Unveiling Far-Infrared Counterparts of Bright Submillimeter Galaxies Using PACS Imaging
We present a search for Herschel-PACS counterparts of dust-obscured, high-redshift objects previously selected at submillimeter and millimeter wavelengths in the Great Observatories Origins Deep Survey North field. We detect 22 of 56 submillimeter galaxies () with a SNR of at 100 m down to 3.0 mJy, and/or at 160 m down to 5.7 mJy. The fraction of SMGs seen at 160 m is higher than that at 100 m. About 50% of radio-identified SMGs are associated with PACS sources. We find a trend between the SCUBA/PACS flux ratio and redshift, suggesting that these flux ratios could be used as a coarse redshift indicator. PACS undetected submm/mm selected sources tend to lie at higher redshifts than the PACS detected ones. A total of 12 sources ( of our SMG sample) remain unidentified and the fact that they are blank fields at Herschel-PACS and VLA 20 cm wavelength may imply higher redshifts for them than for the average SMG population (e.g., ). The Herschel-PACS imaging of these dust-obscured starbursts at high-redshifts suggests that their far-infrared spectral energy distributions have significantly different shapes than template libraries of local infrared galaxies.
Subject headings:Galaxies: high-redshift — Galaxies: individual (GN13 alias HDF850.4) — Galaxies: starburst — Cosmology: observations — Infrared: Galaxies — Submillimeter: galaxies
Several hundred dust-enshrouded high-z sources have been selected through submm/mm imaging with bolometer cameras like SCUBA, LABOCA, AzTEC, MAMBO (e.g., Hughes et al., 1998; Dannerbauer et al., 2002, 2004; Smail et al., 2002; Coppin et al., 2006; Pope et al., 2006; Bertoldi et al., 2007; Perera et al., 2008; Weiß et al., 2009). The large beam size in the (sub)millimeter (e.g., MAMBO: 11″; SCUBA: 15″; LABOCA: 19″) hampers the identification of these so-called Submillimeter Galaxies (SMGs; see for a review Blain et al., 2002) based on bolometer data only. The most obvious choice for obtaining the subarcsecond accurate positions of the dust continuum is (sub)millimeter interferometric continuum observations (e.g., Dannerbauer et al., 2002, 2008; Younger et al., 2007). However the slow mapping speed of (sub)millimeter interferometers does not allow us to study a large number of sources. The most suitable tool for counterpart identification is interferometric observations at radio wavelengths. About of these submm/mm sources have been identified, mainly based on radio observations, and the peak of the redshift distribution of the radio-identified SMG population lies at (Chapman et al., 2005). However, the effect of the bias toward the true redshift distribution introduced by the radio selection technique is still under debate, as SMGs even beyond have been already detected by the VLA at 20 cm (e.g., Daddi et al., 2009a, b; Capak et al., 2008; Schinnerer et al., 2008; Coppin et al., 2009; Knudsen et al., 2010; Morrison et al., 2010). The sample of unidentified SMGs could be either spurious sources or lie at extreme redshifts but no systematic studies on this subsample of SMGs have been conducted so far.
The launch of the Herschel observatory (Pilbratt et al., 2010) promises a
different perspective of SMGs than provided by radio observations only
and offers a higher mapping speed than (sub)mm interferometric
observations. Deep PACS (Poglitsch et al., 2010) imaging at 100 m and
160 m on the Great Observatories Origins Deep Survey North
(GOODS-N) will sample the FIR emission of these dust-enshrouded high-z
objects and enable us to study in detail their far-infrared spectral
energy distribution (SED), redshift distribution, dust temperatures
(Magnelli et al., 2010; Elbaz et al., 2010; Chanial et al., 2010) and dust masses (Santini et al., 2010). In absence
of interferometric observations at mm and cm wavelengths, the
Herschel-PACS beamsize at 100 m (160 m) of 6.7″
(11.0″) provide a more accurate location of the dust emission
than the bolometer data taken with SCUBA, AzTEC, MAMBO or LABOCA.
In this letter we discuss our search for PACS counterparts at 100 m and 160 m, explore the diagnostic potential of Herschel-PACS for the counterpart identification and compare it with the widely used identification approach using VLA observations. In comparison, the Spitzer-MIPS 70 m and 160 m imaging of the GOODS North region performed by Huynh et al. (2007) in the pre-Herschel era detected at relatively high significance only two (one) out of 30 SMGs at 70 m (160 m), at rather low redshifts ( and ). The FIR observations presented here will enable us to study a significant sample of SMGs in the far-infrared wavelength regime.
2. Far-Infrared Association of Submillimeter Galaxies
PACS observations of the GOODS North region at 100 m and 160 m were taken during the Herschel Science Demonstration Phase in autumn 2009 and are part of the Guarantee Time extragalactic PACS survey ’PEP: The PACS Evolutionary Probe’ (PI: D. Lutz). The final images achieve sensitivities of mJy and mJy at 100 m and 160 m respectively. Berta et al. (2010) describe in their appendix the Herschel-PACS observations, data reduction and the blind extraction of PACS sources with SNR. Lutz et al. (in prep.) will present the fluxes of PACS sources. The PACS images are aligned on the GOODS imaging products. Complementary to the Herschel PACS observations, we use for this work data from the VLA at 1.4 GHz (Morrison et al., 2010) and Spitzer-MIPS at 24 m (Dickinson et al., in prep.).
In the past years, several groups surveyed the GOODS North region using the bolometer cameras SCUBA, AzTEC and MAMBO (Hughes et al., 1998; Borys et al., 2003; Pope et al., 2005; Wang et al., 2004; Perera et al., 2008; Greve et al., 2008). These observations discovered about 150 SMGs at 850 m, 1.1 and 1.2 mm. Robustly identified VLA and MIPS 24 m counterparts are already known in the literature for SCUBA and AzTEC sources (Pope et al., 2006; Chapin et al., 2009). In addition, Greve et al. (2008) presented VLA counterparts for 11 out of 30 MAMBO sources.
We search for Herschel-PACS counterparts of SMGs that are either detected with a SNR of or detected by at least two different surveys. 56 SMGs fulfill this criterion which should assure a robust SMG sample to work with. Our sample111In case that an SMG is detected at submm and mm bands, we give preference the SCUBA detection and include it in our sample. For mm-only detected sources we list, if available, the AzTEC otherwise the MAMBO detection. consists of 36 SCUBA, 12 AzTEC and 8 MAMBO sources and for 15 SMGs spectroscopic redshifts (SMG) have been already obtained. We match our SMG sample with the PACS 100 m and 160 m blind catalogue and search for counterparts within a radius of 5.5″ for MAMBO sources, 7.5″ for SCUBA sources and 9″ for AzTEC sources. As a sanity check of the blind catalogue, we inspected by eye the search region in the PACS imaging. The search circles that we have applied correspond to the beam size (FWHM) of the different bolometric datasets and should guarantee that no reliable associations to SMGs will be missed. We also searched the VLA 1.4 GHz map and the MIPS 24 m images for counterparts at the SMG positions.
We calculate the corrected Poissonian probability p that an association of SMGs within the search radius is a chance coincidence. This approach (see for details Downes et al., 1986) corrects the simple Poissonian probability of a detected association for the possibility of associations of different nature but similar probability. The derived probability of PACS and MIPS association is based on raw number counts in GOODS North. We search for VLA counterparts of SMGs down to . However, the VLA source catalogue (Morrison et al., 2010) is only reliable down to 20 Jy (). Therefore, we assess the reliability of VLA counterparts relying on published number counts (e.g., Fomalont et al., 2006). Similar to previous studies, we define following quality criteria for assessing the robustness of identified candidate counterparts. We classify association of SMGs with as secure and with as possible counterparts.
We uncover PACS secure or possible counterparts for 22 SMGs, corresponding to a PACS identification rate of of our whole SMG sample, see also Table 1 for details. Our PACS identification rate of is lower than the rate of found by Magnelli et al. (2010). The reason is that Magnelli et al. (2010) focus on already radio-identified SCUBA and AzTEC sources with spectroscopic redshifts mainly. The top panel of Fig. 1 displays the (sub)millimeter-PACS separation of our association. We did not find evidence of systematic offsets. We find PACS counterparts at a mean (sub)millimeter-PACS positional offset of ″ and ″. This is consistent with a typical submm/mm position error of about 3″ to 5″ and strengthens our choice of the FWHM of the bolometric data as search circle for PACS counterparts. The top panel of Fig. 1 shows clearly a dominance of secure counterparts. Given this statistical approach, associations with offsets much larger than the average might not always be correct.
|PACS m||PACS m|
|# secure association||12||15|
|# possible association||4||5|
|total identification rate||28.6%||35.7%|
|# total association of SMGs||22 (39.3%)|
|# only detected at PACS m||6|
|# only detected at PACS m||2|
|new CPTs identified by PACS||1|
|# PACS and radio blank fields||12 (21.4%)|
We detect 20 (16) SMGs at 160 (100) m and classify 15 (12) of them as secure and 5 (4) as possible counterparts. The number of PACS counterparts at 160 m is higher than at 100 m. This is expected as the 160 m measurements lie close to the FIR peak. We note that based on the corrected Poissonian probability p each PACS detection within the bolometer beam (our search circle) is classified as associated SMG counterpart. Typical PACS fluxes of these dust-obscured high-z sources range between 4.0 mJy to 34.9 mJy at 100 m and 5.0 mJy to 65.0 mJy at 160 m.
We have 43 (28 are secure; 15 are possible) radio-identified SMGs in our sample, about 50% of them are seen at PACS wavelengths. Vice-versa, only one PACS association (secure) is undetected at 1.4 GHz, see detailed discussion of GN13 (alias HDF850.4) at end of Section 3. One-third of PACS detected SMGs have at least two VLA counterparts.
3. Diagnostic Potential of PACS observations of Submillimeter Galaxies
About twice as many SMGs are identified by the VLA than by PACS. VLA observations are more sensitive for sources at redshifts up to . None of the well-known, spectroscopically identified SMGs at — GN20, GN20.2a/b and GN10 (Daddi et al., 2009a, b; Wang et al., 2007, 2009; Dannerbauer et al., 2008) — have been significantly detected in our PACS imaging, see middle panel in Fig. 1. Another highly promising candidate, HDF850.1 (Hughes et al., 1998; Dunlop et al., 2004; Cowie et al., 2009) is also not seen by PACS. However, all these SMGs have radio counterparts. Focusing on our spectroscopic subsample of 15 sources, SMG, we find a trend that the fraction of SMGs detected at PACS bands decreases with redshift (see middle panel in Fig. 1), being explained by the fact that PACS fluxes drop with increasing redshift. No source beyond (GN06) at 100 m and (GN04) at 160 m is significantly detected by PACS. However, we do not find evidence that the PACS detection rate of SMGs correlates with the SCUBA flux density (Fig. 1). In addition, VLA observations provide subarcsecond accurate positions which are essential for the proper identification in the optical and near-infrared bands and follow-up spectroscopic observations. PACS cannot deliver positions at these accuracies. To summarize, ultra-deep VLA observations still remain the best and most effective approach to identify SMG counterparts, even up to redshift .
Using only radio, 850 m and 24 m flux densities, Daddi et al. (2009a) derived accurate () radio-IR photometric redshifts for SMGs. Motivated by this work, we apply the radio-IR photometric redshift code presented in Daddi et al. (2009a) and calculate photometric redshifts by adding our PACS measurements to existing SCUBA, MIPS 24 m and VLA 1.4 GHz flux measurements of our SMG sample. Naively we would have expected to improve the accuracy of the radio-IR photometric redshifts by adding FIR measurements shortwards to the IR dust peak. In Fig. 2 we show the results of our attempt using template libraries of local galaxies (Chary & Elbaz, 2001). In the left panel we show photometric redshifts (adapted from Daddi et al., 2009a) based on radio, MIR and submm observations for SCUBA counterparts with known spectroscopic redshifts in GOODS North which agree fairly well. In the right panel we show photometric redshift estimates adding the PACS measurements to our previous measurements in the radio, MIR and submm. Clearly the obtained accuracy decreases and these photo-z estimates are poorer. Omitting certain measurements (e.g., MIPS 24m and/or VLA 1.4 GHz) gives even worse results. We obtain similar results by using template libraries from Dale & Helou (2002) and the average SED constructed by Michałowski et al. (2010) of 73 spectroscopically identified SMGs (Chapman et al., 2005). This result means that most likely the far-infrared properties of SMGs are different from templates built to describe local galaxies. These observed differences arise most probably from the fact that SMGs are colder than local ULIRGs of same luminosity (Chapman et al., 2005; Magnelli et al., 2010; Elbaz et al., 2010; Chanial et al., 2010). Further investigations are needed to fully exploit the Herschel-PACS imaging in order to obtain accurate photometric redshifts.
We already know the spectroscopic redshift for 15 SMGs in our sample. However, in absence of spectroscopic redshifts for most of our sources, searching for correlations between observed properties is essential for constraining the nature, redshift and evolution of SMGs. We explore if flux ratios involving PACS 100 m and 160 m measurements can be used as a rough redshift indicator. Fig. 3 displays the results of our analysis. The PACS color () of SMGs does not vary with redshift and is consistent with the prediction of models and observed SEDs from Chary & Elbaz (2001), Michałowski et al. (2010), Armus et al. (2007) and Pope et al. (2008). The PACS colors seem to be slightly redder than predicted by the various models and may indicate a difference between the observed FIR-SED of SMGs and templates describing fairly well local infrared galaxies. The composite rest-frame SED for the SMG (Fig. 3) subsample shows their diversity in the far-infrared. Their far-infrared colors are different from the template SEDs, most notably for the local luminous infrared galaxies. This may explain our finding reported above that we are not able to obtain accurate IR-radio photometric redshifts by including PACS measurements.
The PACS fluxes are redshift dependent, whereas the (sub)millimeter flux density at is not (due to the negative K-correction; e.g., Blain et al., 2002), thus one would expect to see a trend between (sub)mm/PACS flux ratio for both PACS bands (see also the predictions based on different templates in Fig. 3). Indeed, we find a fairly clear trend both for respectively versus spectroscopic redshift for the SMG subsample222For SCUBA sources from Pope et al. (2006) we use the deboosted flux. The same we do for AzTEC and MAMBO sources. We converted the AzTEC and MAMBO mm flux densities into SCUBA flux densities by assuming a flux ratio of of 2.5 for a ULIRG-like SED at redshift . over the whole redshift range spanned by our SMGs from . We conclude that the (sub)mm/PACS flux ratio seems to be a useful albeit crude redshift indicator and may help to select/mark SMGs at very high-redshifts. We explored also VLA/PACS and PACS/MIPS 24 m flux ratios versus redshift but did not observe any correlation, being consistent with the fact that VLA, MIPS 24 m and PACS fluxes drop with increasing redshift. Based on this finding, we investigate the (sub)mm/PACS ratio for the remaining 41 SMGs without any reliable redshift. This analysis shows that PACS-detected SMGs tend to have lower flux ratios than the PACS-undetected SMG sample (Fig. 3). This trend seems to be more prominent for the flux density ratio. Our analysis may indicate that indeed as previously discussed PACS-undetected SMGs could tend to lie at higher redshifts and the (sub)mm/PACS flux ratio could be a crucial tool to select the very high-z tail of the SMG population.
The previous analysis bridges with the discovery that 12 out of 56 SMGs () in our sample are blank fields both at PACS and radio wavelengths. The nature of these sources is still not clear. These SMGs could be either spurious sources which is mainly assumed, or at redshifts higher than . Indeed, how reliable are these (sub)mm sources detections? The sample consists of seven submillimeter and five millimeter selected sources (of the latter population, one is seen by AzTEC and four by MAMBO). We checked each source for its reliability. The mm sources are selected with SNR . Four of the seven SCUBA galaxies are listed in both Pope et al. (2006) and Wang et al. (2004). The fact that Pope et al. (2006) and Wang et al. (2004) used (nearly) the same SCUBA data, but performed independent data reduction and source extraction, is giving more weight to the reliability of these detections. As a sanity check we inspected the SCUBA and MAMBO galaxies in the AzTEC 1 mm map. Our analysis suggests that most of these sources are reliable. One of them has a SNR at 850 m. However, we note that only mm interferometric observations will reveal unambiguously the reliability of these sources. Their non-detection both at radio and PACS wavebands in combination with its (sub)mm/PACS flux ratio may suggest that these sources could lie at higher redshifts than the typical SMG population (e.g., ).
We conclude this work by presenting the identification of an SMG counterpart having a PACS but no VLA detection. Daddi et al. (2009a) suspected that the source at , about 7″ away from the SCUBA position, could not be the correct counterpart of GN13 alias HDF850.4 (Pope et al., 2006; Hughes et al., 1998). We detect a faint (), secure counterpart at 100 m, being very close to the bolometer position (). Its reliability is fortified by a 200 Jy strong, secure counterpart at 24 m. Using the flux ratios and as a coarse redshift indicator suggests a redshift of .
- Armus et al. (2007) Armus, L., et al. 2007, ApJ, 656, 148
- Berta et al. (2010) Berta, S., et al. 2010, A&A, in press (astro-ph/1005.1073)
- Bertoldi et al. (2007) Bertoldi, F., et al. 2007, ApJS, 172, 132
- Blain et al. (2002) Blain, A. W., Smail, I., Ivison, R. J., Kneib, J.-P., & Frayer, D. T. 2002, Phys. Rep., 369, 111
- Borys et al. (2003) Borys, C., Chapman, S., Halpern, M., & Scott, D. 2003, MNRAS, 344, 385
- Capak et al. (2008) Capak, P., et al. 2008, ApJ, 681, L53
- Chanial et al. (2010) Chanial, P., et al. 2010, in preparation
- Chapin et al. (2009) Chapin, E. L., et al. 2009, MNRAS, 398, 1793
- Chapman et al. (2005) Chapman, S. C., Blain, A. W., Smail, I., & Ivison, R. J. 2005, ApJ, 622, 772
- Chary & Elbaz (2001) Chary, R., & Elbaz, D. 2001, ApJ, 556, 562
- Coppin et al. (2006) Coppin, K., et al. 2006, MNRAS, 372, 1621
- Coppin et al. (2009) Coppin, K. E. K., et al. 2009, MNRAS, 395, 1905
- Cowie et al. (2009) Cowie, L. L., Barger, A. J., Wang, W.-H., & Williams, J. P. 2009, ApJ, 697, L122
- Daddi et al. (2009a) Daddi, E., et al. 2009a, ApJ, 694, 1517
- Daddi et al. (2009b) Daddi, E., Dannerbauer, H., Krips, M., Walter, F., Dickinson, M., Elbaz, D., & Morrison, G. E. 2009b, ApJ, 695, L176
- Dale & Helou (2002) Dale, D. A., & Helou, G. 2002, ApJ, 576, 159
- Dannerbauer et al. (2002) Dannerbauer, H., Lehnert, M. D., Lutz, D., Tacconi, L., Bertoldi, F., Carilli, C., Genzel, R., & Menten, K. 2002, ApJ, 573, 473
- Dannerbauer et al. (2004) Dannerbauer, H., Lehnert, M. D., Lutz, D., Tacconi, L., Bertoldi, F., Carilli, C., Genzel, R., & Menten, K. M. 2004, ApJ, 606, 664
- Dannerbauer et al. (2008) Dannerbauer, H., Walter, F., & Morrison, G. 2008, ApJ, 673, L127
- Downes et al. (1986) Downes, A. J. B., Peacock, J. A., Savage, A., & Carrie, D. R. 1986, MNRAS, 218, 31
- Dunlop et al. (2004) Dunlop, J. S., et al. 2004, MNRAS, 350, 769
- Elbaz et al. (2010) Elbaz, D., et al. 2010, A&A, in press (astro-ph/1005.2859)
- Fomalont et al. (2006) Fomalont, E. B., Kellermann, K. I., Cowie, L. L., Capak, P., Barger, A. J., Partridge, R. B., Windhorst, R. A., & Richards, E. A. 2006, ApJS, 167, 103
- Greve et al. (2008) Greve, T. R., Pope, A., Scott, D., Ivison, R. J., Borys, C., Conselice, C. J., & Bertoldi, F. 2008, MNRAS, 389, 1489
- Hughes et al. (1998) Hughes, D. H., et al. 1998, Nature, 394, 241
- Huynh et al. (2007) Huynh, M. T., Pope, A., Frayer, D. T., & Scott, D. 2007, ApJ, 659, 305
- Knudsen et al. (2010) Knudsen, K. K., Kneib, J.-P., Richard, J., Petitpas, G., & Egami, E. 2010, ApJ, 709, 210
- Magnelli et al. (2010) Magnelli, B., et al. 2010, A&A, in press (astro-ph/1005.1154)
- Michałowski et al. (2010) Michałowski, M. J., Hjorth, J., & Watson, D. 2010, A&A, in press (astro-ph/0905.4499)
- Morrison et al. (2010) Morrison, G. E., Owen, F. N., Dickinson, M., Ivison, R. J., & Ibar, E., 2010, ApJS, 188, 178
- Perera et al. (2008) Perera, T. A., et al. 2008, MNRAS, 391, 1227
- Pilbratt et al. (2010) Pilbratt, G., et al. 2010, A&A, in press (astro-ph/1005.5331)
- Poglitsch et al. (2010) Poglitsch, A., et al. 2010, A&A, in press (astro-ph/1005.1487)
- Pope et al. (2005) Pope, A., Borys, C., Scott, D., Conselice, C., Dickinson, M., & Mobasher, B. 2005, MNRAS, 358, 149
- Pope et al. (2006) Pope, A., et al. 2006, MNRAS, 370, 1185
- Pope et al. (2008) Pope, A., et al. 2008, ApJ, 675, 1171
- Santini et al. (2010) Santini, P., et al. 2010, A&A, in press (astro-ph/1005.5678)
- Schinnerer et al. (2008) Schinnerer, E., et al. 2008, ApJ, 689, L5
- Smail et al. (1997) Smail, I., Ivison, R. J., & Blain, A. W. 1997, ApJ, 490, L5
- Smail et al. (2002) Smail, I., Ivison, R. J., Blain, A. W., & Kneib, J.-P. 2002, MNRAS, 331, 495
- Wang et al. (2004) Wang, W.-H., Cowie, L. L., & Barger, A. J. 2004, ApJ, 613, 655
- Wang et al. (2007) Wang, W.-H., Cowie, L. L., van Saders, J., Barger, A. J., & Williams, J. P. 2007, ApJ, 670, L89
- Wang et al. (2009) Wang, W.-H., Barger, A. J., & Cowie, L. L. 2009, ApJ, 690, 319
- Weiß et al. (2009) Weiß, A., et al. 2009, ApJ, 707, 1201
- Younger et al. (2007) Younger, J. D., et al. 2007, ApJ, 671, 1531