The challenge of studying the interstellar medium in z~7 galaxies
Kirsten K. Knudsen Chalmers University of Technology (Gothenburg, Sweden)
Collaborators: Darach Watson, Johan Richard, Lise Christensen, Jean-Paul Kneib, Mathilde Jauzac, Benjamin Clement, Anna Gallazzi, Michal Michalowski, David Frayer, Jesus Zavala, Lukas Lindroos, Guillaume Drouart, Suzy Jones, et al. et al... e.g. Bouwens et al. 2011 z = 7.085; Mortlock et al. 2011 10 11 12 13 LFIR ~ 10 10 10 10 L⊙
SFR ~ 1 10 100 1000 M⊙/yr Galaxy cluster field A1689
A1689-zD1, z = 7.5
-0.0065 -0.0042 -0.0019 0.0005 0.0028 0.0051 0.0074 0.0097 0.0121 0.0144 0.0167
-0.0065 -0.0042 -0.0019 0.0005 0.0028 0.0051 0.0074 0.0097 0.0121 0.0144 0.0167 LETTER doi:10.1038/nature14164
A dusty, normal galaxy in the epoch of reionization
Darach Watson1, Lise Christensen1, Kirsten Kraiberg Knudsen2, Johan Richard3, Anna Gallazzi1,4 & Michał Jerzy Michałowski5
Candidates for the modest galaxies that formed most of the stars in As part of a programme to investigate galaxies at z . 7 with the the early Universe, at redshifts z . 7, have been found in large num- X-shooter spectrograph on the Very Large Telescope, we observed the bers with extremely deep restframe-ultraviolet imaging1. But it has candidate high-redshift galaxy, A1689-zD1, behind the lensing galaxy proved difficult for existing spectrographs to characterize them using cluster Abell 1689 (Fig. 1). The source was originally identified10 as a can- their ultraviolet light2–4. The detailed properties of these galaxies didate z . 7systemfromdeepimagingwiththeHubbleandSpitzerspace could be measured from dust and cool gas emission at far-infrared telescopes, with photometry fitting suggesting that it is at z 5 7.6 6 0.4. wavelengths if the galaxies have become sufficiently enriched in dust The galaxy is gravitationally magnified by a factor of 9.3 by the galaxy and metals. So far, however, the most distant galaxy discovered via cluster10. Although it is intrinsically faint, because of the gravitational its ultraviolet emission and subsequently detected in dust emission is amplification, it is one of the brightest candidate z . 7 galaxies known. only at z 5 3.2 (ref. 5), and recent results have cast doubt on whether The X-shooter observations were carried out on several nights between dust and molecules can be found in typical galaxies at z $ 76–8. Here March 2010 and March 2012 with a total time of 16 h on target. we report thermal dust emission from an archetypal early Universe The galaxy continuum is detected and can be seen in the binned spec- star-forming galaxy, A1689-zD1. We detect its stellar continuum in trum (Fig. 2). The Lya cutoff is at 1,035 6 24 nm and defines the redshift spectroscopy and determine its redshift to be z 5 7.5 6 0.2 from a to be z 5 7.5 6 0.2. It is thus one of the most distant galaxies known spectroscopic detection of the Lyman-a break. A1689-zD1 is rep- so far to be confirmed via spectroscopy, and the only galaxy at z . 7 resentative of the star-forming population during the epoch of where the redshift is determined from spectroscopy of its stellar con- reionization9,withatotalstar-formationrateofabout12solarmasses tinuum. The spectral slope is blue; using a power-law fit F l–b, l / per year. The galaxy is highly evolved: it has a large stellar mass and b 5 2.0 6 0.1, where l is the wavelength and Fl isthe flux per unit wave- is heavily enriched in dust, with a dust-to-gas ratio close to that of length. The flux break is sharp, and greater than a factor of ten in depth. the Milky Way. Dusty, evolved galaxies are thus present among the In addition, no line emission is detected, ruling out a different redshift fainter star-forming population at zA1689-zD1:. 7. Dustsolution at z for ~ the 7.5 galaxy. Line emission is excluded to lensing-corrected
RESEARCH LETTER
2 1 0
(arcsec) –1 –2
1.0
0.8
0.6 ) –1
Å 0.4 –2 cm
–1 0.2
erg s erg 0.0 5″ 10″ –19 10
× –0.2
Flux ( 0.50 Figure 1 | The gravitationally lensing galaxy cluster Abell 1689. The colour (1.3699 3 1.1599) is shown, bottom left. Images and noise mapsError were spectrum image is composed with Hubble Space Telescope filters: F105W (blue), F125W primary-beam0.25 corrected before making the signal-to-noise ratio (SNR) maps. (green) and F160W (red). The zoomed box (499 3 499) shows A1689-zD1. Slit positions0.00 for the first set of X-shooter spectroscopy are overlaid in magenta 3 4 4 4 4 4 4 Contours indicate far-infrared dust emission detected by ALMA at 3s,4s, (dashed boxes indicate8.0 × 10 the dither),1.0 × 10 while1.2 × the 10 parallactic1.4 × 10 angle1.6 × 10 was used1.8 × in10 2.0 × 10 and 5s local rms (yellow, positive; white, negative). The ALMA beam the remaining observations (pink dashed lines).Wavelength (Å) Figure 2 | Spectrum of A1689-zD1. The binned one-dimensional (middle the best-fit SED (blue line) are shown. The Lya break is close to the 1Dark Cosmology Centre, NielsWatson, Bohr Institute, UniversityChristensen, of Copenhagen, Knudsen Juliane Maries et Vej al. 32, København2015,panel) and two-dimensional Nature Ø, 2100, Denmark. (upper panel;2Department wavelength of versus Earth distance and Space along Sciences,spectrograph’s Chalmers near-infrared University (NIR)/visual of (VIS) arm split; however, the break is 3 the slit) spectra are shown, with the 68% confidence uncertainty for the clearly detected in the NIR arm alone. A nearby galaxy (z < 2) visible in the Technology, Onsala Space Observatory, SE-439 92 Onsala, Sweden. Centre de Recherche Astrophysiqueone-dimensional de Lyon, spectrum Observatoire in the bottom de Lyon, panel. Universite The redshift´ Lyonz 5 1,7.5 9 Avenue is Charlesbottom Andre part´,69561SaintGenisLaval of the two-dimensional spectrum is detected in both the VIS Cedex, France. 4Istituto Nazionale di Astrofisica-Osservatorio Astrosico di Arcetri, Largo Enrico Fermidetermined 5, 50125 from Firenze, the Lya Italy.break5The at 1,035 Scottish nm. Sky Universities absorption (greyPhysics bands) Alliance, and Instituteand the NIR for arms. Astronomy, University of Edinburgh, Royal Observatory, Edinburgh, EH9 3HJ, UK. depths of 3 3 10219 erg cm–2 s–1 (3s) in the absence of sky emission function at this redshift, meaning that it is among the faint galaxies that lines, making this by far the deepest intrinsic spectrum published of dominate star formation at this epoch9. an object from the epoch of reionization,19 highlighting MARCH the 2015 difficulty | of VOLMosaic 519 observations | NATURE of the | lensing 327 cluster were obtained with the ©2015 Macmillan Publishersobtaining Limited. ultraviolet All rights redshifts reserved for objects at this epoch that are not strongly Atacama Large Millimetre Array (ALMA) in Cycles 0 and 1 with the dominated by emission lines. The restframe equivalent width limits receivers tuned to four 3.8 GHz frequency bands between 211 GHz and are ,4A˚ for both Lya and C III] 1,909 A˚ . Our search space for Lya is 241 GHz. A1689-zD1 is located towards the northern edgeofthe mosaic largely free of sky emission lines; they cover 16% of the range. and is detected at 5.0s with an observed flux of 0.61 6 0.12 mJy in the Fits to the galaxy’s spectral energy distribution (SED) yield a lensing- combined image and at 2.4–3.1s significance in each of the three indi- 9 corrected stellar mass of 1.7 3 10 solar masses (M[) (that is, log(Mw/ vidual observations (Fig. 3). A1689-zD1 is located within the primary z0:15 M[) 5 9:23{0:16), with a best-fitting stellar age of 80 million years beam’s full-width at half-maximum (FWHM) of one pointing and the z0:26 (that is, a light-weighted age t of log[t (yr)] 5 7:91{0:24). The lensing- sensitivity (root mean square, rms) around its position is 0.12 mJy per 10 corrected ultraviolet luminosity is about 1.8 3 10 L[, where L[ is the beam, 42% of the sensitivity of the deepest partof the mosaic. The source solar luminosity, resulting in a star-formation rate (SFR) estimate of is the brightest in the mosaic area of 5 square arcminutes. It coincides 21 2.7 6 0.3 M[ yr based on the ultraviolet emission and uncorrected with the ultraviolet position of A1689-zD1 and is 1.599 away from the for dust extinction, for a Chabrier initial mass function11. A1689-zD1 is next-nearest object in the Hubble Space Telescope image, which is not thus fainter than the turnover luminosity, L*, in the galaxy luminosity detected in the ALMA map. No line emission is convincingly detected
Combined 319-1 319-2 261
Figure 3 | ALMA SNR maps of A1689-zD1. Contours are SNR 5 5, 4, 3, 2 observation 2012.1.00261.S. A1689-zD1 is detected from left to right, at 5.0s, (black, solid), –3, –2 (white, dashed). Images and noise maps were primary- 2.4s, 3.1s, and 3.0s. Natural weighting was used and the visibilities were beam corrected before making SNR maps. Beam sizes are shown at the bottom tapered with a 199 circular Gaussian kernel, resulting in beams of left of each panel. Panels are 899 3 899. The panels show from left to right: 1.3699 3 1.1599, 1.1999 3 1.0999, 1.4399 3 1.1299, 1.4399 3 1.1799 from left the combined data, the two tunings of observation 2011.0.00319.S and to right.
328 | NATURE | VOL 519 | 19 MARCH 2015 ©2015 Macmillan Publishers Limited. All rights reserved More ALMA data: structure - merger or proto-disc?
Knudsen, et al, 2017 More ALMA data: structure - merger or proto-disc? Using UVMULTIFIT:
Two circular Gaussians FWHM ~ 0.5"-0.6" Corrected for lensing: ~ 0.45kpc x 1.9kpc NE
SW
Knudsen, et al, 2017 A1689-zD1: SED
CMB effects: T~40K, beta=1.75: Band 7: 8% , Band 6: 17%
11 LFIR ~ 1.8x10 Lo
SFR(total) ~ 13 Mo/yr log(Mstellar/Mo) ~ 9.3 (+/— 0.13) log(Mdust/Mo) ~ 7.2-7.6 Knudsen et al, 2017 A2744_YD4: pushing to even higher redshift
The Astrophysical Journal Letters, 837:L21 (z6pp )=, 2017 8.38 March 10 Laporte et al. The Astrophysical3. Spectroscopic Journal Follow-up Letters, 837:L21 (6pp), 2017 March 10 Laporte et al. 3.1. X-shooter Observations Given the importance of confirming the presence of dust How manyemission more beyond ofz such8, we undertook systems?? a spectroscopic campaign using X-Shooter/VLT (ID: 298.A-5012, PI: Ellis). Between 2016 November 24 and 27, we secured 7.5 hr on- source integration with excellent seeing (»0.6 arcsec). We used a 5 arcsec dither to improve the sky subtraction and aligned the slit so that a brighter nearby source could verify the blind offset (see Figure 2). The data were reduced using v2.8 of the ESO Reflex software combined with X-Shooter pipeline recipes v2.8.4. We visually inspected all three arms of the X-Shooter (UVB, VIS, NIR) spectrum and identified one emission line at λ=11408.4 Å with an integrated flux of f=1.82±0.46× 10−18 erg s−1 cm−2. By measuring the rms at adjacent wavelengths, we measure the significance as ≈4.0σ. We checked the reliability of the line by confirming its presence on two independent spectral subsets spanning half the total exposureFigure time 1. ALMA(Figure Band4). These 7 continuum half-exposures detection show for the A2744_YD4. line (left) Map combining all frequency channels; (middle left and middle right) independent maps for two fi withequal signi frequencycances of ranges. 2.7 and Contours 3.0, consistent are shown with that at 1, of 2, the 3, 4, and 5σ adopting a noise level from an area of 0.5×0.5 arcmin. (right) HST F160W image with combined Figure 2. Position of the ALMA band 7 detected high-redshift galaxy total exposure. No further emission lines of comparable A2744_YD4 (blue) with respect to other group members (yellow) suggested by signiALMAficance image were contoursfound. We overplotted. explore two interpretations of Zheng et al. (2014). The X-shooter slit orientation is shown with the dashed white line. Although one other member of the group was targeted in the SFR(total)this line at 11408 Å. It~ is either20( 1M) oneo/yr component of the [O II] exposure, no confirming features were found in the data. doubletand atH a redshift= 70 zkm 2.06 s−1, orMpc(2) Ly−1α),at andz= all8.38. magnitudes are quoted in IRAC data obtained in channels 1 (l ~ 3.6 μm) and 2 For (1),0 depending on which9 component of the [O II]λ3727, c M3730thestellar doublet AB ~ system is detected,2x10(Oke we expectM &o Gunn a second1983 line). at either (lc ~ 4.5 μm) with 5σ depths of 25.5 and 25.0, respectively, 11416.9 Å or 11399.8 Å. No6 such emission is detected above carried out under DDT program (ID: 90257, PI: T. Soifer). We Mthedust 1σ flux ~ limit 6x10 of 4.6×10 −M19 cgs.o This would imply a flux ratio for the two components of »2.3.95 Imaging(2.02) at 1( Data2)σ, greater extracted the HST photometry on PSF-matched data using than the range of 0.35–1.5 from theoretical studies (e.g., SExtractor (Bertin & Arnouts 1996) v2.19.5 in double image PradhanHere, et al. 2006 we). describe theLaporte ALMA data et al. in 2017 which a high-z mode using the F160W map for the primary detection Forcandidate(2), although is the detected line is somewhat at 0.84 narrow mm for and Lyα the(rest- public imaging data -1 (Figure 1). To derive the total flux, we applied an aperture frameused width to≈ constrain20 km s ), its its equivalent spectral width energy deduced distribution from (SED). the line flux and the F125W photometry is 10.7±2.7 Å, correction based on the F160W MAG_AUTO measure (see, consistent with the range seen in other z > 7 spectroscopically e.g., Bouwens et al. 2006). The noise level was determined confirmed sources (2.1.Stark Deep et al. 2017 ALMA). We detect Band no 7flux Observations above the noise level at the expected position of either the CIV and using several 0.2 arcsec radius apertures distributed around the [O III]Adoublets deep ALMA at this redshift. Band However, 7 map ( atID the 2015.1.00594, expected PI: Laporte) source. The total Ks magnitude of 26.45±0.33 was obtained position of the C III] doublet, we notice a very marginal (≈2σ) 19 1 2 using a 0.6 arcsec diameter aperture applying the correction featureof the at λ FF=17914.7 clusterÅ ( Abell7.5 ± 0.35 2744×10− centerederg s− cm at− 0.84) mm ( fc=356 seenGHz on two) was individual observed sub-exposures. on 2016 If this July is C duringIII]λ1907 2.5Å hr. The data were estimated in Brammer et al. (2016). The uncertainty was (normally the brighter component) at z 8.396, the Lyα reduced using the CASA pipelineC III= (McMullin et al. 2007) with estimated following a similar procedure to that adopted for the offset of 338±3 km s−1 would be similar to that for a Figure 3. Spectral energy distribution of A2744_YD4. The red curve shows HST data. The Spitzer data were reduced as described in fi +0.09 z=a7.73 natural galaxy weighting(Stark et al. and2017 a). pixel The other size component of 0 04. Figure 1 reveals a the best- tting SED found by Hyperz with zphot=8.42-0.32. The black curve −19 −1 −2 shows a forced low-redshift solution derived when only a redshift interval from wouldsource be fainter with than greater 5.0× than10 erg4.0s signicm ,fi consistentcance with a peak flux of Laporte et al. (2014) using corrected Basic Calibrated Data 0 to 3 is permitted. This has a likelihood >20 times lower. The inner panel with9.9 theoretical±2.3 studies×10 of− this5 Jy doublet/beam(e.g.,( Rubinuncorrected et al. 2004). for magnification). (cBCD) and the standard reduction software MOPEX to displays the redshift probability distribution. Previous spectroscopy of A2774_YD4 was undertaken by fi theThe GLASS uncertainty survey (Schmidt and et al.signi2016fi)cance, who place level a 1σ wasupper computed from the process, drizzle, and combine all data into a nal mosaic. As limitrms on anymeasured Lyα detection across at 4.4 a× representative10−18 erg s−1 cm≈−2,2»×2.42 arcmin field. The shown in Figure 2, four other galaxies are close to We also made use of the Easy and Accurate Zphot from Yale times above our X-Shooter detection. (EAZY; Brammer et al. 2008) software. The SED fits adopted signal is seen within two independent frequency ranges (center A2744_YD4, but only the other source within the X-shooter the standard SED templates from EAZY, as well as those from fi slit is comparably bright to A2744_YD4. We used GALFIT panels in Figure3.2. ALMA1) and Observations the signi cance level is comparable to the galaxy evolutionary synthesis models (GALEV; Kotulla that claimed for Watson et al’s z ~ 7.5 lensed system, although (Peng & Ho 2002) to deblend the two sources and to measure et al. 2009) including nebular emission lines as described by Only a few far-infrared emission lines lines are detectable for fl fi Anders & Fritze-v. Alvensleben (2003). Adopting a large sourcesits observed in the reionizationflux isera six(see, times e.g., Combes fainter.2013) Taking. Only into account the their IRAC uxes. We tted both IRAC ch1 and ch2 images redshift range (0 <