Identifying the Lens Galaxy B1152+199 as a Ghostly Damped Lyman Alpha System by the Cosmic Origins Spectrograph
aa r X i v : . [ a s t r o - ph . GA ] D ec Identifying the Lens Galaxy B 1152+199 as a Ghostly DampedLyman Alpha System by the Cosmic Origin Spectrograph
Xinyu Dai and Bin Chen ABSTRACT
Strong quasar-galaxy lensing provides a powerful tool to probe the inter-stellarmedium (ISM) of the lens galaxy using radiation from the background quasar.Using the Cosmic Origin Spectrograph (COS) on board the Hubble Space Tele-scope, we study the cold ISM properties of the lens galaxy in B 1152+199 at aredshift of z = 0 . α absorption (DLA) system in the near ultraviolet spec-trum; however, our upper limit on the H i column density is several orders ofmagnitude below the expectation. We also marginally detect O i and C ii ab-sorption lines associated with the lens galaxy in the COS spectrum. Thus, thelens galaxy is identified as a ghostly DLA system, and further investigations ofthese ghostly DLA systems would be important to characterize the biases of us-ing DLAs to probe the matter density of the universe. Although preliminary,the most likely explanation of the non-detection of the DLA is because of theLy α emission of the lens galaxy that fills in the absorption trough, with a Ly α luminosity of 4 × erg s − . Subject headings: gravitational lensing: strong — galaxies: ISM — galaxies:individual (B 1152+199) — ultraviolet: ISM
1. Introduction
The inter-stellar medium (ISM) has a primary role in many areas of astronomy, includingstellar formation/evolution, galaxy formation/evolution, physics of active galactic nuclei, and Homer L. Dodge Department of Physics and Astronomy, University of Oklahoma, Norman, OK, 73019,USA, [email protected] Research Computing Center, Department of Scientific Computing, Florida State University, Tallahassee,Florida 32306, USA E ( B − V ) / ∆ N H of the lens galaxies, under the assumption that the differences betweenthe extinction and gas absorption are due to the same parcel of ISM. Chen et al. (2013)found an evolving dust-to-gas ratio of E ( B − V ) /N H = 1 . +0 . − . × − mag cm atom − inlens galaxies (0 < z < . . × − mag cm atom − (Bohlin, Savage & Drake 1978), and a constant dust-to-metalratio with redshift.This paper focuses on the hydrogen and metal absorption lines in a UV spectrum of thelens galaxy in the two-image lens B 1152+199. B 1152+199 was selected as a lens candidatefrom the Cosmic Lens All-Sky Survey, and was confirmed as a two-image lens system with z s = 1 .
019 and z l = 0 .
439 by subsequent follow-up observations (Myers et al. 1999). The
HST
F555W and F814W band images of B 1152+199 show that the lensing galaxy mayresemble an early-type galaxy (Rusin et al. 2002). However, the large amount of cold ISMdetected from the lens galaxy, through studying the absorption/extinction properties ofthe background quasar spectrum, will classify the galaxy as a late-type. In particular,Myers et al. (1999) detected Mg ii and Mg i metal absorption lines at the lens redshiftfrom the optical spectrum of the quasar image A (C. Fassnacht, private communication).Toft, Hjorth & Burud (2000) detected a large extinction for the system, especially in quasar 3 –image B, and El´ıasd´ottir et al. (2006) confirmed this result and measured a large differentialextinction of ∆ E ( B − V ) = 1 . ± .
05 between images A and B with a slope of R V =2 . ± .
1. Dai & Kochanek (2009) and Chen et al. (2013) detected cold ISM in the lensfrom X-ray spectra of the quasar images and measured differential absorption of ∆ N H =(4 . ± . × cm − assuming Solar metallicity. Combining the differential extinction andabsorption measurements, Dai & Kochanek (2009) measured a dust-to-gas ratio of E ( B − V ) /N H = (2 . ± . × − mag cm atom − , slightly higher than the Galactic value1 . × − mag cm atom − (Bohlin, Savage & Drake 1978).Since metals dominate the X-ray absorption cross section, the N H measured from theX-ray spectrum is an effective column density rather than a direct measurement. Ideally adirect measurement of N H using other methods can be vital to break the degeneracy in theX-ray analysis. The original goal of our Cosmic Origin Spectrograph (COS) observation ofB 1152+199 was to directly measure N H through the damped Ly α absorption (DLA) featureassociated with the lens galaxy and also to detect other associated metal absorption lines inthe brighter image A. Here, we present the non-detection of the DLA in B 1152+199. Theobservation and data products are introduced in Section 2, followed by the imaging analysisin Section 3 and the spectral analysis in Section 4. We discuss the results in Section 5. Weassume cosmological parameters of Ω M = 0 .
27, Ω Λ = 0 .
73, and H = 70 km s − Mpc − throughout the paper.
2. Observation
We obtained a low-resolution NUV spectrum of B 1152+199 using COS (Green et al.2012) onboard
HST with the G230L grating on 2014–04–15. We used two central wave-lengths to cover a broader wavelength range and four positions for each central wavelengthto minimize any local systematics in the detector, such as the bad pixels. The detailed obser-vation log is listed in Table 1. The data products were retrieved from STScI after updatingthe wavelength calibration issue in the G230L grating mode in May 2016, and we used thestandard pipeline-produced, background-subtracted spectrum in our subsequent analysis.
3. Imaging Analysis
We analyzed the acquisition image of B 1152+199, taken with the COS/NUV/MirrorBconfiguration with an exposure of 95.6 sec, and only detected image A. Although two pointsources were detected, they are the double-image caused by MirrorB. We first modeled the 4 –Table 1.
HST -COS Observations of B 1152+199.
Target Instrument Grating Wavelength Spectral ExposureCoverage (˚A) Resolution a Time (sec)B 1152+199 COS/NUV/MirrorB None 1200–3300 · · · a The spectral resolution increases with the wavelength.
Table 2. Emission Lines Identified in the COS UV Spectrum.
Line Redshift Rest-Frame FWHM Fluxkm s − − erg cm − s − C iii ± ± β · · · a · · · b ± vi · · · ±
380 2.15 ± α · · · ± ± α · · · ±
846 5.2 ± v · · · ± ± iv · · · ±
661 0.3 ± iv · · · ± ± iv · · · ±
480 3.4 ± a The redshifts of all emission lines are linked. b The line width of Ly β is fixed to be the same as that of thebroad Ly α line. Table 3. Absorption Line Candidates in the COS UV Spectrum.
Line Wavelength Identification redshift Wavelength FWHM Velocity Dispersion Optical Column Density˚A ˚A km s − (Rest) km s − Depth 10 cm − Abs1 1872.3 ± i · · · a · · · a · · · a · · · +0 . − . b Abs2 1917.7 ± ii ± +450 − +260 − +0 . − . · · · · · · Ly α · · · a · · · a · · · a · · · < . < . ± α (or C iv ) 0.8317 (or 0.4377) 2226.8 ± +390 − <
100 0.98 +8 − . ± α ± +1270 − +400 − +0 . − . a The redshifts, wavelengths, and widths of O i and Ly α lines at z = 0 . ii line. b The spectral bin size of 1.5˚A is modeled in calculating the column densities.
Sherpa using two Gaussian components (A1 and A2) for the double-image of image A and a constant background, and obtained an acceptable fit. We thenadded two additional Gaussian components (B1 and B2) for the double-image of image B tothe model. The relative position between B1 and A1 is set by previous
HST measurements(Rusin et al. 2002), and the FWHMs, relative position, and flux ratio between B2 and B1are set to be the same as those for A2 and A1. Essentially, we added one free parameterto the model, the normalization of B1. We constrained the 3 σ upper limit of the flux ratiobetween image B and A as f B /f A < E ( B − V ) = 1 . ± .
05 between images A and B with R V = 2 . ± .
4. Spectral Analysis
Figure 1 shows the background subtracted NUV spectrum (binned by 5˚A) of B 1152+199,a typical broad line quasar spectrum. Since image B is expected to be 9–10 mag fainter thanimage A in the NUV band, the spectrum is essentially that of image A. The actual spectralfitting was performed on the observed spectrum binned by 1.5˚A (Figure 4), slightly largerthan the resolution limit of 1˚A of the observation. We also used
Sherpa to model the spec-trum, and we first empirically fit the spectrum using a polynomial continuum model plusnine Gaussian emission lines. We found that a six-component polynomial is sufficient tomodel the continuum, and the nine detected quasar emission lines are the broad C iii , Ly β ,O vi , Ly α , N v , Si iv , and C iv lines along with narrower components for the Ly α andC iv lines. We set the width of the broad Ly α and Ly β lines to be the same in all thefits. We linked the wavelengths of these lines and jointly fit the redshift of the quasar as z s = 1 . ± . − . z s = 1 . ± . α ab-sorption associated with the lens galaxy (Figure 2). Based on the extinction and X-rayabsorption measurements of B 1152+199, we expect a neutral hydrogen column density of N H ∼ cm − for image A from the lens galaxy, which would result in a damped Ly α line. However, no absorption line is detected at the expected wavelengths 1745–1750˚A as-suming that the lens redshift is between z l =0.436–0.439. We estimated an upper limit on 6 –the neutral hydrogen column density by modeling the 1720–1770˚A spectral segment by apower-law modified by a Voigt absorption profile. We assumed an intrinsic velocity disper-sion of 200 km s − at the rest-frame of the absorber, a typical value for normal galaxies, forthe Gaussian component, and the absorbed model was further smoothed by the 1.5˚A binsize of the observed spectrum. The 68%, 90%, and 99% confidence limits on the neutralhydrogen column density are N H < .
65, 1.6, and 20 × cm − , respectively.After searching for other absorption lines in the spectrum, we identified four candidates(Abs1–4) at wavelengths 1872.3˚A, 1917.7˚A, 2226.8˚A, and 2265.8˚A, respectively. Possibleorigins of these lines include metal absorption lines associated with the lens galaxy, ourown Galaxy, quasar host galaxy, or other Ly α absorbers along the line of sight. We firstexcluded Galactic absorption lines because the wavelengths do not match any strong UVISM absorption lines (Blades et al. 1988). We next considered absorption lines in the quasarhost galaxy, in particular a series of Fe ii absorption lines with rest-wavelengths shorter thanthat of Ly α will fall in the observed wavelength range; however, they also do not match thewavelengths of the absorption line candidates. Thus, we concluded that Abs1–4 are due tosome intervening systems, either the lens galaxy or other intervening clouds. Consideringthe redshift of the lens ( z l ≃ . i λλ ii λλ iv λλ T < K , we excluded C iv from thelens as the interpretation of Abs3 and think it is possibly a Ly α line from a different redshiftat z = 0 . α absorber at z = 0 . i , and C ii absorptionlines and the expected DLA from the lensing galaxy together in Figure 3 to further illustratethe non-detection of the expected DLA feature.We performed a final fit to the spectrum by linking the wavelengths and FWHMsof Abs1 and Abs2, since they are both from the lens galaxy, and reported the best-fitparameters in Table 3. The spectrum, model, and residuals of this fit are shown in Figure 4.We measured the redshift of the lens as z l = 0 . ± . − . i and C ii lines, wecalculated a number of confidence intervals and report the one such that the correspondingoptical depth is consistent with zero. The O i line is detected by 2 . σ (99.77% one-sidedprobability) and the C ii line is detected by 1 . σ (88.69% one-sided probability). Consideredjointly, the presence of NUV metal absorption lines from the lens galaxy is significant at the99.975% confidence (3.5 σ ) level. We also calculated the rest-frame velocity dispersions ofthe absorption lines by subtracting the spectral resolution of ≃ +260 − km s − , consistent with values of typical L ∗ galaxies.
5. Discussion
We have analyzed a COS NUV spectrum of image A of the gravitational lens B 1152+199.We expected to find, based on the absorption seen in the X-ray spectrum, a DLA feature atthe lens redshift with N H ∼ cm − for image A (Dai & Kochanek 2009; Chen et al.2013). We can place a 99% upper limit on the neutral hydrogen column density with N H < × cm − . We do detect weak O i and C ii absorption lines associated withthe lens galaxy at a combined significance of 3.5 σ . B 1152+199 has been well studied in theoptical and X-ray bands, and there are multiple indicators showing that B 1152+199 containsa large amount of ISM, the optical extinction, X-ray absorption, and the presence of Mg ii and Mg i absorption lines. These previous measurements are consistent with our detectionof O i and C ii absorption lines. The non-detection of the DLA feature is surprising, andthus, the lens galaxy of B 1152+199 is identified as a “ghostly” DLA, a DLA revealed byother absorption features but undetected (Fathivavsari et al. 2016). Here, we discuss severalpossible explanations.First, it is unlikely that the lens has an extremely high metal-to-gas ratio, such that weonly detect absorption signatures from the metals or dusts including the extinction, X-rayabsorption, and metal absorption lines. Our 99% N H upper limit is four orders of magnitudebelow the expectation, and such an extreme metal-to-gas ratio is unprecedented. Physically,collisions will always bind and mix the atom/molecules of difference species, and creating aregion devoid of hydrogen atoms but with only metals is difficult. Second, it is also unlikelythat most hydrogen atoms are in the ionized state because of the signatures of cold ISMdetected, the O i , C ii , Mg i , and Mg ii lines. In particular, the ionization potential for Mg i is 7.64 eV lower than that for H i (13.6 eV) and the ionization potential for O i (13.62 eV)is comparable to that of H i , while the O and Mg abundances are much lower. The O i column density is measured to be ∼ × cm − for image A, and for Solar metallicity, theexpected H i column density is ∼ cm − , three orders of magnitude above our 99% limit.Thus, it is difficult to attribute the weak metal absorbers to a satellite galaxy associatedwith the lens. Image B has N H = 4 . × cm − from the X-ray spectrum, and this largeamount of ISM is consistent with a late-type L ∗ galaxy and is difficult to explain it in asatellite as well.A more plausible explanation is that the expected DLA absorption trough is filled bythe Ly α emission from the lens galaxy. Fathivavsari et al. (2016) recently reported another 8 –ghostly DLA system at z = 1 . z = 1 . α emission fromthe quasar broad line region. The authors also assumed that the ghostly DLA is fallingtowards the quasars, which yields an additional redshift such that its measured redshift islarger than that of the quasar. In B 1152+199, the lens galaxy is well separated from thebackground quasar, and the trough is presumably filled by the Ly α emission from the lensgalaxy, and thus the ghostly DLA in B 1152+199 represents a different population. Ly α emission from DLA systems has been detected (e.g., Joshi et al. 2016), usually in the redwing of the DLA absorption profile. The selection method for typical DLAs, however, willmiss the B 1152+199-like ghostly DLAs. The Ly α emission usually has a double-humpedemission profile to escape from the galaxy (e.g., Hansen & Oh 2006; Dijkstra 2014), and inB 1152+199 it is possible that the DLA trough is filled by the wings of the double-humpedemission profile. Depending on the detailed scattering conditions, the two peaks can beseparated by up to a few or more Gaussian widths (e.g., Figure 5 of Dijkstra 2014), and thetotal profile can be broad and shallow. The estimated Ly α luminosity to fill the absorptiontrough is 4 × erg s − , which is close to the break luminosity of the Ly α emitter luminosityfunction (e.g., Dressler et al. 2015). Higher signal-to-noise ratio and resolution spectrum ofB 1152+199 is needed to investigate the details.Regardless of the explanations, the expected DLA is not detected in B 1152+199, wherea large amount of H i is expected. DLAs are important probes to study the baryon distri-bution in the high redshift universe (e.g., Wolfe et al. 2005). Therefore, it is important tocharacterize the statistical properties of the B 1152+199-like ghostly DLAs to evaluate thepotential biases introduce by these systems.We acknowledge C. S. Kochanek for helpful comments and C. Fassnacht for provid-ing more details on the Keck II LRIS spectrum of B 1152+199. Support for the program,HST-GO-13283, was provided by NASA through a grant from the Space Telescope ScienceInstitute, which is operated by the Association of Universities for Research in Astronomy,Inc., under NASA contract NAS 5-26555. REFERENCES
Bohlin, R. C., Savage, B. D., & Drake, J. F. 1978, ApJ, 224, 132Blades, J. C., Wheatley, J. M., Panagia, N., et al. 1988, ApJ, 334, 308 9 –Fig. 1.—
HST -COS-G230L NUV spectrum of B 1152+199, binned by 5˚A. C iii , Ly β , O vi ,Ly α , N v , Si iv , and C iv emission lines from the background quasar at z s = 1 . α absorption feature from the lensgalaxy z l = 0 . i and C ii absorption lines associated with the lens galaxy, and the other twoare other Ly α absorbers along the line of sight at different redshifts. 10 –Fig. 2.— The NUV spectrum (squares) between 1720–1770˚A of B 1152+199. A DLA systemwas expected at 1748.8˚A (observed frame), which is not detected in the spectrum. The solidcurves show a power law model absorbed by Voigt profiles with N H = 0, 10 , 10 , 10 ,10 , 10 cm − , from the top to the bottom, further smoothed by the 1.5˚A resolution ofthe spectrum. We measured a 1 σ upper limit for the neutral hydrogen column density of N H < . × cm − . 11 –Fig. 3.— Zoomed in NUV spectrum of B 1152+199 close to the absorbers associated withthe lens galaxy. The top panel shows the non-detection of the expected Ly α absorption. Themiddle and bottom panels show the O i and C ii lines from the lens galaxy. The verticallines show the expected wavelengths of these lines for a lens redshift of z l = 0 . F l u x D en s i t y ( E r g s − c m − A − ) −2e−1502e−154e−156e−15 Wavelength (Angstrom) S i g m a −4−202 Fig. 4.—