MUSE Illuminates Channels for Lyman Continuum Escape in the Halo of SBS 0335-52E
E.C. Herenz, M. Hayes, P. Papaderos, J.M. Cannon, A. Bik, J. Melinder, G. Östlin
AAstronomy & Astrophysics manuscript no. sbs0335_letter_FINAL c (cid:13)
ESO 2018November 6, 2018 L etter to the E ditor VLT/MUSE Illuminates Possible Channels for Lyman ContinuumEscape in the Halo of SBS 0335-52E (cid:63)
E. C. Herenz , M. Hayes , P. Papaderos , J. M. Cannon , A. Bik , J. Melinder , and G. Östlin Department of Astronomy, Stockholm University, AlbaNova University Centre, SE-106 91, Stockholm, Sweden Instituto de Astrofísica e Ciências do Espaço - Centro de Astrofísica da Universidade do Porto, Rua das Estrelas, 4150-762 Porto,Portugal Department of Physics & Astronomy, Macalester College, 1600 Grand Avenue, Saint Paul, MN 55105, USAReceived ...; accepted ...
ABSTRACT
We report on the discovery of ionised gas filaments in the circum-galactic halo of the extremely metal-poor compact starburstSBS 0335-052E in a 1.5 h integration with the MUSE integral-field spectrograph. We detect these features in H α and [O iii ] emis-sion down to a limiting surface-brightness of 5 × − erg s − cm − arcsec − . The filaments have projected diameters of 2.1 kpc andextend more than 9 kpc to the north and north-west from the main stellar body. We also detect extended nebular He ii λ z star-forming galaxies, the ionised gas in this galaxy is dispersion dominated. We argue that the filaments were created via feedbackfrom the starburst and that these ionised structures in the halo may act as escape channels for Lyman continuum radiation in thisgas-rich system. Key words.
Galaxies: starburst – Galaxies: haloes – Galaxies: individual: SBS 0335-052E – Techniques: imaging spectroscopy
1. Introduction
Pinpointing the sources that are relevant for reionising the Uni-verse at z (cid:38) E γ ≥ z galaxies are scarce (e.g.Rutkowski et al. 2017). Moreover, the processes enabling theescape of this so-called Lyman continuum (LyC) radiation fromthe interstellar- and circum-galactic medium (CGM) are not wellconstrained. In principle feedback from supernovae can carveionised gas cavities through the CGM, thereby creating channelsthrough which the LyC can leak into the intergalactic medium(Fujita et al. 2003). Observational inferences of such processeshave to be made from the neutral and the ionisied gas phase inthe CGM (Bik et al. 2015). At high redshifts, however, direct ob-servations of the CGM in its neutral phase are currently impossi-ble (Obreschkow et al. 2011) while directly detecting its ionisedphase remains challenging (Rauch et al. 2016; Wisotzki et al.2016; Finley et al. 2017). Fortunately, observations of nearbygalaxies with properties comparable to systems at the highestredshifts provide a promising alternative for detailed examina-tions of the relevant physical processes in the early universe (fora recent review see Hayes 2015).The SBS 0335-052 system at a redshift of z = . d =
54 Mpc) consists of a pair of extremely metal-deficient star-forming dwarf galaxies separated by 22 kpc (in projection) that (cid:63)
Based on observations collected at the European Organisation forAstronomical Research in the Southern Hemisphere under ESO pro-gramme 096.B-0690. has long been recognised as a special laboratory for such studies(Izotov et al. 1990). In fact, with 12 + log(O / H) (cid:46) .
15 (3% ofthe solar value) the western galaxy – SBS 0335-052W (discov-ered by Pustil’nik et al. 1997) – is one of the most metal-pooremission line galaxy known (Izotov et al. 2009). Observationsof galaxies with such low oxygen abundance are rare and onlya handful of objects comparable to SBS 0335-052W are known(Guseva et al. 2017; Izotov et al. 2017). The brighter easterngalaxy – SBS 0335-052E – shows only a slightly higher nebularoxygen abundance, and thus also belongs to the elusive group ofextremely metal-poor compact star-forming galaxies. HST ob-servations of this galaxy reveal six young ( t < −
25 Myr)super-star clusters (SSCs, with masses M SSC > M (cid:12) and star-formation rates SFR SSC (cid:46) (cid:12) yr − ) that are within ∼
500 pc ofeach other (Thuan et al. 1997; Adamo et al. 2010). Observationsof the 21cm line revealed that the two galaxies are surrounded bya large HI complex (Pustilnik et al. 2001). This HI halo exhibitselongated tidal tails to the east and to the west as well as a faintdi ff use bridge connecting both galaxies (Ekta et al. 2009). Theinferred neutral hydrogen column density towards brightest star-forming regions in SBS 0335-052E is 2 × cm − . Analysis ofHST UV spectroscopy with GHRS (Thuan & Izotov 1997) andCOS (James et al. 2014) points at an even higher neutral columnof 5 − × cm − . Thus, the optical depth for photons above1 Ry is 10 or higher, implying total absorption of LyC radiationfrom the young stellar population in SBS 0335-052E along oursightline.In this letter we report on a discovery of low surface-brightness (SB) filamentary tails in H α and [O iii ] emission thatemanate to the north (N) and north-west (NW) of SBS 0335- Article number, page 1 of 5 a r X i v : . [ a s t r o - ph . GA ] O c t & A proofs: manuscript no. sbs0335_letter_FINAL
2. Observations, data reduction, and creation ofnarrow-band images
VLT / MUSE (Bacon et al. 2010) observations of SBS 0335-052Ewere obtained in service mode under clear skies on November16th and 17th, 2015. The seeing varied between 0.9 (cid:48)(cid:48) and 1.2 (cid:48)(cid:48) .We used the wide-field mode without the second-order block-ing filter. The total exposure time was 5680 s (8 ×
710 s integra-tions). No separate sky exposures were taken. We reduced andcalibrated the data using the MUSE data reduction system (DRSWeilbacher et al. 2014) version 1.6.2. After reduction we wereequipped with a sky-subtracted, wavelength, and flux calibrateddatacube covering a field of view (FoV) of 1 (cid:48) × (cid:48) over a spec-tral range from 4600Å–9370Å. From this cube we producedan “emission-line” only datacube by subtracting an in spectraldirection median-filtered (180Å filter width) “continuum-only”cube (see Sect. 4.1 in Herenz & Wisotzki 2017). We then synthe-sised 8Å wide narrow-band images centred around the emissionlines H α , [O iii ] λ ii λ E γ ≥ E γ ≥ . E γ ≥ α and[O iii ] (after applying the correction described below) and Fig-ure 2 for He ii .Due to the large spatial extent of the H α and [O iii ] emis-sion the standard sky correction resulted in over subtraction ofH α and [O iii ] flux. This resulted in negative zero-point o ff setsin the H α and [O iii ] narrow-band images and, moreover, vis-ible H α and [O iii ] emission line features in the DRS calcu-lated sky spectrum. After ensuring that no real skylines werepresent at the a ff ected wavelengths, we removed those featuresfrom the calculated sky spectrum by linear interpolation over thebracketing wavelengths. Feeding this modified spectrum backinto the DRS removed the over subtraction for H α completely,while in the [O iii ] band a remaining small average o ff set of2 × − erg s − cm − had to be corrected manually.
3. Results
In the high SB region of our H α and [O iii ] images (Fig. 1) weresolve in great detail the super-shell structure( r ≈ . α as faint as 5 × − erg s − cm − arcsec − . Moreover, the spatiallyresolved nature of our MUSE data allows to map the filamen-tary morphology of the extended low-SB emission for the firsttime. The filaments appear to continue beyond the MUSE FoV,permitting us only to provide lower limits of their radial extend:9.7 kpc (35.1 (cid:48)(cid:48) ) and 9.3 kpc (33.7 (cid:48)(cid:48) ) for the NW and N filament,respectively.The filamentary “ears” di ff er in their [O iii ] / H α flux ratios.The NW ear is only barely detected in [O iii ], while it bright-ens in H α towards the edges of the FoV, so that [O iii ] / H α (cid:46) /
10. The northern ear, however, is clearly detected in [O iii ],with [O iii ] / H α ≈ /
4. These values are much lower com-pared to the central high-SB region. There (i.e., within the1.25 × − erg s − cm − arcsec − H α SB contour) we measure[O iii ] / H α = .
99 and [O iii ] / H β = .
97. The [O iii ] / H β (or [O iii ] / H α ) ratio depends on metallicity and on the ionisation pa-rameter (i.e. the ratio of ionising photon density to gas density).The lower ratio found in the ears could be due to lower densityand / or lower metallicity in the CGM. Unfortunately, other thanonly marginal detections in [O iii ] λ β , we do not findother nebular lines from the filaments in our data. Hence, wecan not make strong inferences on the physical conditions in theears. Even H β is not detected significantly enough to permit thecalculation of H α / H β ratios in those low-SB regions that wouldbe needed to asses the presence of dust.Compared to its main stellar body, SBS 0335-052E also ex-hibits significantly extended He ii λ ii image are co-spatialwith the more evolved SSCs, indicating that here the He ii emis-sion is related to post-main sequence stars (Wolf-Rayet stars)and / or their remnants (as first obtained by Izotov et al. 2006).Moreover, towards the NW the He ii emission is more elon-gated with a brightening on the rim of the shell. This pointstowards a locally enhanced UV radiation field caused by theshell shockfront (Izotov et al. 2001; Thuan & Izotov 2005).The total He ii flux, obtained by integrating over all spaxelswithin the 2 . × − erg s − cm − arcsec − SB contour, is F HeII = (2 . ± . × − erg s − cm − . Other possible sources for pro-ducing He ii ionising E γ > α line in each spaxel.The resulting line of sight velocity field is shown in Figure 3.Within the high-SB region our velocity field agrees with theresults and Moiseev et al. (2015) who traced H α emissionto ∼ − erg s − cm − arcsec − . These authors interpreted thecentral blue-shifted arc-like features as kinematic disturbancescaused by the expansion of the super-shell. The shell structuremanifests itself also in a locally double peaked H α profile, whichis however only resolved in the VLT / GIRAFFE data from Izo-tov et al. (2006) that was taken at ∼ × higher spectral reso-lution compared to MUSE. The newly detected filaments con-nect seamlessly in velocity space to the central velocity field.Especially, the ear towards the north shows the same blueshiftas the shell. The extended kinematical imprint of the shell in thefilaments argues for them being created by feedback from thestarburst. To quantify the ionised gas kinematics we calculatethe shearing velocity v shear and the intrinsic velocity dispersion σ (see review by Glazebrook 2013). Following the proceduresdetailed in Herenz et al. (2016) we obtain v shear = . − and σ = . − . Our derived σ is in excellent agree-ment with the measurement by Moiseev et al. (2015): σ = . ± . − . As v shear /σ = . <
1, SBS 0335-052Equalifies as a dispersion dominated system. Dispersion domi-nated kinematics are commonly found among high-redshift star-forming galaxies (e.g., Turner et al. 2017). Herenz et al. (2016)showed that v shear /σ < ffi cient, criterion for galaxies to have Lyman α photonsescaping. The escape of Lyman α photons can be theoreticallylinked to the leakage of LyC emission (Verhamme et al. 2015, forrecent observational evidence see Verhamme et al. 2017). How-ever, towards our line of sight Ly α is strongly absorbed (Thuan& Izotov 1997; James et al. 2014), and there are only tentativesignatures of a possible extended low-SB Ly α halo (Östlin et al. We correct for instrumental broadening using the wavelength depen-dent width of the MUSE line spread function given in Fig. 5 of Husseret al. (2016).Article number, page 2 of 5. C. Herenz et al.: Ionised gas tails in the halo of SBS 0335-52E F H α [ − e r g s − c m − ] F [ O III ] [ − e r g s − c m − ] Fig. 1. H α ( left ) and [O iii ] λ right ) narrow band images of SBS 0335-052E created from the MUSE datacube. East is left and North is up.The colour scale within the high-SB region encodes fluxes from 10 − to 10 − erg s − cm − for H α and 8 × − to 8 × − erg s − cm − for [O iii ],while for the outer low surface-brightness region the encoding is shown by the colour bar on the right. An asinh-scaling is used. SB contours aredrawn at [2 . , , . , × − erg s − cm − arcsec − for H α and [0 . , , . , × − erg s − cm − arcsec − for [O iii ]. To highlight thelow-SB features the images have been smoothed with a σ = (cid:48)(cid:48) ) Gaussian kernel. F H e II [ − e r g s − c m − ] Fig. 2. He ii narrow band image magnified on the central region. Theyellow contour is the 1 . × − erg s − cm − arcsec − contour from theH α image (Fig. 1, left). Colour scale in the inner high-SB region extendsfrom 3 × − to 3 × − erg s − cm − arcsec − , while for the outer regionit is shown by the colour bar on the right. An asinh-scaling is used. Con-tours correspond to [2 . , , × − erg s − cm − arcsec − . The in-set in the bottom left shows the emsission line free HST WFC3 / F550Mimage from the galaxy within the region indicated by the red square. α ) escape in this galaxy.Lastly, we also point out that the galaxy is embedded in dif-fuse extended H α emission. Detections of di ff use ionised halogas far away from the main stellar body have been presentedfor individual spiral galaxies (e.g., Hlavacek-Larrondo et al.2011) and in stacks of SDSS spectra (Zhang et al. 2016). Mea-surements of this di ff use component could provide constraintson the multi-phase nature of the CGM. We claim a signifi-cant detection out to 15 (cid:48)(cid:48) (4.1 kpc) from the main stellar bodyat ∼ − erg s − cm − arcsec − (lowest iso-SB contour in Fig-ure 1). Constructing a SB profile without including the regionsof the ears, we even measure di ff use emission out to 30 (cid:48)(cid:48) at (cid:38) − erg s − cm − arcsec − . However, given the compactness v r a d [ k m s − ] Fig. 3. H α line of sight velocity field with H α SB contours from Fig. 1.A background [O ii ] λ z = .
78 coincident with H α contaminated the fit towards the north and hence the respective regionis masked (black circle). . We spatially smoothed the datacube with atophat filter ( r = (cid:48)(cid:48) ) to enhance the signal to noise for the fit. of the high-SB region – 95% of the total encircled energy arewithin 2 (cid:48)(cid:48) – and that the SB-profile spans more than six ordersof magnitude within 30 (cid:48)(cid:48) , di ff use scattered light from telescopeoptics and / or the atmosphere can not be neglected as a cause forthe observed halo (Sandin 2014). As detailed by Sandin (2014)an accurate determination of the MUSE point-spread out to radii (cid:38) (cid:48)(cid:48) would be needed to accurately account for this e ff ect, butsuch a measurement is not available.
4. Discussion and Conclusion
Star formation in SBS 0335-052E is concentrated in six SSCs(Thuan et al. 1997) that have ages (cid:46)
10 Myr and masses ∼ × − M (cid:12) . The two oldest ( ∼ ∼
10 – 20 × M (cid:12) ) SSCs are in the north and the twoyoungest ( (cid:46) ∼ × M (cid:12) ) are inthe south (Thompson et al. 2009; Adamo et al. 2010). The star-burst might have been triggered either by the recent encounter Article number, page 3 of 5 & A proofs: manuscript no. sbs0335_letter_FINAL with its western companion or by a weak tidal interaction withthe massive spiral NGC 1376 ( M ∼ M (cid:12) ), which is found ata projected distance of 150 kpc (Pustilnik et al. 2001). The SSCsshow an age-sequence within 490 pc (1.9 (cid:48)(cid:48) ) from NW to south-east (SE). The age gradient points towards a scenario where feed-back from the oldest cluster in the NW ignited subsequent star-formation (SF) in the SE regions. This picture of unidirectionalpropagating SF as the main driver of the buildup of SBS 0335-052E is strengthened by an age gradient in faint stellar clustersdetected with an unsharp-masking technique out to (cid:38) i images ( ∼ (cid:48)(cid:48) beam) by Ektaet al. (2009) show indeed that the highest column densities( N HI ∼ cm − ) are confined to the south-east, i.e. oppositethe direction of the expanding shell. With increased sensitivity atlower spatial resolution ( ∼ (cid:48)(cid:48) beam), H i at lower column density( ∼ cm − ) aligned with the direction of the shell becomesvisible. Both these observational facts are consistent with the in-voked scenario of feedback driven SF propagation in the high-density SE regions, and the feedback driven compression of neu-tral gas at the shell front towards the regions of lower density inthe NE. At even lower resolution and higher sensitivity the im-ages by Ekta et al. (2009) reveal a tidal feature slightly west ofthe NW ionised filament with N HI ≈ × cm − . While thespatial resolution at this sensitivity appears insu ffi cient to infer adirect spatial connection between filaments and neutral halo gas,it still shows that the bulk of N HI (cid:38) cm − gas must be withinthe MUSE FoV, and that both the N and NW filament extend intoregions of lower neutral columns. Thus, while we do not coverthe full extend of the filaments, they appear to have penetratedcompletely through the bulk of the neutral gaseous halo of thesystem. Consequently, the reduced neutral fraction within thoseionised filaments could advocate an escape of LyC photons fromthe starburst into the intergalactic medium.To conclude, we have presented arguments for the ionised fil-aments of SBS 0335-052E towards the N and NW being causedby SF-driven feedback, while feedback driven SF propagationproceeds towards the opposite direction in this galaxy. Inter-estingly, a large fraction of extremely metal-deficient star-burstgalaxies show cometary morphologies. This is regarded as evi-dence for unidirectional SF propagation being common in suchsystems (Papaderos et al. 2008). Moreover, 7%-10% of high- z galaxies show similar morphologies (Straughn et al. 2006). Inorder to explain cometary morphologies (also known as “tad- pole” galaxies) also other mechanisms have been put forth,namely early-stage mergers or stream-driven gas accretion fromthe cosmic-web (e.g., Rhoads et al. 2005; Sánchez Almeida et al.2014; Straughn et al. 2015; Lagos et al. 2016). Nevertheless, wespeculate that if propagating SF is indeed an important driver ofdwarf galaxy formation at high redshift, a highly anistropic LyCescape into small solid angles would be expected. In this caseonly a small fraction of LyC leakers will be directly detectable.More importantly, accounting for this e ff ect might be importantwhen budgeting the ionising photon output from low-mass star-forming galaxies in order to asses their contribution for reionis-ing the universe. Acknowledgements.
We thank P. M. Weilbacher for feedback on handling thesky-oversubtraction issue described in Sect. 2. M.H. acknowledges the supportof the Swedish Research Council, Vetenskapsrådet and the Swedish NationalSpace Board (SNSB), and is Fellow of the Knut and Alice Wallenberg Foun-dation. This work was supported by Fundação para a Ciência e a Tecnolo-gia (FCT) through national funds and by FEDER through COMPETE by thegrants UID / FIS / / / FIS-AST / / / / / CP1191 / CT0002.
References
Adamo, A., Zackrisson, E., Östlin, G., & Hayes, M. 2010, ApJ, 725, 1620Bacon, R., Accardo, M., Adjali, L., et al. 2010, in Society of Photo-OpticalInstrumentation Engineers (SPIE) Conference Series, Vol. 7735, Society ofPhoto-Optical Instrumentation Engineers (SPIE) Conference SeriesBik, A., Östlin, G., Hayes, M., et al. 2015, A&A, 576, L13Cooper, J. L., Bicknell, G. V., Sutherland, R. S., & Bland-Hawthorn, J. 2008,ApJ, 674, 157Ekta, B., Pustilnik, S. A., & Chengalur, J. N. 2009, MNRAS, 397, 963Faisst, A. L. 2016, ApJ, 829, 99Finley, H., Bouché, N., Contini, T., et al. 2017 [ ]Fujita, A., Martin, C. L., Mac Low, M.-M., & Abel, T. 2003, ApJ, 599, 50Glazebrook, K. 2013, PASA, 30, 56Guseva, N. G., Izotov, Y. I., Fricke, K. J., & Henkel, C. 2017, A&A, 599, A65Hayes, M. 2015, PASA, 32, 27Herenz, E. C., Gruyters, P., Orlitova, I., et al. 2016, A&A, 587, A78Herenz, E. C. & Wisotzki, L. 2017, A&A, 602, A111Hlavacek-Larrondo, J., Marcelin, M., Epinat, B., et al. 2011, MNRAS, 416, 509Husser, T.-O., Kamann, S., Dreizler, S., et al. 2016, A&A, 588, A148Izotov, I. I., Guseva, N. G., Lipovetskii, V. A., Kniazev, A. I., & Stepanian, J. A.1990, Nature, 343, 238Izotov, Y. I., Cha ff ee, F. H., & Schaerer, D. 2001, A&A, 378, L45Izotov, Y. I., Guseva, N. G., Fricke, K. J., & Papaderos, P. 2009, A&A, 503, 61Izotov, Y. I., Schaerer, D., Blecha, A., et al. 2006, A&A, 459, 71Izotov, Y. I., Thuan, T. X., Guseva, N. G., & Liss, S. E. 2017 [ ]James, B. L., Aloisi, A., Heckman, T., Sohn, S. T., & Wolfe, M. A. 2014, ApJ,795, 109Lagos, P., Demarco, R., Papaderos, P., et al. 2016, MNRAS, 456, 1549Moiseev, A. V., Tikhonov, A. V., & Klypin, A. 2015, MNRAS, 449, 3568Obreschkow, D., Heywood, I., & Rawlings, S. 2011, ApJ, 743, 84Östlin, G., Hayes, M., Kunth, D., et al. 2009, AJ, 138, 923Papaderos, P., Guseva, N. G., Izotov, Y. I., & Fricke, K. J. 2008, A&A, 491, 113Papaderos, P., Izotov, Y. I., Fricke, K. J., Thuan, T. X., & Guseva, N. G. 1998,A&A, 338, 43Prestwich, A. H., Tsantaki, M., Zezas, A., et al. 2013, ApJ, 769, 92Pustilnik, S. A., Brinks, E., Thuan, T. X., Lipovetsky, V. A., & Izotov, Y. I. 2001,AJ, 121, 1413Pustil’nik, S. A., Lipovetsky, V. A., Izotov, Y. I., et al. 1997, Astronomy Letters,23, 308Rauch, M., Becker, G. D., & Haehnelt, M. G. 2016, MNRAS, 455, 3991Rhoads, J. E., Panagia, N., Windhorst, R. A., et al. 2005, ApJ, 621, 582Rutkowski, M. J., Scarlata, C., Henry, A., et al. 2017, ApJ, 841, L27Sánchez Almeida, J., Morales-Luis, A. B., Muñoz-Tuñón, C., et al. 2014, ApJ,783, 45Sandin, C. 2014, A&A, 567, A97Schaerer, D., Izotov, Y. I., Verhamme, A., et al. 2016, A&A, 591, L8Straughn, A. N., Cohen, S. H., Ryan, R. E., et al. 2006, ApJ, 639, 724Straughn, A. N., Voyer, E. N., Eufrasio, R. T., et al. 2015, ApJ, 814, 97Thompson, R. I., Sauvage, M., Kennicutt, R. C., et al. 2009, ApJ, 691, 1068Thuan, T. X., Bauer, F. E., Papaderos, P., & Izotov, Y. I. 2004, ApJ, 606, 213Thuan, T. X. & Izotov, Y. I. 1997, ApJ, 489, 623Thuan, T. X. & Izotov, Y. I. 2005, ApJS, 161, 240Thuan, T. X., Izotov, Y. I., & Lipovetsky, V. A. 1997, ApJ, 477, 661 Article number, page 4 of 5. C. Herenz et al.: Ionised gas tails in the halo of SBS 0335-52E
Turner, O. J., Cirasuolo, M., Harrison, C. M., et al. 2017, ArXiv e-prints[ arXiv:1704.06263 ]Verhamme, A., Orlitová, I., Schaerer, D., & Hayes, M. 2015, A&A, 578, A7Verhamme, A., Orlitová, I., Schaerer, D., et al. 2017, A&A, 597, A13Weilbacher, P. M., Streicher, O., Urrutia, T., et al. 2014, in Astronomical Soci-ety of the Pacific Conference Series, Vol. 485, Astronomical Data AnalysisSoftware and Systems XXIII, ed. N. Manset & P. Forshay, 451Wisotzki, L., Bacon, R., Blaizot, J., et al. 2016, A&A, 587, A98Zhang, H., Zaritsky, D., Zhu, G., Ménard, B., & Hogg, D. W. 2016, ApJ, 833,276]Verhamme, A., Orlitová, I., Schaerer, D., & Hayes, M. 2015, A&A, 578, A7Verhamme, A., Orlitová, I., Schaerer, D., et al. 2017, A&A, 597, A13Weilbacher, P. M., Streicher, O., Urrutia, T., et al. 2014, in Astronomical Soci-ety of the Pacific Conference Series, Vol. 485, Astronomical Data AnalysisSoftware and Systems XXIII, ed. N. Manset & P. Forshay, 451Wisotzki, L., Bacon, R., Blaizot, J., et al. 2016, A&A, 587, A98Zhang, H., Zaritsky, D., Zhu, G., Ménard, B., & Hogg, D. W. 2016, ApJ, 833,276