Figura atribución del contenido al continente

In this chapter we present a sample of X-ray selected AGN that have been observed in two ALMA Band-7 programs during Cycle 1 and Cycle 2. The Cycle 1 sample was selected to have L2−8keV> 1042 erg s−1 at redshifts of 1.5< z <3.2 (see Mullaney et al.

2015; Harrison et al. 2016). The Cycle 2 sample was selected to uniformly sample the L2−8keV–z plane covering the redshifts of 1.5< z <3.2 and 1043<L2−8keV. 1045erg s−1.

The sample was restricted within the areas covered by the Herschel observational pro- grams PEP/GOODS-H (Lutz et al. 2011; Elbaz et al. 2011) and HerMES (Oliver et al. 2012) in the fields of GOODS-S, and COSMOS, that are our main sources of the FIR photometry covering the wavelengths of 70 – 500µm. In both ALMA programs the tar- geted sources were chosen to have insufficient Herschel photometry (i.e., undetected in most bands and with high upper limits) to successfully constrain the IR SED and decom- pose it to the star-forming and AGN components. Consequently, our sample consists of mostly Herschel, and sometimes Spitzer, undetected sources with poor SFR constraints. The selected sample of X-ray AGN is a sub-sample of that presented in Chapter 3. We make use of the photometric counterparts assigned to the X-ray AGN in that work for our analysis, in combination to the ALMA observations at 870µm.

Since the original definition of the samples for the ALMA programs, there have been new redshift catalogues of the CDF-S and C-COSMOS from Hsu et al. (2014) and March- esi et al. (2016) respectively. We make use of the updated redshifts from the updated cat- alogues, and a number of sources no longer lie within the parameter space of the original selection, with 89% lying within, 3.7% lying below, and 7.3% lying above the original redshift selection.

In this chapter we analyse all of the X-ray AGN that were observed, including serendip- itous detections within the primary beam. We restrain the sample to only sources with

5.2. Sample & Observations 135

Figure 5.2.1: X-ray hard band (HB; 2–8keV) luminosity (L2−8keV) as a function of red-

shift. Here we plot the 109 X-ray AGN observed with ALMA, including 101 originally targeted and 8 serendipitous detections. We highlight the ALMA detected sources with black centers.

z> 1, resulting in 109 X-ray AGN with ALMA observations, 101 originally targeted, and 8 serendipitously detected X-ray AGN. In Fig. 5.2.1 we plot the L2−8keV as a function

of redshift for the sample studied here, with updated redshift values, and highlight the ALMA Band-7 detected sources. In total there are 5 sources with z < 1 covered by the ALMA program, all in the field of GOODS-S, that are not included in the analysis of this chapter.

5.2.1

ALMA 870um observations

The sample of 109 X-ray AGN were observed during Cycle 1 and Cycle 2 with a band- width of 7.5GHz centered at 351GHz, with 55 sources in CDF-S and 54 sources in C- COSMOS.

The data were processed and imaged following the methods of Simpson et al. (2015); also see Harrison et al. (2016). We used the COMMON ASTRONOMY SOFTWARE APLICATION (CASA; version 4.4.0; McMullin et al 2007), and the CLEAN routine provided within CASA. The raw data was calibrated using the ALMA data reduction pipeline. The results were visually inspected, and when deemed necessary, the pipeline

calibration process was repeated with additional data flagging. We cleaned the images by first creating “dirty” images. We then identify the sources with SNR≥5, which we mask, and then repeat the cleaning process down to 1.5σ. We measure the noise in the resulting cleaned images, and repeat the above process around the sources with SNR≥4. Finally, for the final cleaned images, we applied natural weighting and a Gaussian taper. The synthesized beams are of the size of (0.8” − 0.9”) × 0.7”, with noise levels of 0.1–0.8 mJy/beam in CDF-S1, and 0.08–0.23 mJy/beam in C-COSMOS.

From the product images, Scholtz et al. (in prep) created catalogues of targeted sources and serendipitous detections. From the maps we extracted all the peaks of at least 2.5σ and matched to the optical counterparts from Hsu et al. (2014) and Marchesi et al. (2016) catalogues. To estimate the probability of spurious peaks being matched to our sources, we created negative maps (negatives of the original maps), from which we estimate the density of noise peaks at different SNR values. The surface density of noise peaks matched to the X-ray objects for SNR bins of 2.5 – 3, 3 – 4, and >4 are 0.01, 0.027, and 6×10−4 objects per arcsec2, corresponding to 2.41, 0.89 and 0.052 overall spurious matches respectively, for the search radius of 0.5”. For peaks of SNR > 4 we increased the matching radius to 1” (corresponding to 0.24 expected spurious objects). Finally, the catalogue includes all targeted sources including spurious detections. We use an SNR>2.5 limit to define detected sources; however we note that the majority of our detected sources have SNR>3 (∼73%; Scholtz et al. in prep). If a source remains unde- tected we chose to take a more conservative limit of 3×RMS as the flux limit. In total we find that 40/109 (36.7%) of our sources are detected by ALMA.

5.2.2

MIR and FIR photometry

For our SED fitting analysis, we exploit available photometry in the wavelength range of 3.6 – 500µm, provided by observations carried out by: Spitzer-IRAC at 3.6–8µm; Spitzer- IRS at 16µm; Spitzer-MIPS at 24µm; Herschel-PACS at 70, 100, 160µm; and Herschel- SPIRE at 250, 350, 500µm, in addition to the ALMA photometry outlined above.

1_{The large noise levels of 0.8mJy/beam correspond to a sub-sample of targets that were observed at}

higher resolution than that requested (i.e, 0.3” instead of 1” resolution). Therefore, for these observations the images had to be heavily tapered to a resolution of ∼0.8”, resulting in increased noise levels.

5.3. Method: IR SED fitting 137