -12 AA resolution
Deep in the Desert • 10 hr LRIS-B integration for 25 obj, 20 hr for 5 obj • 4 -5 AA resolution, S/N=20 -25 / res. element • IMF, stellar metallicity, outflow properties (c. B 58)
Deep in the Desert • 10 hr LRIS-B integration for 25 obj, 20 hr for 5 obj • 4 -5 AA resolution, S/N=20 -25 / res. element • IMF, stellar metallicity, outflow properties (c. B 58)
Deep in the Desert • 10 hr LRIS-B integration for 25 obj, 20 hr for 5 obj • 4 -5 AA resolution, S/N=20 -25 / res. element • IMF, stellar metallicity, outflow properties (c. B 58)
Implications of Winds • ICM metals in clusters of galaxies: these are after all the progenitors of rich environments today… • “Entropy floor” in galaxy clusters? • Feedback regulates star formation, suppresses subsequent star formation while wind-affected regions cool (~1 -2 Gyr) • Suppress formation of small galaxies within ~200 kpc of large ones? . • The existence of winds is not surprising—what may be surprising is their strength and the suggestion that relatively massive galaxies (as opposed to dwarfs) may dominate the effects on the IGM, and not all of the effects are confined to very high redshifts (z>>3).
Measuring Redshifts: z~3 High redshift • Lya em/abs, IS abs at z>2. 5 • At z =1. 4 -2. 5, these features are in the near UV, while strong restframe optical emission lines have shifted into the near-IR
Redshift Desert Low redshift • Emission-line z • [OII], [OIII], Hb, Ha • At z > 1. 4, [OII] moves past 9000 AA, while Lya below 4000 AA at z<2. 3 SDSS galaxy at z=0. 09
Redshift Desert Low redshift • Abs-line z • 4000 AA break, Ca H&K, Mg • At z > 1. 4, 4000 AA break moves past 9000 AA SDSS galaxy at z=0. 38
Keck/LRIS-B Efficiency • LRIS-B + 400/3400 grism --> 40% efficiency from 3800 -5000 AA Unsmoothed • Low night-sky background in near -UV (~3 mag fainter than at 9000 AA) (Steidel et al. 2004)
Measuring Redshifts: z~2 • Low- and highionization outflow lines • He II emission, CIII] emission • Fewer galaxies have Lya emission (57% have no Lya) than in z~3 sample (cf. SMG!) (Steidel et al. 2004)
Measuring Redshifts: z~2 • Low- and highionization outflow lines • He II emission, CIII] emission • Fewer galaxies have Lya emission (57% have no Lya) than in z~3 sample (cf. SMG!) (Steidel et al. 2004)
Redshift Distributions LBG: z~3 (940) SLBG=1. 7/arcmin 2 n. LBG=1. 4 x 10 -3 Mpc-3 BX: z=2 -2. 5 (816) SBX=5. 2/arcmin 2 n. BX=2. 0 x 10 -3 Mpc-3 Unsmoothed BM: z=1. 5 -2. 0 (118) SBM=3. 8/arcmin 2 n. BM=1. 7 x 10 -3 Mpc-3 750 gals at z=1. 4 -2. 5 BX/BM/LBG with R<=25. 5 (Steidel et al. 2004; Adelberger et al. 2004)
Direct Abundance Determination Z=1/25 Zsun • Use [OIII] 4363/(5007, 4959) to get Te, [SII] to get ne • Problem: 4363 weak, even in local low-Z gals; star-forming gals are not very metal-poor--NO HOPE at high redshift (figure from van Zee 2000)
Abundance Indicators: Bright (R 23) • High-Z branch: R 23 decreases as Z increases • Low-Z branch: R 23 decreases as Z decreases • Uncertainty in which branch R 23 corresponds to (Kobulnicky et al. 1999) • Systematic differences from direct method
- Slides: 80