z ∼ 7 GALAXIES IN THE HUDF: FIRST EPOCH WFC3/IR RESULTS^*

While great progress has been made in the study of galaxies up to z ∼ 6, even after more than a decade of Lyman break galaxy (LBG) surveys, the samples of UV-selected candidate z ≳ 7 galaxies are still very modest (e.g., Bouwens et al. 2004b, 2008; Yan & Windhorst 2004; Bradley et al. 2008; Richard et al. 2008; Oesch et al. 2009a; Zheng et al. 2009; Ouchi et al. 2009; Castellano et al. 2009). Due to their small number, the z ≳ 7 luminosity function (LF), and in particular its faint-end slope, is still very uncertain. An accurate measurement of this is of fundamental importance, however, not only for the early stages of galaxy formation, but also for the understanding of the epoch of reionization and the role of galaxies therein.

For sources of such early cosmic times the intergalactic medium absorbs most of the light blueward of Lyα, which is redshifted into the near-infrared (NIR), where the sensitivities and fields of view of the detectors have lagged behind the optical CCD cameras. Extremely deep NIR imaging is needed to detect these galaxies. Moreover, due to limited spatial resolution of earlier IR cameras, it has not been possible so far to measure other basic quantities of z ∼ 7 galaxies such as sizes.

With the installation of the Wide Field Camera 3 (WFC3) in Service Mission 4 of the Hubble Space Telescope (HST) a new era of high-redshift galaxy surveys has begun. With its large area, high sensitivity, and small pixel size, it allows us to probe faint objects much more efficiently than HST's previous NIR camera Near Infrared Camera and Multi-Object Spectrometer (NICMOS). Additionally, WFC3's improved filter set over NICMOS helps us to exclude contaminating interloper objects and to estimate more precise photometric redshifts for these sources. Furthermore, it allows for the first time the detection of galaxies at z ∼ 8, which we report on in an accompanying paper (Bouwens et al. 2009b).

In this Letter, we present early results on our search for z ∼ 7 galaxies from the first epoch data of the HUDF09 survey. In Section 2, we describe the data, the z ∼ 7 candidate selection and its efficiency. In Section 3, we present new constraints on the LBG LF at z ∼ 7. We adopt Ω_M = 0.3, Ω_Λ = 0.7, H₀ = 70 km s⁻¹ Mpc⁻¹, i.e., h = 0.7. Magnitudes are given in the AB system (Oke & Gunn 1983).

2.1. The HUDF09: First Epoch Data

2.2. z₈₅₀-dropout Candidate Selection

The summed J₁₂₅+Y₁₀₅ image is used to detect galaxies with the SExtractor program (Bertin & Arnouts 1996) and to measure their colors in isophotal apertures after the optical images have been PSF-matched to the WFC3/IR exposures. Total magnitudes are measured in 2.5 Kron (AUTO) apertures and small corrections of 0.1 mag have been applied to account for flux loss in the wings of the PSF.

Only sources with signal-to-noise ratio (S/N) larger than 5 within 025 radius apertures in both J₁₂₅- and Y₁₀₅-images are considered. We have ensured that our weight maps correctly reproduce the appropriate noise properties in these apertures and measured fluxes <1σ were replaced by a 1−σ upper limit.⁹

Galaxies are selected from the SExtractor catalog based on the Lyman Break technique (e.g., Steidel et al. 1996; Giavalisco et al. 2004; Bouwens et al. 2007) requiring

$$\begin{eqnarray*} &&\begin{array}{c} \dsty (z_{850}-Y_{105})>0.8 \\ \ms \dsty (z_{850}-Y_{105})>0.9 + 0.75 (Y_{105}-J_{125}) \\ \ms \dsty (z_{850}-Y_{105})>-1.1 + 4(Y_{105} - J_{125}) \\ \ms \dsty {\rm S}/{\rm N}(J_{125}) > 5 \wedge {\rm S}/{\rm N}(Y_{105})>5 \\ \ms \dsty {\rm S}/{\rm N}(V_{606}) <2 \wedge {\rm S}/{\rm N}(i_{775}) <2.\end{array} \end{eqnarray*}$$

These criteria select galaxies from z ∼ 6.4–7.3 with a median redshift of 〈z〉 = 6.8; see the inset of Figure 1.

Zoom In Zoom Out Reset image size

After rejecting spurious sources such as diffraction spikes of stars and one probable supernova, we find 16 z ∼ 7 candidates between J₁₂₅=26.0and29.0 mag. Their properties are listed in Table 1, and an image of all candidates is shown in Figure 2. As can be seen from Table 1, the J₁₂₅ − H₁₆₀ colors of these sources are extremely blue, in particular toward faint J₁₂₅ magnitudes, which results in UV-continuum slopes as steep as β ≲ −2.5. This could imply that these sources are dominated by very metal-poor stars with little to no dust extinction, or alternatively exhibit initial mass functions skewed toward massive stars. All these possibilities would have strong implications for the ionizing flux produced by these galaxies, which may be larger than what is assumed in current calculations of reionization (see Bouwens et al. 2009d).

Zoom In Zoom Out Reset image size

Table 1. Photometry (AB mag) of the z ∼ 7 z₈₅₀-dropout Sources^a

ID	α	δ	J125	z850 − Y105	Y105 − J125	J125 − H160	S/NJ125	Reference
UDFz-42566566	03:32:42.56	−27:46:56.6	25.95 ± 0.04	1.48 ± 0.14	0.16 ± 0.04	−0.08 ± 0.04	25.8	1, 2, 3, 4, 5, 6
UDFz-44716442	03:32:44.71	−27:46:44.2	26.90 ± 0.10	>2.43	0.66 ± 0.12	−0.25 ± 0.09	17.2	×
UDFz-38807073	03:32:38.80	−27:47:07.3	26.90 ± 0.06	2.34 ± 0.61	0.41 ± 0.08	0.14 ± 0.06	26.2	1, 2, 4, 5, 6, 7
UDFz-39557176	03:32:39.55	−27:47:17.6	27.19 ± 0.08	1.95 ± 0.70	0.51 ± 0.12	0.04 ± 0.09	14.8	2, 5
UDFz-42577314	03:32:42.57	−27:47:31.4	27.22 ± 0.20	1.61 ± 0.33	0.30 ± 0.14	−0.16 ± 0.12	11.1	2, 5, 6
UDFz-39586565	03:32:39.58	−27:46:56.5	27.74 ± 0.18	0.99 ± 0.27	−0.11 ± 0.15	−0.32 ± 0.18	9.3	−
UDFz-37228061	03:32:37.22	−27:48:06.1	27.76 ± 0.14	>2.18	0.46 ± 0.16	−0.08 ± 0.13	11.3	−
UDFz-43146285	03:32:43.14	−27:46:28.5	27.81 ± 0.20	>2.13	0.27 ± 0.16	0.04 ± 0.14	10.3	×
UDFz-36777536	03:32:36.77	−27:47:53.6	27.83 ± 0.16	1.06 ± 0.23	−0.23 ± 0.11	−0.13 ± 0.13	13.5	−
UDFz-37446513	03:32:37.44	−27:46:51.3	27.86 ± 0.16	1.05 ± 0.27	−0.11 ± 0.13	−0.06 ± 0.14	11.4	−
UDFz-40566437	03:32:40.56	−27:46:43.7	27.97 ± 0.17	1.32 ± 0.40	−0.07 ± 0.15	−0.43 ± 0.20	7.7	−
UDFz-41057156	03:32:41.05	−27:47:15.6	28.08 ± 0.16	1.46 ± 0.51	0.18 ± 0.16	−0.16 ± 0.16	10.7	−
UDFz-36387163	03:32:36.38	−27:47:16.3	28.27 ± 0.18	0.98 ± 0.29	−0.07 ± 0.15	−0.12 ± 0.17	9.8	−
UDFz-37807405	03:32:37.80	−27:47:40.5	28.36 ± 0.16	>1.97	0.06 ± 0.21	−0.27 ± 0.24	6.8	−
UDFz-39736214	03:32:39.73	−27:46:21.4	28.56 ± 0.20	>1.95	0.04 ± 0.23	−0.41 ± 0.28	5.9	−
UDFz-38537519	03:32:38.53	−27:47:51.9	28.96 ± 0.24	>1.81	0.08 ± 0.26	−0.62 ± 0.36	5.8	−
UDF-34537360b	03:32:34.53	−27:47:36.0	25.96 ± 0.03	3.35 ± 0.57	0.18 ± 0.03	0.21 ± 0.03	50.0

Notes. ^aThe J₁₂₅ photometry was derived from elliptical apertures of 2.5 Kron radii, and has been corrected to total magnitudes by 0.1 mag offsets, while colors were measured in isophotal apertures. Limits correspond to 1σ. ^bThis object is most likely a supernova, as it is very bright and has a stellar profile. It should have been securely detected in the previous NICMOS images, and we therefore exclude it from any further analysis. References. (1) Bouwens & Illingworth 2006; (2) Bouwens et al. 2004b; (3) Yan & Windhorst 2004; (4) Coe et al. 2006; (5) Labbé et al. 2006; (6) Bouwens et al. 2008; and (7) Oesch et al. 2009a: (×) not covered by NICMOS data and (−) no previous detection.

Download table as:  ASCIITypeset image

It is reassuring that all z ∼ 7 galaxy candidates, which have been identified in previous work, are confirmed to be secure high-redshift candidates (see Table 1). We show in Figure 3 a comparison of the NICMOS observations with the WFC3 data for the brightest candidates which were covered by both instruments in order to visualize the enormous improvement in data quality provided by WFC3/IR. The new data allow us to probe much fainter limits. While in previous studies only two galaxies were identified beyond a magnitude of 27.5, of which one is only marginally detected (Bouwens et al. 2008; Oesch et al. 2009a), the current WFC3 sample includes 11 such faint objects, resulting in much better constraints on the LF at z ∼ 7.

Zoom In Zoom Out Reset image size

2.3. Sources of Sample Contamination

As in Oesch et al. (2007, 2009a), completeness, C, and magnitude-dependent redshift selection probabilities, S, for our sample are derived from simulations in which we add artificial galaxies into the real images and rerun our detection and selection procedure. The input SED distribution for these simulations is chosen to be slightly steeper than the measured slopes for z ∼ 6 galaxies (β = −2.2; Stanway et al. 2005; Bouwens et al. 2009c) to account for the very blue colors seen in our sample (β ∼ −2.5, see also Bouwens et al. 2009d), and a log-normal size distribution with a size scaling of (1 + z)⁻¹ (Ferguson et al. 2004; Bouwens et al. 2004a; Oesch et al. 2009b) is assumed.

The LF brightens significantly from z ∼ 6 to z ∼ 4 (e.g., McLure et al. 2009a; Bouwens et al. 2007; Bouwens & Illingworth 2006; Oesch et al. 2007; Yoshida et al. 2006; Ouchi et al. 2004; Steidel et al. 1999), following M_*(z) = −21.02 + 0.36(z − 3.8) (Bouwens et al. 2008), and previous studies suggested a similar trend to z ∼ 7 (e.g., Bouwens et al. 2008; Oesch et al. 2009a; Ouchi et al. 2009). From our simulations we estimate that 31 ± 8 objects should be detected in our small area survey, assuming no evolution of the LF since z ∼ 6 (error bars include uncertainties in the LF parameters). The cosmic variance is expected to add 35%–40% (Trenti & Stiavelli 2008). While the expected number of sources is much larger than the 16 observed, the difference is not yet statistically robust. Clearly, the combination of our survey with wider area data will provide much stronger constraints on the changes from z ∼ 6 to z ∼ 7 in the future.

To better quantify the evolution of the LBG population, we show the stepwise LF based on our data in Figure 4. Absolute magnitudes are computed assuming all candidates to lie at z ∼ 6.8 and the effective volume is estimated as $V_{\rm eff}(m) = \int _0^\infty dz \frac{dV}{dz} S(z,m)C(m)$.

Zoom In Zoom Out Reset image size

Due to the small area of the present data we are not able to fit all parameters of the Schechter function from our data set alone. We therefore include the points from wide area Subaru searches for z ∼ 7 galaxies by Ouchi et al. (2009) in order to constrain the bright end of the LF. The fitting is done as in Oesch et al. (2009a) maximizing the Poissonian likelihood for the observed number of sources in each magnitude bin.

The value of ϕ_* of the UV LF has been found to be remarkably constant with redshift, and we fix its value to ϕ_* = 1.4 × 10⁻³ Mpc⁻³ mag⁻¹ as measured at z ∼ 6 (Bouwens et al. 2007). This results in M_* = −19.91 ± 0.09 and α = −1.77 ±  0.20, which gives a first constraint on the faint-end slope at z ∼ 7. This Schechter function is shown in Figure 4 as a black solid line.

The data are consistent with a faint-end slope that is largely unchanged over the 2.5 Gyr from z ∼ 7 to z ∼ 2 with a value of α ∼ −1.7 (e.g., Reddy et al. 2009; Bouwens et al. 2007; Oesch et al. 2007), and the measured value of M_* is in very good agreement with the extrapolation of Bouwens et al. (2008). Note also that, within the error bars, the LF function parameters are consistent with the ones derived by Ouchi et al. (2009).

The first observations with the WFC3/IR camera have demonstrated the amazing improvement over previous NIR instruments. Nevertheless, they have confirmed our prior NICMOS detections at z ∼ 7. In the first epoch data of the HUDF09, which comprises only about one third of the final data set, we have identified a robust sample of 16 z ∼ 7 galaxy candidates down to flux limits as low as 29 mag. With NICMOS about 100 orbits were needed per z ∼ 7 galaxy candidate found (Bouwens et al. 2009a); WFC3 is ∼40 times more efficient and requires only 2.4 orbits per candidate (for fields with existing deep optical data). This remarkable ability to detect high redshift galaxies extends to z ≳ 8 (see Bouwens et al. 2009b)

Most of the z ∼ 7 candidates appear very compact with an average observed half-light radius of ∼016 (from SExtractor apertures) indicating that the typical size of a starburst galaxy at z ∼ 7 is of the order of ≲1 kpc (for a more extensive analysis of galaxy sizes, see Oesch et al. 2009b).

Due to the depth of our observations we can put a first constraint on the faint end of the LF at z ∼ 7 of α = −1.77 ± 0.20. This is consistent with no evolution over the whole time span from z ∼ 7 until z ∼ 2 and such a steep faint-end slope has strong implications for reionization, providing support to a scenario in which low-luminosity galaxies provide the bulk of the flux for ionizing the universe (see also Oesch et al. 2009a; Kistler et al. 2009). The combined surface brightness of our sources is μ_J ≃ 26.1 mag arcmin⁻² which is brighter than the minimum required for reionization at ∼7 for Population II stars μ_min ≃ 27.2 (Stiavelli et al. 2004a, 2004b). The factor ∼3 difference suggests that even considering an escape fraction ≪1, these galaxies still contribute significantly to reionization.

The full HUDF09 data set, complemented with planned wide area surveys, will allow another leap forward in the measurement of the z ∼ 7 LF in the near future. The HST is again enabling a revolutionary advance in our understanding of the build-up of galaxies in the universe.

We especially thank all those at NASA, STScI, and throughout the community who have worked so diligently to make Hubble the remarkable observatory that it is today. The servicing missions, like the recent SM4, have rejuvenated HST and made it an extraordinarily productive scientific facility time and time again, and we greatly appreciate the support of policy makers, and all those in the flight and servicing programs who contributed to the repeated successes of the HST servicing missions. P.O. acknowledges support from the Swiss National Foundation (SNF). This work has been supported by NASA grant NAG5-7697 and NASA grant HST-GO-11563.01. This research has benefitted from the SpeX Prism Spectral Libraries, maintained by Adam Burgasser at http://www.browndwarfs.org/spexprism.

Facilities: HST (ACS/NICMOS/WFC3) - Hubble Space Telescope satellite