Online material

Appendix A: Absence of gravitational lensing

Whether the submm/radio companion is a distinct object or a gravitational image of the quasar host galaxy is an important question because in the absence of magnification, BR1202–0725 would be the most luminous binary CO and far-IR source in the Universe. The observational facts do indeed speak against gravitational lensing. The difference in linewidths of the NW and SE sources, the absence of optical or UV emission from the NW source, and the multiple CO and dust components in each source, are all difficult to reconcile with a gravitational lensing scenario to account for the appearance of BR1202–0725. In the following discussion of gravitational lensing, we definitely rule out the hypothesis that the SE and NW sources are multiple lensed images of the same background source.

A.1. Different linewidths

The main piece of evidence comes from the greatly different molecular and atomic emission line profiles observed toward the NW and SE sources. These NW and SE CO and carbon-line profiles are so different in their widths and peak velocities, that they cannot be gravitational images of a single object. Differences in velocities and linewidths might occur in multiple lensed images of optical lines, if, for example, a multiply-imaged quasar were locally magnified by a micro-lensing event that would affect only one of the images. Such an effect cannot affect the submm CO and carbon lines, however, because these low brightness, high luminosity lines are emitted over a large (≥2-kpc), relatively cool (50 to 100 K) region, and unlike the small, hot, accretion disk of a quasar, cannot be affected by micro-lensing. Differential image magnification between the NW and SE CO sources can also be ruled out; although the SE and NW CO sources have quite different linewidths, their dust continuum fluxes are nearly the same. The dust sources and the molecular line sources are observed at the same wavelengths, and their measured sizes (~0.5″) are roughly the same, so there is no evidence for any differential magnification between the SE and NW sources at submm wavelengths.

A.2. No optical counterpart of the NW galaxy

The second piece of evidence comes from the dissimilarity of the optical and mm/submm images of the quasar and its main companion, the NW source. The SE submm dust and molecule component is associated with an optically-bright quasar whose light is not seen at the position of the NW submm component. The nearly unsurmountable problem is to make a gravitational lensing model that would produce a single-spot optical image and a double-spot submm image. To first order, gravitational lensing is achromatic if the source size is the same at different wavelengths.

But the quasar optical light (rest-frame UV) and its broad lines come from the black hole accretion disk on a scale of 10^-2 pc, while the low-brightness-temperature CO lines and submm dust emission must be of kpc-scale size to be detectable at all at high redshifts. The molecular disk area is thus 10¹⁰ times larger than the quasar accretion disk area, and this can lead to dramatic differences in lensing amplification and lensed image shapes between optical and submm wavelengths.

Two examples are IRAS F10214+4724, and APM 08279+5255. In F10214, the quasar is too obscured to be visible, but the AGN optical narrow-line region is magnified by a factor of 100, while the much larger submm molecular disk is magnified by only a factor of 10 (see models by Broadhurst & Lehár 1995; Trentham 1995; Eisenhardt et al. 1996). In APM 08279, the quasar’s rest-frame UV through mid-IR is also magnified by factor of 100, while the cold part of the submm molecular disk is magnified by much less (see model by Egami et al. 2000, their Fig. 9). But even in these models with large differences in magnification between the optical and submm, it is nearly impossible to simultaneously produce a double image at submm wavelengths, and only a single image at optical wavelengths.

In a lensing scenario, one would have to imagine a galaxy on the line of sight somehow producing a double image only of the molecular disk, but not of the optical quasar, and then, in addition to this first lens that images only the molecular disk, a second lens producing a strong micro-lensing of the quasar to make an optically bright, single, quasar image, with no micro-lensing amplification of the much larger molecular disk. This overly-complicated scenario would not work at near-IR wavelengths of 3.6 and 4.6 μm (rest wavelengths 6300 and 8070 Å), however, because the emitting region, the hot outer parts of the circumnuclear ring around the quasar, are large compared to a typical micro-lensing Einstein radius (e.g., Sluse et al. 2011).

Contradicting this scheme, the Spitzer archive images at 3.6 and 4.6 μm show that the quasar is indeed much brighter (>100 times) than any object toward the NW, in the near-IR. In fact, the light that is seen in the NW in Spitzer data is actually the extended starlight continuum from the Lyα 1 companion galaxy, which is also near-IR bright in the R, I, and K bands (see discussion of their Spitzer data by Hines et al. 2006). The submm dust and molecular disks in the SE and NW galaxies that are discussed in the present paper cannot be detected in the Spitzer near-IR images, because they are too cold; their flux is hopelessly far down the Wien side of the blackbody curve, and they do not radiate at all at rest-frame optical wavelengths (Fig. 8).

All these difficulties to find a lensing model that would produce a single-spot optical quasar image and a double-spot submm image, such as a galaxy-mass lens for a double-image molecular disk only, simultaneously with stellar-mass micro-lensing for the quasar only, that doesn’t really work for the near-IR images, lead us to conclude that a lensing explanation is a blind alley.

A.3. Absence of a candidate deflector to make a double image

Deep optical images from the Hubble Space Telescope (HST) and VLT do not show any evidence of a massive foreground object capable of deflecting the light of the quasar of BR1202–0725 into two widely-separated images. Nor do they show any evidence of any arc-like structure, neither on galaxy-lensing angular scales, nor on cluster-lensing angular scales.

Because the SE and NW sources have similar fluxes in the submm continuum and in the CO, [CI], and [CII] emission lines, the most natural hypothesis for gravitational lensing would be a fairly circularly symmetric mass distribution located close to the mid-point of the 4″ line connecting the two sources. Deep ESO-NTT imaging (Fontana et al. 2000) of the system along with BVrIK data (Giallongo et al. 1998) allow us to put tight upper limits on the deflector brightness.

To produce an image splitting of Δθ = 4″, a deflector at redshift z_L, that for simplicity we approximate as a Singular Isothermal Sphere, would need to have a velocity dispersion given by: (A.1)where w = d_LS/d_S is the ratio of the angular diameter distances between the deflector and the source and between the observer and the source. We get a sensible value of σ ≲  350 km s^-1 only if z_lens ≲ 1. Furthermore, from the Faber-Jackson relation (e.g., Faber et al. 1987; Djorgovski & Davis 1987), one can estimate the rest-frame luminosity of such a deflector and, for a SED typical of a massive galaxy, one can predict the redshifted r-band magnitude. For any redshift z_lens ≲ 1, the deflector would have a magnitude ≤ 24 and be bright enough to be detectable in the HST image. We thus conclude that, unless the deflector is very atypical relative to more local lensing galaxies (e.g., Auger et al. 2010), one would be able to detect it.

The brightest near-IR object in the vicinity of BR1202–0725 is the faint Lyα 1 companion galaxy located between the NW and SE sources, which has an r magnitude of 24.3. It cannot play the role of a deflector, as it has the same redshift as BR1202–0725 and is much too faint. The second, even fainter, Lyα 2 companion galaxy lies 3″ southwest of BR1202–0725, but it is again at the same redshift as the quasar (Hu et al. 1996, 1997).

Finally, there is no arc-like feature on deep optical images that would indicate very strong magnification of the bright quasar. Such an arc-like feature might arise if the quasar (only) were on the infinite-magnification caustic of a lens so close to the line of sight to the quasar itself, that the lensing galaxy would be

visible neither against the quasar light, nor even on images where the quasar light is subtracted out, such as the quasar-subtracted images of Fontana et al. (1996) and Hu et al. (1996).

In summary, there is no evidence, at any wavelength, for a galaxy-mass lens along the line of sight to BR1202-0725, that could produce a submm double image.

A.4. Absence of a nearby cluster lens for overall magnification

Apart from trying to explain the submm double source by lensing by a galaxy-mass deflector, one might ask whether the entire BR1202-0725 complex could be lensed by foreground cluster of galaxies. In this case, the NW and SE submm sources are indeed independent galaxies, but each of them, and the optical quasar, might be amplified by a comparable amount, by a large, low-redshift foreground lens, i.e., a cluster of galaxies. Here, the answer is no. The BR1202-0725 region has been the object of a deep-field survey, the NTT BR1202 field, together with an overlapping region, the NTT Deep Field, resulting in a total field of 2.3′ × 4′ (Fontana et al. 2000). These two overlapping deep fields were surveyed in UBVr bands and in the near-IR IJHK bands. Magnitude limits were of order 26 in r and 21.7 in K. Although there is a peak in the redshift distribution at z ~ 0.6 (Giallongo et al. 1998), there is no evidence for a cluster of galaxies that could act as an overall magnifier for the BR1202-0725 complex as a whole. Nor is there any systematic lengthening of galaxy images into ellipses or arcs that would be signatures of large-scale magnifcation across the field.

© ESO, 2012