Detection of an ionized gas outflow in the extreme UV-luminous star-forming galaxy BOSS-EUVLG1 at z=2.47
J. ?lvarez-Márquez, R. Marques-Chaves, L. Colina, I. Pérez-Fournon
AAstronomy & Astrophysics manuscript no. 39375corr © ESO 2021February 4, 2021
Detection of an ionized gas outflow in the extreme UV-luminousstar-forming galaxy BOSS-EUVLG1 at z=2.47
J. Álvarez-Márquez , R. Marques-Chaves , L. Colina , , and I. Pérez-Fournon , Centro de Astrobiología (CSIC-INTA), Carretera de Ajalvir, 28850 Torrejón de Ardoz, Madrid, Spain; e-mail: [email protected] International Associate, Cosmic Dawn Center (DAWN) Instituto de Astrofísica de Canarias, C / Vía Láctea, s / n, E-38205 San Cristóbal de La Laguna, Tenerife, Spain Universidad de La Laguna, Dpto. Astrofísica, E-38206 San Cristóbal de La Laguna, Tenerife, Spain
ABSTRACT
BOSS-EUVLG1 is the most ultraviolet (UV) and Ly α luminous galaxy to be going through a very active starburst phase detectedthus far in the Universe. It is forming stars at a rate (SFR) of 955 ±
118 M (cid:12) yr − . We report the detection of a broad H α componentcarrying 25% of the total H α flux. The broad H α line traces a fast and massive ionized gas outflow characterized by a total mass,log( M out [ M (cid:12) ]) , of 7.94 ± out ) of 573 ±
151 km s − and an outflowing mass rate ( ˙ M out ) of44 ±
20 M (cid:12) yr − . The presence of the outflow in BOSS-EUVLG1 is also supported by the identification of blueshifted UV absorptionlines in low and high ionization states. The energy involved in the H α outflow can be explained by the ongoing star formation, withoutthe need for an active galactic nucleus (AGN) to be included in the scenario. The derived low mass-loading factor ( η = ± ≤ − hasthe capacity to escape the gravitational potential and to enrich the surrounding circumgalactic medium at distances above several tensof kpc. The ionized phase of the outflow does not carry su ffi cient mass or energy to play a relevant role in the evolution of the hostgalaxy nor in the enrichment of the intergalactic medium. As predicted by some recent simulations, other phases of the outflow couldbe responsible for most of the outflow energy and mass in the form of hot X-ray emitting gas. The expected emission of the extendedX-ray emitting halo associated with the outflow in BOSS-EUVLG1 and similar galaxies could be detected with the future ATHENAX-ray observatory, however, there are no methods at present that would assist in their spatial resolution. Key words.
Galaxies: starburst – galaxies: high-redshift – galaxies: kinematics and dynamics – galaxies: ISM – intergalactic medium
1. Introduction
Cosmological simulations exploring the formation and evolutionof galaxies in the early universe invoke outflows to explain theobserved stellar mass function of galaxies. While outflows pro-duced by luminous accreting black holes are favoured for thehigh-mass end, outflows associated with intense starbursts areclaimed for the low-mass end (e.g., Somerville & Davé 2015;Naab & Ostriker 2017; Nelson et al. 2019; Mitchell et al. 2020;Valentini et al. 2020). These outflows do not only control thegrowth of stellar mass in galaxies across redshifts, they are alsoresponsible for the metal enrichment of the circum- and inter-galactic medium (Kim et al. 2020).Observationally, outflows in galaxies are a universal phe-nomenon that have been detected at both low (e.g., Arribas et al.2014; Cicone et al. 2016; Concas et al. 2019) and high red-shifts (e.g., Freeman et al. 2019; Förster Schreiber et al. 2019;Swinbank et al. 2019; Finnerty et al. 2020), covering all typesof galaxies, ranging from those featuring an active galactic nu-cleus (AGN) (e.g., Harrison et al. 2014; King & Pounds 2015;Fiore et al. 2017; Harrison et al. 2018) to those featuring star-bursts (e.g., Heckman et al. 1990; Veilleux et al. 2005, 2020).The complex, intrinsic multi-phase nature (molecular, neutral,and ionized) of galactic outflows has been traced at all wave-lengths from the X-rays (Strickland et al. 2004), to ultraviolet(Steidel et al. 2010), optical (Arribas et al. 2014; Cazzoli et al. 2016; Cicone et al. 2016; Concas et al. 2019), near-infrared (Hill& Zakamska 2014; Emonts et al. 2017), far-infrared (Veilleuxet al. 2013), and millimeter wavelengths (Cicone et al. 2014;Pereira-Santaella et al. 2018; Spilker et al. 2020b,a; Veilleuxet al. 2020).Most of the studies of ionized outflows at high- z have fo-cused on galaxies located on the main sequence (MS) of star-forming galaxies (SFG), namely, galaxies with average star for-mation rates (SFRs) of several to a few tens of M (cid:12) yr − (FörsterSchreiber et al. 2019; Freeman et al. 2019; Swinbank et al.2019). Ionized gas outflows appear to be frequent at the peak ofthe cosmic star formation history (i.e. z ∼ D H α survey (Förster Schreiber et al. 2019),about a third of the non-AGN dominated SFGs show evidenceof ionized gas outflows with masses (M out ) of 10 to 10 M (cid:12) ,outflowing mass rates ( ˙ M out ) of few tens M (cid:12) yr − , and low mass-loading factors ( η = ˙ M out / SFR) of ∼ ∼ ffi cient to explain low-mass galaxy formation e ffi ciency,additionally asserting that cold molecular and hot X-ray emittingphases are expected to dominate the mass, energy, and momen-tum of the outflows as observed in low-z galaxies (e.g., Veilleuxet al. 2020; Strickland et al. 2004). Article number, page 1 of 8 a r X i v : . [ a s t r o - ph . GA ] F e b & A proofs: manuscript no. 39375corr
Studies of outflows in high-z galaxies with extreme star for-mation, that is, with SFRs of several hundreds to thousand M (cid:12) yr − , are far more limited. A recent study of massive, dusty star-forming galaxies (DSFG) at z > max ∼ − ˘1 ), mass outflow-ing rates of several M (cid:12) yr − , and mass-loading factors just be-low one (Spilker et al. 2020a). However, the presence of the(partially) ionized outflowing component was not detected inthose sources with a high quality [CII]158 µ m line and with clearmolecular outflow, as traced by the OH119 µ m absorption line.High-velocity outflows have also been detected in z > − . There-fore, high-velocity outflows appear to be common in both in-frared (i.e., dusty) and ultraviolet (i.e., dust-poor) luminous star-forming galaxies at high redshifts. However, the multi-phasestructure and their impact on the formation and evolution ofgalaxies is far from clear. Detailed multi-wavelength studies ofindividual prototypes, or small samples of galaxies, are requiredto quantify and establish the relevance of each of the main out-flow components (cold dense molecular, warm (partially-) ion-ized, or hot di ff use X-rays) in the stellar buildup of galaxies andin the metal enrichment of the intergalactic medium (IGM).Very recently, an extremely UV luminous galaxy, BOSS-EUVLG1, at a redshift of 2.47, was discovered in the studyof Marques-Chaves et al. (2020; hereafter, MC20 ) withinthe extended Baryon Oscillation Spectroscopic Survey (Abol-fathi et al. 2018). BOSS-EUVLG1 is a compact (size ≤ ∼ ∼ (cid:12) yr − , with low metallicity and dust content(12 + log(O / H) = ± (cid:39) α luminosi-ties ( M UV < − .
5; log( L Ly α [erg s − ]) > (cid:12) yr − . In addition, BOSS-EUVLG1 has a spe-cific star formation (sSFR) ∼ ±
15 Gyr − , a factor 30 timeshigher than 10 M (cid:12) main-sequence galaxies at redshifts 2-2.5(Whitaker et al. 2015) but in the range of the H α -excess galaxiesat redshifts ∼ Spitzer
UltraVISTA ultra-deep stripes survey (SMUVS survey; Caputiet al. 2017). An important fraction (15%) of the SMUVS galax-ies in the 9.4 ≤ log(M stellar [ M (cid:12) ]) ≤ (cid:12) yr − and sSFR well above 24 Gyr − , and up to 120 Gyr − . TheSMUVS galaxies in the starburst phase account for more than50% of the cosmic SFR density at redshifts ∼ α -excessgalaxies, and could be important contributors to the SFR densityat intermediate and high redshifts. BOSS-EUVLG1, therefore,brings on the opportunity to investigate the presence and prop-erties of ionized gas outflows for the first time in this class ofextreme star-forming galaxies.In this article, we report the detection of the ionized phaseof the outflow in BOSS-EUVLG1 using the H α emission line.Section 2 describes the observations and calibrations of the H α spectrum. Section 3 presents the analysis of H α emission andrest-frame UV absorption lines. The results of the analysis and interpretation as outflowing material are presented in Section 4,including the overall kinematic and mass ( § § § § § § ATHENA ( § H = . − Mpc − and Ω m = .
307 (Planck Collaboration et al. 2016) as cosmologicalparameters and the Salpeter initial mass function (IMF; Salpeter1955) throughout this paper. A factor of 1.66 (Newman et al.2012) is used to convert the SFRs and stellar masses (M stellar )from a Chabrier (2003) to a Salpeter (1955) IMF when usingsome published data.
2. Observations
Near-infrared (Near-IR) long-slit spectroscopy of BOSS-EUVLG1 was obtained with the
Espectrógrafo MultiobjetoInfra-Rojo (EMIR) instrument on the 10.4m Gran Telesco-pio Canarias (GTC). Observations were taken in two servicemode nights – 2020 February 10 and May 8 – in bright Moonand subarcsec seeing ( (cid:39) (cid:48)(cid:48) full width half maximum) condi-tions, as part of the GTC program GTCMULTIPLE2E-19A (PI:R. Marques-Chaves). The observations used the K grism with adispersion of 1.71 Å pix − and a 0.8 (cid:48)(cid:48) -wide long-slit, providinga spectral resolution of 82 km s − at ∼ µ m. The slit was cen-tered on a bright ( K = . ± .
02, AB) reference star (SDSSJ122039.27 + (cid:39) (cid:48)(cid:48) SW of BOSS-EUVLG1) and ori-ented so as to encompass BOSS-EUVLG1, with a sky positionangle of PA = ◦ (north toward east). The total exposure timeon-source was 6400 s, split into 40 individual exposures of 160 seach, using an ABBA nodding pattern with 6 (cid:48)(cid:48) o ff set along theslit.Data reduction was performed using the EMIR pipeline(P y E mir ) . Near-IR 2D spectra were flux calibrated using theHIP 59174 telluric standard star observed in the second night.To account for slit-losses, the fluxes of BOSS-EUVLG1 and thereference star were matched to those obtained from photome-try. The 1D spectrum of H α was extracted applying an optimalextraction algorithm (Horne 1986) in an aperture of 3 (cid:48)(cid:48) . Figure1 shows the final 2D calibrated and 1D extracted spectrum ofBOSS-EUVLG1.
3. Data analysis
The H α emission of BOSS-EUVLG1 exhibits an asymmetricline profile characterized by a main component with a blue-shifted wing (Figure 1). We implement a two-component Gaus-sian fit, using a Python version of the MPFIT routine (Markwardt2009) , to characterize the H α emission. The [N II]6548,6583Ådoublet and a continuum emission, together with the instrumen-tal broadening, are also included in the fit. Figure 1 shows theresults of the two-components fit with the narrow and a blue-shifted broad emission of H α . Additionally, we test the possi-bility to fit the profile with just one Gaussian component but itfails to reproduce its asymmetry, yielding to a χ
50% largerthan using the two-component fit. The errors on the parameters https://pyemir.readthedocs.io/en/latest/index.html MPFIT Python version used: https://github.com/segasai/astrolibpy/tree/master/mpfit
Article number, page 2 of 8. Álvarez-Márquez et al.: Ionized gas outflow in BOSS-EUVLG1 are estimated using Monte Carlo. The noise of the spectrum ismeasured as the root mean square (rms) of the continuum sur-rounding the H α + [NII] lines. This noise is used to generate newspectra (3000), where a random Gaussian noise with a sigmaequal to the rms is added to the original spectrum before thelines are fitted again.The H α narrow component has a flux (F N ) of 1 . ± . × − erg s − cm − and a full width at half maximum (FWHM N )of 241 ±
11 kms − . The broad component is characterized byan H α flux (F B ) of 4 . ± . × − erg s − cm − , a width(FWHM B ) of 511 ±
145 kms − , and a velocity o ff set (V peak ) of − ±
87 kms − (i.e., blueshift) with respect to the narrow com-ponent. In addition, the [N II]6583Å emission appears as a weakline with a measured flux of (5 . ± . × − erg s − cm − for the best fit. These results are consistent, within the uncertain-ties, with the slight asymmetries in the low-resolution spectrumof the [O iii ] 4959 , ∼ iii ] 4959 , ii ii ii iv , α line. Asseen in this figure, the centroids of low- and high-ionization ISMabsorption lines are clearly blueshifted with respect to the sys-temic redshift by ∆ v low − ion = − ±
75 km s − and ∆ v high − ion = − ±
75 km s − , respectively. The errors account for the un-certainties of the spectral bin size and the standard deviation ofthe centroid of each individual ISM line. Due to the low spec-tral resolution of the rest-frame UV spectrum ( R ∼ M out , V max , column density, covering fraction, etc.), is not possible asit requires a better spectral resolution. However, they do presentfurther evidence for the existence of outflows in this galaxy and,therefore, lend their support to the interpretation of the broad H α component as outflowing material.
4. Results
The narrow component of H α is assigned to the emission in thehost galaxy while the broad component is identified with out-flowing ionized gas. The main properties of this outflowing ma-terial (i.e., outflowing mass, velocity of the bulk of the outflowand its maximal velocity) are derived from the flux, o ff set veloc-ity, and line profile of the broad H α line component, assumingthe electron density is known. For BOSS-EUVLG1, the electrondensity is derived from the low-ionization [O ii ] 3729 , ± − (Kew-ley et al. 2019). This value provides the mean electron densityof the ionized gas in the galaxy. Since no direct measurement F l u x [ e r g / s / c m / Å ] = 1.5 Spectral LSFBOSS-EUVLG1 spec.Best-fit modelH Narrow comp.H Broad comp.[NII]6548,6583 Å Velocity [km/s] R e s i d u a l s Fig. 1.
Observed H α emission of BOSS-EUVLG1. Upper panel showsthe near-IR 2D calibrated spectrum centered on the H α emission. Bot-tom panel illustrates the observed H α emission line profile of BOSS-EUVLG1, together with the two-components Gaussian fit and its resid-uals. Gray line: 1D observed H α spectrum. Black line: Best two-component Gaussian H α fitted model. Red and blue lines: Narrow andbroad H α components. Green line: [N II]6548,6583Å lines. Gray dottedline: Illustration of observed spectral line-spread function, derived froma sky line at 2.25 µ m, normalized to the peak of the H α emission, andcentered within a velocity of -1200km / s. The χ , calculated from -1000to 750 km s − , is also included. of the electron density in the outflow is available within ourdata, this value of the density will also be assumed for out-flowing material in all the subsequent analysis and derivationof the physical quantities. This is a valid assumption. The me-dian electron density of the outflows traced by the low ionization[SII]6717,6731Å lines in a large sample of low-z U / LIRGs is afactor of 1.23 larger than the density of the ionized gas in thesegalaxies (Arribas et al. 2014).The peak velocity (V peak ) of the broad H α component isblueshifted by − ±
87 km s − . High-spatial resolution zoom-in simulations of the ionized gas velocity field traced by the H α emission line in z ∼ α line component in two-third of the simu-lated galaxies and with blueshifted velocities of 110 ±
20 kms − (Ceverino et al. 2016). This result is explained as a directconsequence of the extended structure of the outflowing mate-rial, coming from the near-side of the galaxy, above the planeof the galaxy (Ceverino et al. 2016). These simulations covergalaxies with stellar masses 1-4 × M (cid:12) , similar to that ofBOSS-EUVLG1, but with SFR ∼ (cid:12) yr − , namely, fac-tors of 16-50 lower than the SFR derived for BOSS-EUVLG1. Asimilar spatial and velocity structure can be expected in BOSS-EUVLG1 where, due to its higher SFR, the outflowing materialcould be launched to distances further out, and up to several kpcabove the plane of the galaxy. Article number, page 3 of 8 & A proofs: manuscript no. 39375corr
The observed H α outflowing velocity is substantially lowerthan that measured using the UV absorption lines ( ∆ v low − ion = − ±
75 km s − and ∆ v high − ion = − ±
75 km s − ), but itis in agreement with the typical blueshifted velocity observed inLyman Break Galaxies at intermediate redshifts ( − ±
16 kms − , Steidel et al. 2010). Such di ff erences are expected becausethe outflowing gas is stratified, that is, regions of di ff erent ioniza-tion state (neutral versus ionized gas, low- versus high-ionizationISM) show di ff erent outflowing velocities (e.g., Erb 2015; Rupke2018; de la Cruz et al. 2020).The kinematics of the H α outflow is characterized by thepeak velocity o ff set and the outflow velocity ( V out = V peak + × σ B ). For a FWHM B of 511 ±
145 km s − and a V peak of -139 ±
87 km s − , the V out correspond to 573 ±
151 km s − . FollowingColina et al. (1991), the total outflowing ionized mass (M out ∝ L(H α ) / n e ) corresponds to log( M out [ M (cid:12) ]) = . ± .
15, derivedfrom the extinction corrected (E(B-V) = α luminosity. A more conservative estimate of the total outflow-ing ionized mass can be derived by assuming only the ionizedgas with velocities that are well above the expected dispersionor rotational velocities derived from the narrow H α component.Considering velocities only above 241 km s − (FWHM N ) andthe o ff set velocity of the outflowing component, the lower limitto the outflowing mass would correspond to about a third of thetotal mass. Therefore, the outflowing ionized mass would be inthe mass range of 7.42 < log( M out [ M (cid:12) ]) < . The outflowing ionized mass rate is given as K × M out × V out × size − , where K is a factor that depends on the geometry, withvalues between 1 and 3 for an elongated and spherical filled out-flow, respectively (Lutz et al. 2020). Here, we adopt a value ofone as the spatially resolved outflows in nearby galaxies supportin general elongated geometries (Pereira-Santaella et al. 2018)and the limited spatial resolution here does not provide informa-tion on the geometry of the outflow. For a V out of 573 ±
151 kms − and an outflow size comparable to the UV-continuum source(i.e., 1.2 kpc, MC20), the outflowing mass rate corresponds to44 ±
20 M (cid:12) yr − .The SFR derived from the observed H α flux corrected bythe internal extinction (E(B-V) = ±
118 M (cid:12) yr − for a Salpeter IMF (Salpeter 1955) and a SFR = × − × L ( H α ) (Kennicutt 1998). The corresponding mass-loading factor ( η = ˙ M out / SFR) is 0.05 ± α emis-sion line, appears irrelevant in the quenching of the extreme on-going star formation in BOSS-EUVLG1. However, other phasescould serve as the dominant phases of the outflow, carrying mostof its mass and energy, therefore acting as the major contribu-tors to the removal of gas and quenching of star formation (seediscussion in § For young starbursts, the rates of outflowing mass, kinetic en-ergy, and momentum liberated by supernovae explosions andstellar winds are a function of the SFR. Following Veilleuxet al. (2005), these quantities correspond to 248 M (cid:12) yr − , 6 . ± . × erg s − , and 4 . ± . × dynes for a SFR of 955 M (cid:12) yr − , respectively. The measured rate of kinetic energy(0 . × ˙ M out × V out ) and momentum ( ˙ M out × V out ) of the H α outflowcorresponds to 4.6 ± × erg s − and 1.6 ± . × dynes,respectively. The ionized outflow represents in BOSS-EUVLG1a minor fraction of the outflowing mass (18%), kinetic energy(0.7%), and momentum (3.3%) rates liberated by the starburst.There is therefore no need to invoke the presence of an AGN inthe nucleus to explain the energetics of the warm ionized out-flow. However, unless the starburst in BOSS-EUVLG1 is ex-tremely ine ffi cient in converting the available energy into out-flows in the surrounding interstellar medium, contributors to theoutflow could still be present and locked in other phases such asin the hot, X-ray emitting gas (see § Independent of the energy source, gas outflows in galaxies arerelevant, not only as potential mechanisms of star formationquenching, but also as one of the main mechanisms for ex-pelling gas from galaxies into the intergalactic medium, withsubsequent metal enrichment (Somerville & Davé 2015). Anestimate of the fraction of the outflowing material in BOSS-EUVLG1 that is capable of escaping the gravitational poten-tial of the galaxy and enrich the IGM can be derived assum-ing only the outflowing material with velocities above the es-cape velocity. For a galaxy represented by an isothermal spheretruncated to a maximum radius (R max ), the escape velocity for adynamical mass (M dyn ) within a radius (R) is given as V esc = (cid:112) × M dyn × G × (1 + ln [ R max / R ]) / (3 × R ) (e.g., Arribas et al.2014 and references therein). BOSS-EUVLG1 has a dynamicalmass of 1.8 × M (cid:12) , derived for a R of 1.2 kpc and velocitydispersion from the narrow H α component of 102 km s − (Bel-locchi et al. 2013). For a R max of 10 kpc, the escape velocitycorresponds to 372 km s − .The velocity of the outflow (V out =
573 km s − ) is 1.5 timeslarger than the escape velocity. Thus, only a small fraction ofthe total ionized outflowing material (15%, i.e., ∼ × M (cid:12) ),with velocities well above V esc , would potentially be able to es-cape and enrich the surrounding IGM at distances larger than10 kpc. Most of the outflowing material would not escape thegravity of the host galaxy and would be likely to fall back into thesystem, providing the raw material for subsequent delayed starformation episodes, as in low-z infrared luminous star-forminggalaxies (Arribas et al. 2014; Ceverino et al. 2016). We note,however, that H α only traces the warm (10 K) ionized gascomponent of the outflow. In fact, the peak velocities of theblueshifted low- and high-ionization UV ISM absorption lines( ∆ v low − ion ∼ −
305 km s − and ∆ v high − ion ∼ −
415 km s − ) arefactors of 2 and 3 larger than the corresponding H α (V peak = − ±
85 km s − ). This already demonstrates the existence ofoutflowing material that is rich in metals, with velocities higherthan those of the H α emitting gas – velocities that are highenough to escape into the IGM and contribute to its metal en-richment.
5. Discussion
BOSS-EUVLG1 is an extreme starburst galaxy with a star-forming rate of 955 ±
118 M (cid:12) yr − , which can be consid-ered an ultraluminous galaxy based on its UV luminosity (i.e., L ∼ L (cid:12) ). Its SFR is similar to those measured in high − z Article number, page 4 of 8. Álvarez-Márquez et al.: Ionized gas outflow in BOSS-EUVLG1 N o r m a li z e d F l u x H spectrumH NarrowH Broad
Velocity [km/s]
UV High-IonUV Low-Ion M UV [AB] v ] -25-24-23-22-21-20-400-300-200-1000 BOSS-EUVLG1LAEsLBGs ( U V L o w - I o n ) [ k m s Fig. 2.
View of the UV absorption lines and the UV peak velocity of BOSS-EUVLG1. Left panel: Comparison between the H α emission and theUV ISM absorption lines. Upper left panel: Normalized H α spectrum (black line), together with the narrow (red line) and broad (blue line) H α components. Bottom left panel: UV ISM absorption lines for two di ff erent ionization states. Orange line: Combination of UV ISM absorption linesof partially neutral elements (Si ii ii ii iv , α component are shown as color-coded vertical lines. Right panel: Peak velocity of theUV low-ionization ISM absorption lines as a function of the UV absolute magnitude. Yellow star: BOSS-EUVLG1. Blue dots: Lyman alphaemitters (Patrício et al. 2016; Marques-Chaves et al. 2020b). Red squares: Lyman break galaxies (Pettini et al. 2000; Quider et al. 2009, 2010;Dessauges-Zavadsky et al. 2010; Marques-Chaves et al. 2018). dust-enshrouded extreme starbursts (Reuter et al. 2020), andabout 10 to 100 times higher than that of massive star-forminggalaxies at intermediate redshifts (Förster Schreiber et al. 2019;Swinbank et al. 2019; Freeman et al. 2019; Newman et al.2012), and low − z luminous and ultraluminous infrared galaxies(U / LIRGs, Arribas et al. 2014). Moreover, BOSS-EUVLG1 isdeficient in terms of dust and is underluminous in the infrared(log( L IR [L (cid:12) ]) < / UV luminosity ratio oflog( L IR / L UV ) < − α ).The ratio of the H α broad- to narrow-component flux (F B / F N ) inBOSS-EUVLG1 is 0 . ± .
12. This value is lower than the ra-tios measured in low- and intermediate-z star-forming galaxieswhere the broad line component carries a larger fraction of theflux with typical F B / F N values in the 0.4 to 0.8 range, indepen-dent of the SFR albeit with a large scatter (upper left panel ofFigure 3). However, the F B / F N ratio as a function of the stellarmass follows the expected trend for star-forming galaxies (up-per right panel of Figure 3). These results clearly indicate thatthe ionized outflow component in BOSS-EUVLG1 is low for itsspecific star formation rate (sSFR = ±
15 Gyr − , MC20).BOSS-EUVLG1 is an extreme starburst galaxy with sSFR about30-50 times higher than that of Main Sequence star-forminggalaxies at z ∼ α emission. It appears as if the strength of the out-flow is dominated by the stellar mass of the system – and notby the strength of the star formation as given by the SFR. Thisis somehow unexpected given that in a starburst-dominated out-flow, the mass, momentum, and energy rates of the outflow aredirectly related to the total star formation rate (Heckman et al.1990; Veilleux et al. 2005). In addition, outflows in z ∼ B / F N values closeto 1 for all galaxies with SFR surface brightness above 1 M (cid:12) yr − kpc − and low values of about 0.2 for the galaxies belowthe SFR brightness threshold, independent of the stellar mass(Newman et al. 2012). According to its sSFR, SFR, and upperlimit on the size of the H α emission, BOSS-EUVLG1 would beclassified in the category of galaxies with strong outflows char-acterized by a F B / F N ratio of about 1. However, the measuredvalue of 0.34 is closer to that of weak outflows in starbursts witha low-SFR surface brightness. Explanations of the low F B / F N value in BOSS-EUVLG1 could invoke a qualitatively di ff erentmulti-phase structure and composition of the interstellar mediumand outflows in IR-luminous, namely, dustier, more molecular,and metal-rich; and in UV-luminous starbursts such as BOSS-EUVLG1, namely, dust-, molecular-, and metal-poor (see dis-cussion in § § Spitzer
UltraVISTA ultra-deepstripes survey (SMUVS survey; Caputi et al. 2017) has identi-fied a number IR-luminous, H α -excess galaxies at redshifts of ∼ ≤ log(M stellar [ M (cid:12) ]) ≤ (cid:12) yr − and sSFR well above 24 Gyr − , and up to120 Gyr − . Future H α spectroscopy of EUVLGs and SMUVS Article number, page 5 of 8 & A proofs: manuscript no. 39375corr F B / F N
10 100 1000SFR ([M /yr])2004006008001000 V o u t [ k m / s ] BOSS-EUVLG1 (This work)V out
SFR (This work)V out
SFR (Arribas+14)LIRGs low-z (Arribas+14)ULIRGs low-z (Arribas+14) log(M stellar [M ])9.5 10.0 10.5 11.0log(M stellar [M ])
SFGs 0.6 Fig. 3. Observational properties of the ionized outflow in BOSS-EUVLG1 (yellow star) together with samples of star-forming galaxies (SFGs) atintermediate redshifts including 0.6 < z < D survey (Förster Schreiber et al. 2019, values of V out are a privatecommunication), SFGs at z ∼ ∼ z ∼ α linecomponents (F B / F N ) as a function of the SFR (left) and M stellar (right). Bottom panels show the outflow velocity as a function of the SFR (left)and M stellar (right). The continuum line in the V out to SFR relation defines the best fit ( V out = × S FR . ) to the BOSS-EUVLG1, low- andintermediate-z data points. The dashed line represents the V out to H α -derived SFR relation derived for the low-z sample of U / LIRGs without AGNs(Arribas et al. 2014). galaxies in the starburst phase from the ground and with JWST(for z > 3) are required to confirm the strength of outflows inextreme starbursts (i.e., very large sSFR) and investigate it indepth to identify the potential di ff erences between the classes ofIR- and UV-luminous.The outflow velocity in BOSS-EUVlG1 is in agreementwith the expected value as extrapolated from previous V out ∝ SFR α relations derived in low-z U / LIRGs ( α = α ) of 0.23, which is consistent with thepredicted dependence with the bolometric luminosity of the star-burst (V out ∝ L . bol , Heckman et al. 2000). However, the out-flow velocity in BOSS-EUVLG1, 573 km s − , stands out againstthe measured weak dependence of the outflowing velocities asa function of the stellar mass. Velocities, V out , lower than 400km s − are measured in galaxies with log(M stellar [M (cid:12) ]) of 10.0(lower right panel of Figure 3). In summary, the observed prop-erties of the ionized outflowing gas in BOSS-EUVLG1 appear to be a consequence of its extremely high sSFR, that is, a verymassive, young, and compact starburst in a galaxy with relativelylow stellar mass. On the one hand, the V out follows the functionaldependence observed in star-forming galaxies at low and inter-mediate redshifts. On the other hand, the amount of outflowingmaterial, as identified by its low F B / F N , appears to be consistentwith the ratio expected by its low stellar mass.The blueshifts identified in the UV ISM absorption lines ofBOSS-EUVLG1 are also indicative of large-scale galactic out-flows, similar to that observed in many other star-forming galax-ies at low (e.g., Heckman et al. 2000; Rubin et al. 2014) andhigh redshift (e.g., Shapley et al. 2003; Steidel et al. 2010).The right panel of Figure 2 shows a comparison between theblueshift of the low-ionization ISM absorption lines (relative tothe systemic redshift) and the UV absolute magnitude M UV ofBOSS-EUVLG1 and other UV-luminous star-forming galaxiesat 2 < z < ∆ v canbe measured on an individual basis (Pettini et al. 2000; Quideret al. 2009, 2010; Dessauges-Zavadsky et al. 2010; Patrício et al.2016; Marques-Chaves et al. 2018, 2020b)). This figure clearlyshows a trend among ∆ v low − ion and M UV , indicating that UV-luminous star-forming galaxies, such as BOSS-EUVLG1, ex- Article number, page 6 of 8. Álvarez-Márquez et al.: Ionized gas outflow in BOSS-EUVLG1 perience outflows caused by supernova explosions and stellarwinds, with the peak velocity of these outflows increasing withthe absolute UV luminosity of the galaxy.A more direct comparison of other relevant physical quanti-ties of the ionized outflow, such as the rates of outflowing mass,kinetic energy, and momentum, as well as the mass-loading fac-tors, are di ffi cult to establish. These quantities have a strong de-pendence on the value adopted for the electron density and sizeof the H α emitting nebulae. In some samples, such as the MOS-DEF survey of z ∼ − is adopted for the outflowing material. Other studies of outflowsin mean sequence star-forming galaxies at z ∼ − for the composite spectra of di ff erent galaxies of the sample. Fi-nally, the recently completed KMOS-3D survey of z ∼ . − . − from the stack spectra of the star-forming driven outflows.The electron densities measured or adopted in all these interme-diate redshift samples are factors of ∼ − ) measured in BOSS-EUVLG1, while the sizesare consistently larger by factors of 3-10, making any compari-son between the ionized outflow in BOSS-EUVLG1 with othergalaxy samples rather uncertain in absolute terms. However, ifthe di ff erences in the size and electron density were real, thesewould indicate that the outflow in BOSS-EUVLG1 is much morecompact and involves a much denser gas than in mean sequencestar-forming galaxies at intermediate redshifts. Accurate mea-surements in both IR-luminous and UV-luminous star-formingare needed to establish whether or not these two population ofgalaxies have intrinsically di ff erent physical properties in theiroutflows. The outflows predicted in cosmological simulations of high-zgalaxies are extended over regions of several kpc in size (e ff ec-tive radius of 1 to 7 kpc) but with maximal velocities of lessthan 200 km s − (Ceverino et al. 2016), that is, a factor 2-3smaller than the V out measured in BOSS-EUVLG1. However,according to most recent high angular resolution (100-200 pc)hydrodynamical simulations (TNG50; Nelson et al. 2019), out-flowing maximal velocities tend to increase with the stellar massof the galaxy, such that V max of 500-600 km s − are expectedin z ∼ M (cid:12) . Ac-cording to these simulations, the mass-loading factor would beabout 4 for a typical 10 M (cid:12) galaxy at z ∼ 2, namely, a factorof about 100 larger than the value derived for BOSS-EUVLG1( η = ± α line used in this study traces only the warm 10 K optical emit-ting gas, while it is known that some outflows carry also rele-vant amounts of mass, momentum, and energy in their molecularphase (Veilleux et al. 2020), and in the di ff use hot X-ray emit-ting gas (e.g. M82; Strickland & Heckman 2009; Lopez et al.2020; also Heckman & Thompson 2017 for a review). In fact,according to the TNG50 simulations, the ouflowing mass ratefor galaxies with masses of 3 × M (cid:12) , or less, is dominatedby the di ff use X-ray emitting gas with the peak of the outflowingmass at temperatures above 10 K (Nelson et al. 2019). The X-ray emitting mass ouflowing rate in these galaxies is predictedto be at least a factor of 10 higher than that of the warm emittinggas (Nelson et al. 2019). It could well be that the H α outflowidentified in BOSS-EUVLG1 could be tracing only the warm, lower velocity phase of a much powerful outflow with a totalmass-loading factor (i.e., all velocities and gas phases) close toone or even higher (Nelson et al. 2019) and, therefore, relevantfor the quenching or temporal delay of the star formation in thesegalaxies. ATHENA, the future European Space Agency X-ray satellite,could be able to detect the X-ray emission associated with themassive starburst at the redshift of BOSS-EUVLG1. One ofits instruments, the Wide Field Imager (WFI, Meidinger et al.2018), will be able to detect sources with a flux of 5 × − erg s − cm − in the SOFT X-ray (0.5 to 2 keV) with a signal-to-noise ratio (S / N) ratio of 10 σ in 10 hours (Athena Commu-nity O ffi ce, private communication). At the redshift of BOSS-EUVLG1, this limit corresponds to a luminosity L X − SOFT = × erg s − . Well-known nearby starbursts such as M82 andNGC3079 have extended X-ray emitting halos of several kpc(Strickland et al. 2004) with L X − SOFT of 4.1 × erg s − and8.3 × erg s − , and total integrated (i.e., halo plus disk andnucleus) L X − SOFT of 4.3 × erg s − and 3.2 × erg s − ,respectively. These galaxies are forming stars at a rate of 8-9 M (cid:12) yr − (Strickland et al. 2004), about hundred times less than inBOSS-EUVLG1. Therefore, if the e ffi ciency in the conversionof the radiative and mechanical energy of the starburst into heat-ing of the interstellar medium in BOSS-EUVLG1 is similar tothat of M82 and NGC3079, a L X − SOFT of 3-4 × erg s − forthe integrated emission would be expected, and therefore couldbe detected with ATHENA. This luminosity has to be consid-ered as a lower limit to the total X-ray emission if the ratio ofthe H α to SOFT (0.3 − H α / L X − SOFT )for the extended emission in starbursts (Colina et al. 2004) istaken into account. The L H α / L soft − X ratio covers a wide range ofvalues from about 0.25 (NGC253), to 0.5 (NGC3079, Arp220),and 6 (M82). Applying similar ratios to BOSS-EUVLG1, a to-tal soft X-ray luminosity in the 8 to 180 × erg s − wouldbe expected for the outflowing ionized gas with an H α lumi-nosity of 3.1 × erg s − . However, the imaging capabilitiesof the WFI instrument precludes any direct, spatially resolved,detection of the extended emission associated with the outflow-ing material. WFI will be providing X-ray images over a 40 x40 arcmin field-of-view with an angular resolution (FWHM)of about 5 arcsec (Meidinger et al. 2018), that is, about 40 kpcat the redshift of BOSS-EUVLG1. Our H α spectroscopic ob-servations place an upper limit of a few kpc to the size of theextended emission. In addition, X-ray emitting halos is nearbystarbursts have sizes of a few kpc (M82, NGC253, NGC3079)to about 25 kpc (Arp220). In summary, the detection with a highsignificance of the X-ray emission in starburst galaxies similar toBOSS-EUVLG1 at redshifts of 2-3 will be possible in the futurewith the ATHENA satellite, but the X-ray emitting halos associ-ated with the starburst-generated outflowing material will not bespatially resolved. 6. Summary and Outlook We report the detection of a massive and fast ionized gas outflowin BOSS-EUVLG1, the most UV luminous star-forming galaxydetected so far in the Universe. BOSS-EUVLG1 is the bright-est member of a new class of galaxies, the Extremely UV Lu-minous Galaxies (EUVLGs), with (unobscured) star formationrates above 450 M (cid:12) yr − , recently detected in the BOSS survey. Article number, page 7 of 8 & A proofs: manuscript no. 39375corr The H α line in emission has been used to characterized the ion-ized outflow of BOSS-EUVLG1. H α presents a broad compo-nent blueshifted by -139 ± 87 km s − and with a width (FWHM)of 511 ± 145 km s − . The presence of the outflow in BOSS-EUVLG1 is also supported by blueshifted UV ISM absorptionlines. The H α outflow is described by a broad-to-narrow com-ponent flux ratio (F B / F N ) of 0.34 ± out ) of 573 ± 151 km s − , a total mass, log( M out [ M (cid:12) ]) , of 7.94 ± M out ) of 44 ± 20 M (cid:12) yr − .The velocity of the outflow in BOSS-EUVLG1 follows the ex-pected value in the V out -SFR diagram predicted from the rela-tion derived for low- and intermediate-z star-forming galaxies.The fraction of the flux carried out by the outflow is within thevalue expected for its stellar mass in the F B / F N ratio to stellarmass diagram.The momentum and energy involved in the ionized gas out-flow do not require the presence of an additional energy sourceother than the starburst itself, such as an AGN. The velocity ofthe outflow (V out ) is such (573 km s − ) that only a small fraction( ≤ η = ± Athena . The properties of the ionized outflow in the classof extreme UV-luminous galaxies (EUVLGs) recently identifiedin the BOSS survey will contribute to characterizing it further,using the H α emission line as a tracer. Acknowledgements. The authors thank to the anonymous referee for usefulcomments, to Natascha Forster-Schreiber for kindly sharing the data from theKMOS-3D survey used in the plots, and to Francisco Carrera and the ATHENACommunity O ffi ce for providing us with the expected sensitivity for the fu-ture WFI / ATHENA instrument. This work is based on observations made withthe Gran Telescopio Canarias (GTC) installed in the Spanish Observatorio delRoque de los Muchachos of the Instituto de Astrofísica de Canarias, in the islandof La Palma. This work was supported by the Spanish State Research Agency(AEI) under grants ESP2015-65597-C4-4-R, ESP2017-83197, ESP2017-86852-C4-2-R, PID2019-106280GB-I00, and MDM-2017-0737 Unidad de Excelencia”María de Maeztu”- Centro de Astrobiología (CSIC-INTA). References