Astronomical Science

Breaking Cosmic Dawn: The Faintest Galaxy Detected by the VLT

Maruša Bradač1 after the Big Bang when early light z > ~ 7 gal­axies using the dropout tech- Eros Vanzella2 sources produced enough energetic nique (Steidel et al., 1996), both in Nicholas Hall1 ­photons to ionise the neutral hydrogen, blank fields (the Hubble Ultra-Deep Field Tommaso Treu3 making the Universe transparent to visi- [HUDF] and the Cosmic Assembly Adriano Fontana4 ble light. This era is referred to as reioni- ­Near-infrared Deep Extragalactic Legacy Anthony H. Gonzalez5 sation and is also the era of the formation Survey [­CANDELS]; e.g., Bouwens et al., Douglas Clowe6 of the first galaxies. But when exactly 2012), and behind galaxy clusters (e.g., Dennis Zaritsky7 did reionisation occur and how long did Hall et al., 2012). One of the most obvious Massimo Stiavelli8 it last? What were the sources responsi- limitations of the dropout technique, how- Benjamin Clément7 ble for ionising the neutral gas? Was it ever, is that unambiguously confirming the first galaxies? Now, for the first time, the object’s redshift usually requires we can peer far enough into space to spectroscopic follow-up. This is hard to 1 Department of Physics, University of detect these galaxies spectroscopically do for the typically faint high-z sources. California, Davis, USA and answer these questions. 2 INAF, Osservatorio Astronomico di However much more important than the ­Trieste, Italy Observations of galaxies at these early redshift confirmation, spectroscopy 3 Department of Physics, University of times are challenging, not only due to ­provides information on properties of the California, Santa Barbara, USA the large distance to these objects, but interstellar and intergalactic media (ISM 4 INAF, Osservatorio Astronomico di also due to their lower luminosity (gal­ and IGM). In particular, Lyman-α emission Roma, Monteporzio, Italy axies had fewer stars when they first from sources close to the reionisation 5 Department of Astronomy, University of formed). The quest to discover the most era is a valuable diagnostic, given that it Florida, Gainesville, USA distant (i.e., first) galaxies has advanced is easily erased by neutral gas within 6 Department of Physics & Astronomy, rapidly in the last decade. With the help and around galaxies. The observed Ohio University, Athens, USA of a galaxy cluster acting as a cosmic strength of Ly-α in distant galaxies is a 7 Steward Observatory, University of ­telescope and magnifying the back- gauge of the time when reionisation was ­Arizona, Tucson, USA ground Universe, we have observed the completed (Robertson et al., 2010). Fur- 8 Space Telescope Science Institute, faintest galaxy ever detected spectro- thermore, we expect Ly-α Emitters (LAEs) ­Baltimore, USA scopically at these extremely large dis- to be predominantly dust-free galaxies;­ tances. It was detected using the FORS2 hence their numbers should increase with spectrograph on the Very Large Telescope redshift until the state of the IGM becomes Astronomers have been debating the (VLT). With an extremely long ex­­­posure neutral, at which point their numbers origin of galaxies for centuries and time (16 hours on a mask, a total of should decline. the timeline of the evolution is now well 22 hours of observing time) and assisted established. However, despite the by nature’s own telescope (The Bullet A powerful way to detect emission lines ­success of galaxy formation models, Cluster) we were able to push to the limits from faint sources is to use galaxy clus- some questions remain largely unan­ of the light-gathering power of the VLT ters as cosmic telescopes (e.g., Treu, swered. Chief among them is how the which allowed us to detect Lyman-α [2010] for a recent review). Gravitational first galaxies formed. When the Uni­ emission from this “normal” (i.e., not ultra- lensing magnifies solid angles while pre- verse was in a highly opaque state, luminous) galaxy (Bradač et al., 2012). serving colours and surface brightness. these first galaxies most likely cleared Thus, sources appear brighter than in the the cosmic fog and broke the cosmic absence of lensing. The advantages of dawn. However, there are large uncer­ Beyond simple counting cosmic telescopes are that we can probe tainties as to how this transition oc­ deeper (due to magnification), sources curred and what the exact role of the Detection of high-redshift galaxies is are practically always enlarged and iden- first galaxies was. Using galaxy clusters ­performed by searching for the redshifted tification is further eased if sources are as cosmic telescopes that magnify Lyman break using broadband photo­ multiply imaged. Typically, one can gain background objects, we were able to metry (Steidel et al., 1996). These so- several magnitudes of magnification, thus detect hydrogen emission from a galaxy called Lyman Break Galaxies (LBGs, see enabling the study of intrinsically lower at a redshift of z = 6.740 ± 0.003 using e.g., Vanzella et al., 2009) are the best luminosity galaxies that we would other- FORS2 on the VLT. To date, this is studied and the largest sample of galax- wise not be able to detect with even the the faintest galaxy with a successfully ies at redshifts z > ~ 5. One can identify largest of current telescopes. measured spectrum and as such z ≈ 5 objects by their non-detection in an important beacon for cosmic dawn. the V-band and blueward: such objects are referred to as V-band dropouts. The combined power of VLT and the ­Similarly, objects at z ≈ 6 are associated “Silver­ Bullet” Cosmic dawn through cosmic telescopes with i-band non-detection, z ≈ 7–8 with z-band, and z > ~ 8 objects would be Our targets were selected from deep The cosmic Dark Ages are thought to observed as J-band dropouts. Substan- Ad­vanced Camera for Surveys (ACS) and have ended around 500 million years tial progress has been made in detecting Wide Field Camera 3 (WFC3) data, both

The Messenger 150 – December 2012 53 Astronomical Science Bradač M. et al., The Faintest Galaxy Detected by the VLT

on the Hubble Space Telescope (HST). Out of the ten z-band dropout candidates Breaking cosmic dawn Ten z-band dropouts were found by targeted, we detected an emission line at Hall et al. (2012). The aim of the VLT pro- 9412 Å with 5σ significance (see Figure 1) While this work presents only a single gramme 088.A-0542 (PI Bradač) was in one target. The line is detected in two spectroscopic detection at z > 6.5, to target all these extraordinary targets ­different FORS2 masks and is broader it nonetheless probes a very important with the FORS2 spectrograph. ­Ordi- than cosmic rays or residuals due to sky region of parameter space. The intrin- narily the hope of detecting any of these subtraction; hence we are confident that sic (unlensed) line flux of this object is sources even with a programme of the line is not an artefact (see Figure 2). ~ 2−3 times fainter than the, until now, 22 hours would be impossible, as their No other emission lines are detected in faintest spectroscopic detection of an intrinsic magnitudes are H > 27 AB mag. the spectrum (7700–10000 Å); if the LAE at z ~ 7 (Schenker et al., 2012). Its However with the help of magnification object were at lower redshift one would intrinsic H(160W)-band magnitude is int from the Bullet Cluster, which is one of have expected detections of several mH(160W) = 27.57 ± 0.17, corresponding to the best cosmic telescopes (due to its lines. Based on this lack of other lines an intrinsic luminosity of 0.5 L* at this red- large mass and elongation, the magnifi- and on the spectral energy distribution fit shift (see Table 1). cations are large), the observed magni- using imaging, we exclude other alter­ tudes of these ten sources are H = 25–27 native explanations and conclude that But more important than the record AB mag. Of course, this does not make the line is most likely Ly-α at z = 6.740 ± breaking is the fact that these sources the observing easy per se! 0.003. are excellent beacons of cosmic

Figure 1. A distant source observed from when the Universe was less than 10% of its current age. The object was found behind the Bullet Cluster. The top row shows an image of the cluster (from HST and WFC3), and a zoom-in on the source in a colour composite from an optical image and an infrared image. The bottom row shows a confirmation spec- trum (from FORS2) of the object (sky emission above, object spectrum with Ly-α emission below). Colour Optical Composite 800–950 nm Infrared 950–1600 nm

SKY 900 nm 980 nm

MASK 1+2 Lyα

Figure 2. Upper: 2-D spectrum of the dropout galaxy (sky emission above, spectra from individual masks MASK 1 below). Lower: 1-D spectrum of the object. The sky spectrum has been rescaled by a factor of 300 and offset for display purposes, and the regions where MASK 2 skylines are more intense have been marked with transparent vertical bars. Strong residuals in the sky MASK 1+2 subtraction are evident and correspond to the more intense sky lines. Note that the detected line is broader than residuals of sky subtraction, confirming –19 Ly the reality of the feature. ) 5 × 10 α /Å 2 spectrum m

/c 0 /s rg (e )

λ –19 f( –5 × 10 sky

9 000 9200 9400 9600 9800 Wavelength (Å)

54 The Messenger 150 – December 2012 ­reionisation. Measuring the rest frame R.A. 104.63015 Table 1. Imaging and spectroscopic properties of z -band dropout #10 equivalent width (EW) distribution of LAEs Dec. –55.970482 850 from Hall et al. (2012). as a ­function of redshift and luminosity mH160W 26.37 ± 0.16 is a powerful tool to study reionisation. mJ110W 26.5 ± 0.3

The EW distribution changes with redshift (J110W — H160W) 0.10 ± 0.15 and source luminosity. Simulations sug- (z850LP — J110W ) 1.57 ± 0.68 a gest that reionisation is the key factor mV606 > 28.75 (texp = 2336s) driving this trend (Dayal & Ferrara, 2012), mi775W > 28.60 (texp = 10150s) because, unlike continuum photons, the mI814W > 29.00 (texp = 4480s)

Ly-α photons that escape the galactic mKs > 26.65 (texp = 3.75hr) environment are attenuated by the neutral μ 3.0 ± 0.2 hydrogen in the IGM. With a measure- int +0.17 mH(160W) 27.57 –0.17 ment of the EW distribution in LAEs we can therefore help distinguish between λ 9412 Å effects of ISM dust and neutral IGM z 6.740 ± 0.003 b –17 2 and study the epoch of reionisation (see f (0.7 ± 0.1 ± 0.3) × 10 erg/s/cm f 3.3+ 1.0 × 10–20 erg/s/cm2/Å also Treu et al., 2012). λ,c – 0.8 f int (0.23 ± 0.03 ± 0.10 ± 0.02) × 10–17 erg/s/cm2 f int 1.1+ 0.4 × 10–20 erg/s/cm2/Å Our source is the faintest (in line flux) λ,c – 0.3 W (Ly- ) 30+12 Å detected thus far and is only the second rest a – 21 firm spectroscopic detection of a sub-L* source at z > 6.5 (the other example is also a lensed galaxy at z = 7.045 discov- Extreme Imaging Reaching Out to Acknowledgements ered by Schenker et al. [2012]). z > ~ 7) will open up new parameter Support for this work was provided by NASA space and will allow us to study the prop- through HST-GO-10200, HST-GO-10863, and HST- erties (e.g., star formation rates and GO-11099 from STScI, from ASI–INAF, from the NSF, Future is made bright by gravitational ­stellar masses) of a large number of gal- Sloan and Packard Foundations as well as KITP. lensing axies at z ~ 7 for the first time. The ­presence (or absence) of an established References With future observations of dropout stellar population will be measured in objects magnified by cosmic telescopes the targets and these findings will allow Bouwens, R. J. et al. 2012, ApJ, 754, 83 we will further increase the sample. As us to identify the dominant sources of Bradač, M. et al. 2012, ApJ, 755, L7 Dayal, P. & Ferrara, A. 2012, MNRAS, 421, 2568 noted above, measuring the EW distri­ the bulk of ionising photons necessary to Hall, N. et al. 2012, ApJ, 745, 155 bution of LAEs as a function of redshift drive reionisation. McLean, I. S. et al. 2010, SPIE, 7735, and luminosity is a very powerful tool to Robertson, B. E. et al. 2010, Nature, 468, 49 study reionisation, because the latter is Furthermore, in the local Universe, Schenker, M. A. et al. 2012, ApJ, 744, 179 Steidel, C. C. et al. 1996, ApJ, 462, L17 likely to be the key factor driving the trend star formation is associated with mole­ Treu, T. 2010, ARA&A, 48, 87 of EW and luminosity. The main missing cular gas, and therefore to understand Treu, T. et al. 2012, ApJ, 747, 27 observational ingredient is a measure- star formation properties we would like to Vanzella, E. et al. 2009, ApJ, 695, 1163 ment of the EW distribution for both lumi- trace molecules in the cold ISM of high nous and sub-L* galaxies at the red- redshift galaxies. ALMA is opening a new shifts of reionisation; future surveys and window for this endeavour. It is designed facilities will achieve exactly that (e.g., to trace molecular gas in galaxies up to X-shooter on VLT and the new spectro- z ~ 5 in the brightest cases. With the graph ­MOSFIRE on Keck; see McLean assistance of gravitational lensing, how- et al., 2010). ever, these limits can be pushed to the redshifts near reionisation. ALMA’s Perhaps even more importantly, however, incredible resolution can also be further we will also add the wavelength cover- improved with lensing, giving us access age to the observations. A first major to the sub-kpc regime. This will allow improvement will be performed with deep us to characterise the spatial distribution Spitzer observations. There is a large and kinematics of the cold ISM in these Exploration Science programme extremely high redshift, but not ultra- approved for Spitzer Cycle 9. The pro- luminous galaxies, something that would gramme (Spitzer UltRaFaint SUrvey - not have been possible without the help SURF’S Up: Cluster Lensing and Spitzer of gravitational lensing.

The Messenger 150 – December 2012 55