The Host Galaxy and Redshift of the Repeating Fast Radio Burst FRB 121102

0.56 MB·

by  Shriharsh P. Tendulkar

Draft version January 6, 2017
TypesetusingLATEXtwocolumnstyleinAASTeX61
THE HOST GALAXY AND REDSHIFT OF THE REPEATING FAST RADIO BURST FRB 121102
S. P. Tendulkar,1 C. G. Bassa,2 J. M. Cordes,3 G. C. Bower,4 C. J. Law,5 S. Chatterjee,3 E. A. K. Adams,2
S. Bogdanov,6 S. Burke-Spolaor,7,8,9 B. J. Butler,7 P. Demorest,7 J. W. T. Hessels,2,10 V. M. Kaspi,1
T. J. W. Lazio,11 N. Maddox,2 B. Marcote,12 M. A. McLaughlin,8,9 Z. Paragi,12 S. M. Ransom,13 P. Scholz,14
A. Seymour,15 L. G. Spitler,16 H. J. van Langevelde,12,17 and R. S. Wharton3
7
1
1Department of Physics and McGill Space Institute, McGill University, 3600 University St., Montreal, QC H3A 2T8, Canada
0
2 2ASTRON, the Netherlands Institute for Radio Astronomy, Postbus 2, NL-7990 AA Dwingeloo, The Netherlands
3Cornell Center for Astrophysics and Planetary Science and Department of Astronomy, Cornell University, Ithaca, NY 14853, USA
n
4Academia Sinica Institute of Astronomy and Astrophysics, 645 N. A’ohoku Place, Hilo, HI 96720, USA
a
J 5Department of Astronomy and Radio Astronomy Lab, University of California, Berkeley, CA 94720, USA
5 6Columbia Astrophysics Laboratory, Columbia University, New York, NY 10027, USA
7National Radio Astronomy Observatory, Socorro, NM 87801, USA

] 8Department of Physics and Astronomy, West Virginia University, Morgantown, WV 26506, USA
E
9Center for Gravitational Waves and Cosmology, West Virginia University, Chestnut Ridge Research Building, Morgantown, WV 26505
H
10Anton Pannekoek Institute for Astronomy, University of Amsterdam, Science Park 904, 1098 XH Amsterdam, The Netherlands
.
h 11Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109, USA
p 12Joint Institute for VLBI ERIC, Postbus 2, 7990 AA Dwingeloo, The Netherlands
o- 13National Radio Astronomy Observatory, Charlottesville, VA 22903, USA
r 14NationalResearchCouncilofCanada,HerzbergAstronomyandAstrophysics,DominionRadioAstrophysicalObservatory,P.O.Box248,
t
s Penticton, BC V2A 6J9, Canada
a 15Arecibo Observatory, HC3 Box 53995, Arecibo, PR 00612, USA
[
16Max-Planck-Institut fu¨r Radioastronomie, Auf dem Hu¨gel 69, Bonn, D-53121, Germany

2 17Leiden Observatory, Leiden University, PO Box 9513, 2300 RA Leiden, The Netherlands
v
0
(Received January 6, 2017; Revised January 6, 2017; Accepted January 6, 2017)
0
1 Submitted to ApJ
1
0
ABSTRACT
.
1
Thepreciselocalizationoftherepeatingfastradioburst(FRB121102)hasprovidedthefirstunambiguousassociation
0
7 (chance coincidence probability p(cid:46)3×10−4) of an FRB with an optical and persistent radio counterpart. We report
1 on optical imaging and spectroscopy of the counterpart and find that it is an extended (0(cid:48).(cid:48)6−0(cid:48).(cid:48)8) object displaying
:
v prominentBalmerand[OIII]emissionlines. Basedonthespectrumandemissionlineratios,weclassifythecounterpart
i as a low-metallicity, star-forming, m = 25.1AB mag dwarf galaxy at a redshift of z = 0.19273(8), corresponding to
X r(cid:48)
a luminosity distance of 972Mpc. From the angular size, the redshift, and luminosity, we estimate the host galaxy
ar to have a diameter (cid:46) 4kpc and a stellar mass of M∗ ∼ 4−7×107M(cid:12), assuming a mass-to-light ratio between
2 to 3M L−1. Based on the Hα flux, we estimate the star formation rate of the host to be 0.4M yr−1 and a
(cid:12) (cid:12) (cid:12)
substantial host dispersion measure depth (cid:46)324pccm−3. The net dispersion measure contribution of the host galaxy
to FRB121102 is likely to be lower than this value depending on geometrical factors. We show that the persistent
radio source at FRB121102’s location reported by Marcote et al. (2017) is offset from the galaxy’s center of light
by ∼200mas and the host galaxy does not show optical signatures for AGN activity. If FRB121102 is typical of
the wider FRB population and if future interferometric localizations preferentially find them in dwarf galaxies with
Correspondingauthor: S.P.Tendulkar,C.G.Bassa
[email protected];[email protected]
2 Tendulkar et al.
low metallicities and prominent emission lines, they would share such a preference with long gamma ray bursts and
superluminous supernovae.
Keywords: stars: neutron – stars: magnetars – galaxies: distances and redshifts – galaxies: dwarf –
galaxies: ISM
The host of FRB121102 3
1. INTRODUCTION tendedopticalcounterpartatthelocationwithachance
coincidenceprobabilityof≈3×10−4 —thefirstunam-
Fast radio bursts (FRBs) are bright (∼Jy) and short
biguous identification of multi-wavelength counterparts
(∼ms) bursts of radio emission that have dispersion
to FRBs. Independently, Marcote et al. (2017) used the
measures (DMs) in excess of the line of sight DM con-
European VLBI Network (EVN) to localize the bursts
tribution expected from the electron distribution of our
and the persistent source and showed that both are co-
Galaxy. To date 18 FRBs have been reported — most
located within ∼12 milliarcseconds.
ofthemdetectedattheParkestelescope(Lorimeretal.
Here we report the imaging and spectroscopic follow-
2007; Thornton et al. 2013; Burke-Spolaor & Bannister
up of the optical counterpart to FRB121102 using the
2014; Keane et al. 2012; Ravi et al. 2015; Petroff et al.
8-m Gemini North telescope.
2015; Keane et al. 2016; Champion et al. 2016; Ravi
et al. 2016) and one each at the Arecibo (Spitler et al.
2014) and Green Bank telescopes (Masui et al. 2015). 2. OBSERVATIONS AND DATA ANALYSIS
A plethora of source models have been proposed to
The location of FRB121102 was observed with the
explain the properties of FRBs (see e.g. Katz 2016, for
GeminiMulti-ObjectSpectrograph(GMOS)instrument
a brief review). According to the models, the excess
at the 8-m Gemini North telescope atop Mauna Kea,
DM for FRBs may be intrinsic to the source, placing it
Hawai’i. ImagingobservationswereobtainedwithSDSS
within the Galaxy; it may arise mostly from the inter- r(cid:48),i(cid:48)andz(cid:48)filterson2016October24,25,andNovember
galacticmedium,placingasourceofFRBsatcosmolog- 2, under photometric and clear conditions with 0(cid:48).(cid:48)58 to
icaldistances(z ∼0.2−1)oritmayarisefromthehost 0(cid:48).(cid:48)66 seeing. Exposure times of 250s were used in the
galaxy, placing a source of FRBs at extragalactic, but r(cid:48) filter and of 300s in the i(cid:48) and z(cid:48) filters with total
not necessarily cosmological, distances (∼100Mpc). exposuresof1250sinr(cid:48),1000sini(cid:48)and1500sinz(cid:48). The
Since the only evidence to claim an extragalactic ori-
detectors were read out with 2×2 binning, providing a
gin for FRBs has been the anomalously high DM, some pixelscaleof0(cid:48).(cid:48)146pix−1. Theimageswerecorrectedfor
models also attempted to explain the excess DM as a
abiasoffset,asmeasuredfromtheoverscanregions,flat
part of the model, thus allowing FRBs to be Galactic.
fieldedusingskyflatsandthenregisteredandco-added.
AllFRBsobservedtodatehavebeendetectedwithsin-
The images were astrometrically calibrated against
gle dish radio telescopes, for which the localization is of
theGaia DR1Catalog(GaiaCollaborationetal.2016).
orderarcminutes,insufficienttoobtainanunambiguous To limit the effects of distortion, the central 2.(cid:48)2×2.(cid:48)2
associationwithanyobject. Todate,noindependentin- subsection of the images were used. Each of the r(cid:48), i(cid:48),
formationabouttheirredshift,environment,andsource and z(cid:48) images were matched with 35 – 50 unblended
couldbeobtainedduetothelackofanaccuratelocaliza-
stars yielding an astrometric calibration with 7 – 9mas
tionof FRBs. Keaneet al.(2016) attempted toidentify
root-mean-square (rms) position residuals in each coor-
the host of FRB150418 on the basis of a fading radio
dinate after iteratively removing ∼ 4−5 outliers. The
source in the field that was localized to a z = 0.492
error in the mean astrometric position with respect to
galaxy. However, later work identified the radio source
the Gaia frame is thus ∼1−2mas.
as a variable active galactic nucleus (AGN) that may
We used the Source Extractor (Bertin & Arnouts
not be related to the source (Williams & Berger 2016;
1996)softwaretodetectandextractsourcesinthecoad-
Bassa et al. 2016; Giroletti et al. 2016; Johnston et al. ded images. The r(cid:48) and i(cid:48) images were photometri-
2017).
cally calibrated with respect to the IPHAS DR2 cat-
Repeated radio bursts were observed from the loca-
alog (Barentsen et al. 2014) using Vega-AB magnitude
tion of the Arecibo-detected FRB121102 (Spitler et al.
conversions stated therein. We measure isophotal in-
2016; Scholz et al. 2016), with the same DM as the first
tegrated magnitudes of m = 25.1±0.1AB mag and
r(cid:48)
detection, indicatingacommonsource. Asdiscussedby
m = 23.9±0.1AB mag for the optical counterpart of
i(cid:48)
Spitler et al. (2016), it is unclear whether the repeti-
FRB121102. The error value includes the photometric
tion makes FRB121102 unique among known FRBs, or
errorsandrmszero-pointscatter. Ongoingobservations
whether radio telescopes other than Arecibo lack the willprovidefullphotometriccalibrationing(cid:48),r(cid:48),i(cid:48),and
sensitivity to readily detect repeat bursts from other z(cid:48) bands and will be reported in a subsequent publica-
known FRBs.
tion.
Chatterjeeetal.(2017)usedtheKarlG.JanskyVery
SpectroscopicobservationswereobtainedwithGMOS
Large Array (VLA) to directly localize the repeated on 2016 November 9 and 10 with the 400linesmm−1
bursts from FRB121102 with 100-mas precision and re- grating (R400) in combination with a 1(cid:48)(cid:48) slit, covering
portedanunresolved,persistentradiosourceandanex- the wavelength range from 4650 to 8900˚A. A total of
4 Tendulkar et al.
Figure 1. The co-added spectrum of the host galaxy of FRB121102, the reference object, and the sky contribution (scaled
by 10% and offset by −3µJy). The spectra have been resampled to the instrumental resolution. Prominent emission lines are
labelled with their rest frame wavelengths. Black horizontal bars denote the wavelength ranges of the filters used for imaging.
Most of the wavelength coverage of the z(cid:48) band is outside the coverage of this plot.
nine1800sexposuresweretakenwith2×2binning,pro- tra of the FRB counterpart, the reference object as well
vidingaspatialscaleof0(cid:48).(cid:48)292pix−1 andaninstrumental as the sky background, we then simultaneously fit the
resolution of 4.66˚A, sampled at 1.36˚Apix−1. The con- spatial profile at the location of the counterpart and at
ditions were clear, with 0(cid:48).(cid:48)8 to 1(cid:48).(cid:48)0 seeing on the first the location of the reference object on top of a spatially
night, and 0(cid:48).(cid:48)9 to 1(cid:48).(cid:48)1 on the second. To aid the spec- varyinglinearpolynomialforeachcolumninthedisper-
tralextractionoftheveryfaintcounterpart,theslitwas sion direction.
orientedatapositionangleof18◦.6,containingthecoun- Wavelength calibrations were obtained from arc lamp
terparttoFRB121102aswellasanm =24.3 ABmag, exposures, modelling the dispersion location to wave-
r(cid:48)
m = 22.7AB mag foreground star, located 2(cid:48).(cid:48)8 to the length through 4th order polynomials, yielding rms
i(cid:48)
South (shown later in Figure 3). residualsofbetterthan0.2˚A.Theindividualwavelength
The low signal-to-noise of the spectral trace of the calibrated spectra were then combined and averaged.
FRB counterpart on the individual bias-corrected long- The instrumental response of the spectrograph was cal-
slit spectra complicated spectral extraction through the ibrated using an observation of the spectrophotometric
optimal method by Horne (1986). Instead, we used standard Hiltner600 (Hamuy et al. 1992, 1994), which
a variant of the optimal extraction method of Hynes was taken on 2016 November 7 as part of the stan-
(2002) by modelling the spectral trace of the reference dard Gemini calibration plan with identical instrumen-
object by a Moffat function (Moffat 1969) to determine talsetupasthescienceobservations. Theflux-calibrated
thepositionandwidthofthespatialprofileasafunction spectrum of the reference object gives a spectroscopic
ofwavelength. Becauseoftheproximityofthereference AB magnitude of m ≈ 22.6, about 11% higher than
i(cid:48)
objecttotheFRBcounterpart(20pix), weassumethat derived from photometry. Given that the spectrophoto-
the spatial profile as a function of wavelength is iden- metric standard was observed on a different night with
tical for both. We note that though the counterpart is worse seeing (1(cid:48).(cid:48)4), we attribute this difference to slit
slightly resolved in the imaging observations, the worse losses and scale the flux of the observed spectra of the
seeing during the spectroscopic observations (by a fac- reference object and the FRB counterpart by a factor
tor 1.2 to 1.9) means the seeing dominates the spatial 0.89.
profile. The residual images validate this assumption;
3. RESULTS AND ANALYSIS
no residual flux is seen once the extracted model is sub-
tracted from the image. To optimally extract the spec- The final combined and calibrated spectrum is shown
in Figure1. Besides continuum emission, which is
The host of FRB121102 5
Table 1. Emission line properties.
Line Obs. Flux Width(σ) Aλ/AV
(ergcm−2s−1) (˚A) (mag)
Hβ 0.118(11)×10−16 1.91(19) 0.941
[OIII]λ4959 0.171(10)×10−16 1.75(11) 0.921
[OIII]λ5007 0.575(11)×10−16 1.89(4) 0.911
[OI]λ6300 <0.009×10−16 0.670
[NII]λ6549 <0.021×10−16 0.625
Hα 0.652(9)×10−16 2.02(3) 0.622
[NII]λ6583 <0.030×10−16 0.619
[SII]λ6717 0.040(6)×10−16 2.4(4) 0.596
[SII]λ6731 0.024(6)×10−16 2.4(6) 0.593
Note—Observedemissionlinepropertiesfromfitting
normalizedGaussianstotherest-wavelengthhostgalaxy
spectrum. Upperlimits(3-σ)onlinefluxesassume
Gaussianwidthsofσ=2˚A.TheabsorptionAλ/AV at
theobservedlinewavelengthsistakenfromCardellietal.
(1989). Toobtainunabsorbedlinefluxes,multiplyby
100.4(Aλ/AV)AV,whereAV istheGalacticabsorption
towardsFRB121102.
weakly detected in the red part of the spectrum, four
strong emission lines are clearly visible and are iden-
tified as Hα, Hβ, and [O III] λ4959 and [O III] λ5007
indicatingthattheopticalcounterpartisastar-forming
galaxy. The corresponding weighted mean redshift is
z =0.19273±0.0008. Weaker emission lines from [SII]
λλ6717,6731arealsodetected. The[NII]λλ6549,6583
and the [OI] λ6300 lines are not seen.
Gaussian fits to the emission lines in the rest frame
yieldthefluxand1-σ widthvalueslistedinTable1. We Figure 2. BPT(Baldwinetal.1981)diagramsof[NII]/Hα
estimate rest frame equivalent widths for the strongest and [SII]/Hα against [OIII]/Hβ for the SDSS DR12 (Alam
emission lines; 392±102˚A for [O III] λ5007 and 290± et al. 2015) galaxy sample with significant (>5-σ) emission
55˚A for Hα. lines. The black symbol with error bars denotes the lo-
cation of the host galaxy of FRB121102. The solid and
The ratios of measured line fluxes for [O III]/Hβ
dashed lines denote the demarcations between star-forming
against[NII]/Hαand[SII]/Hα—thewell-knownBald- and AGN dominated galaxies, respectively (Kewley et al.
win,Phillips&Terlevich(BPT)diagram(Baldwinetal. 2001,2006;Kauffmannetal.2003). Theregionbetweenthe
1981) — are shown in Figure 2. The line ratios of the twocurvescorrespondstocompositeobjectswithAGNand
host galaxy of FRB121102 are compared to those from star formation.
the SDSS DR12 galaxy sample (Alam et al. 2015). The
locations below and to the left of the solid and dashed with the point-spread-function, was fitted against the
greylinesindicatethattheemissionlinesareduetostar spatialprofileofthecounterpart. Forthei(cid:48)-bandimage,
formation and not due to AGN activity (Kewley et al. the best fit has an effective radius of Re =0(cid:48).(cid:48)41±0(cid:48).(cid:48)06,
2001,2006;Kauffmannetal.2003). NotethattheBPT a S´ersic index of n = 2.2 ± 1.5, and an ellipticity of
diagram line ratios are insensitive to reddening (from b/a=0.25±0.13. Thelowersignal-to-noiseofthecoun-
the Milky Way as well as the host itself). terpartinther(cid:48)andz(cid:48)imagesdidnotpermitmeaningful
We use the galfit software (Peng et al. 2002, 2010) results. Instead, we directly fit the spatial profile in all
to constrain the morphology of the optical counterpart. three bands with a two-dimensional elliptical Gaussian
A S´ersic profile (Σ(r) = Σee−κ[(r/Re)1/n−1]), convolved profile. Inthecaseofthei(cid:48)-bandimage,thefitprovides
6 Tendulkar et al.
a position and effective radius, taken as the Gaussian
σ, consistent with the S´ersic profile convolved with the
point-spread-function. The results of the fits are shown
in Figure3.
The position and extent of the host galaxy, as ap-
proximatedwiththetwo-dimensionalellipticalGaussian
profile, agrees well in the r(cid:48) and i(cid:48) bands (semi-major
axis σ = 0(cid:48).(cid:48)44 with ellipticity b/a = 0.68), while the
a
z(cid:48)-band has a slightly offset position and appears larger
(σ =0(cid:48).(cid:48)59withb/a=0.45). Weattributethisdifference
to the fact that the the r(cid:48) and i(cid:48) bands are dominated
by the bright emission lines of Hα, Hβ, [O III] λ4959
and [O III] λ5007, while the redder z(cid:48)-band traces the
continuum flux of the host galaxy. As such, the mor-
phology suggests that the host galaxy has at least one
HII region at a slight offset from the galaxy center.
Finally, the bottom right panel of Figure3 plots the
Gaussian centroids on the International Celestial Refer-
ence Frame (ICRF) through the astrometric calibration
of the r(cid:48), i(cid:48), and z(cid:48) images against Gaia. The posi-
Figure 3. The top left, top right and bottom left pan-
tional uncertainties in each axis are the quadratic sum
els show respective 7(cid:48).(cid:48)4×7(cid:48).(cid:48)4 subsections of the GMOS r(cid:48),
of the astrometric tie against Gaia (of order 2mas) and
i(cid:48) and z(cid:48) images, centered on the optical counterpart to
the centroid uncertainty on the image (between 20 and
FRB121102. Each image has been smoothed by a Gaussian
50mas). The Gaia frame is tied to the ICRF defined with a width of 0(cid:48).(cid:48)2, while the plus sign and ellipse denote
via radio VLBI to a ∼1mas precision (Mignard et al. the position and extent of a two-dimensional Gaussian fit
2016), much smaller than the centroid uncertainty. We to the spatial profile of the counterpart. The i(cid:48)-band image
findthatthepositionofthepersistentradiosourceseen also shows the narrower S´ersic fit by galfit. The bottom
with the EVN at an observing frequency of 5GHz with right panel combines the positional and morphological mea-
surements from the different bands on an astrometric frame
a 1-mas precision (Marcote et al. 2017), is offset from
of 1(cid:48)(cid:48)×1(cid:48)(cid:48) in size. The colors are identical to those used in
thegalaxycentroidsby186±68and163±32masinthe
the other panels. The large ellipses denote the extent of the
line-dominatedr(cid:48) andi(cid:48) images,and286±64masinthe
Gaussian and S´ersic fits, while the small ellipses denote the
continuum-dominatedz(cid:48) image. Thoughoffsetfromthe 1-σ absolute positional uncertainties. The location of the
centroids, the persistent radio source is located within persistent counterpart as measured with the EVN at 5GHz
the effective radii of the different bands. by Marcote et al. (2017) is represented by the black cross.
The uncertainty in the EVN location is much smaller than
the size of the symbol.)
4. DISCUSSION AND CONCLUSIONS
The observations presented here confirm the interpre-
WeusetheSchlegeletal.(1998)estimateoftheGalac-
tation by Chatterjee et al. (2017) that the extended tic extinction along this line of sight1, E = 0.781.
B−V
optical counterpart associated with FRB121102 is the
UsingR =3.1,wefindA =2.42,andusetheCardelli
V V
host galaxy of the FRB. Our measurement of the red-
et al. (1989) Galactic extinction curve to correct the
shift z = 0.19273 is consistent with the DM-estimated
spectrum with band extinctions of A = 2.15,A =
r(cid:48) i(cid:48)
value of z < 0.32 (Chatterjee et al. 2017) and to-
DM 1.63, and Az(cid:48) = 1.16mag. We note that the Schlafly
gether with the very low chance superposition probabil-
et al. (2010); Schlafly & Finkbeiner (2011) recalibrated
ity,firmlyplacesFRB121102atacosmologicaldistance,
extinction model predicts a slightly lower extinction of
ruling out all Galactic models for this source.
E = 0.673. The results described below are insen-
B−V
In the following discussion, we assume the cosmolog-
sitive to differences in the extinction at this level. We
ical parameters from the Planck Collaboration et al.
donotapplyk-correctiontothemagnitudesastheyare
(2016)asimplementedinastropy.cosmology(Astropy not needed for the precision discussed here.
Collaboration et al. 2013), giving a luminosity distance
of D = 972Mpc, and 1(cid:48)(cid:48) corresponding to projected
L
properandcomovingdistancesof3.31kpcand3.94kpc, 1 From the IRSA Dust Extinction Calculator http://irsa.
respectively. ipac.caltech.edu/applications/DUST/
The host of FRB121102 7
4.1. Burst Energetics y=log ([NII]λ6584/[SII]λλ6717,6731)
10
TheredshiftmeasurementallowsustoputFRB121102’s +0.264log ([NII]λ6584/Hα).
10
energetics on a firmer footing, confirming the energy
scale of 1038erg(δΩ/4π)(A /0.1Jyms)(∆ν/1GHz) cal- Asthe[OII]λ3727lineisoutsideourspectralcoverage
ν
culated by Chatterjee et al. (2017) using a distance and[NII]λ6584isnotdetected,wecanonlysetanupper
limit to the metallicity. Using the extinction-corrected
scale of 1Gpc. Here A and ∆ν are the fluence and
ν
line fluxes, we measure,
bandwidth, respectively, at observing frequency ν and
δΩ is the opening angle of the bursts. A more detailed
R ≥0.77,
analysis of energetics of individual bursts detected by 23
the VLA and their rates will be reported in Law C. J. N2≤−1.34,
et al. (in preparation). O3N2≥2.1,
y≤−0.66,
4.2. Physical Properties of the Host
ThehostofFRB121102isasmallgalaxywithadiam- where the limits are calculated from the 3-σ limit on
eter of (cid:46)4kpc, inferred from the continuum-dominated [NII]λ6584fluxandassumingthelowerlimitfortheun-
z(cid:48)-band image. The absolute magnitudes, including measured [OII]λ3727 flux to be zero. This corresponds
the emission line fluxes and after correcting for the to a 3-σ metallicity limit of log ([O/H]) + 12 < 8.4
10
Milky Way’s extinction, are M = −17.0AB mag and (Kewley & Dopita 2002), < 8.4 (Pettini & Pagel 2004,
r(cid:48)
M = −17.7AB mag, identifying the host as a dwarf , N2), < 8.4 (Pettini & Pagel 2004, O3N2)2 and < 8.1
i(cid:48)
galaxy. (Dopita et al. 2016, not including scatter). We con-
From Table 1, the Hα luminosity of the host galaxy, vert these into the oft-used KK04 scale (Kobulnicky
corrected for Milky Way extinction, is L = 2.9 × & Kewley 2004) using the conversions of Kewley &
Hα
1040ergs−1. The corresponding star formation rate is Ellison (2008). All measurements are consistent with
SFR(Hα) = 7.9 × 10−42M yr−1 × (L /ergs−1) = log ([O/H])+12(cid:46)8.7 in the KK04 scale. The metal-
(cid:12) Hα 10
0.23M yr−1 (Kennicutt et al. 1994). This value does licityofthehostislow—lessthan∼15%ofallgalaxies
(cid:12)
not completely account for the extinction of Hα pho- brighter than M < −16 have metallicity lower than
B
tons in the host galaxy. The correction suggested by 8.7 (Graham & Fruchter 2015). This set of galaxies ac-
Kewley et al. (2002) is SFR(IR)=2.7×SFR(Hα)1.3 ≈ countforlessthan20%ofthestarformationofthelocal
0.4M yr−1 (inthe8–1000µmband). Thisisconsistent Universe.
(cid:12)
withthe3-σ upperlimitof<9M yr−1 estimatedfrom The host properties are similar to those of extreme
(cid:12)
theALMAnon-detectionofthehostat230GHzassum- emissionlinegalaxies(EELGs; Ateketal.2011),young,
ing a submillimeter spectral index α = 3 (Chatterjee low-mass starbursts which have emission lines of rest-
et al. 2017). frame equivalent widths greater than 200˚A.
The mass-to-light ratio Υ is dependent on the star
∗
formation history and the initial mass function for star 4.3. Ionized Gas Properties in the Host
formation. As an estimate, we use ΥR∗ ≈2−3M(cid:12)L−(cid:12)1 The Balmer lines from the host also allow us to esti-
based on the dynamics of dwarf galaxies with high star matethepropertiesitsionizedISManditscontribution
formation rates (Lelli et al. 2014), implying a stellar to the total DM of FRB121102.
mass M∗ ∼ 4 − 7 × 107 M(cid:12). As dwarf galaxies are The Hα surface density for the galaxy with flux FHα,
usuallygas-rich(e.g. Papastergisetal.2012),weexpect semi-major axis a, and semi-minor axis b is
that this estimate is a lower limit to the host baryonic
F
mass. We also note that dwarf galaxies are typically S(Hα)= Hα,
dark matter dominated (Cˆot´e et al. 2000), and so the πab
total dynamical mass is likely to be larger. ≈6.8×10−16ergcm−2s−1arcsecond−2,
We use the R23 (Kewley & Dopita 2002), N2, O3N2 ≈120Rayleigh, (1)
(Pettini&Pagel2004),andtherecentlydefineddiagnos-
ticofDopitaetal.(2016,labelledhereasy)toestimate wherewehaveusedtheextinctioncorrectedfluxFHα =
the metallicity where, 2.6×10−16ergcm−2s−1 and the semi-major and minor
R =log (([OII]λ3727+[OIII]λλ4959,5007)/Hβ),
23 10
2 WenotethatthePettini&Pagel(2004)calibrationhashigh
N2=log10([NII]λ6584/Hα), scatterforO3N2(cid:38)2butthelimitquotedhereincludesthescat-
O3N2=log ([OIII]λ5007)/[NII]λ6584×Hα/Hβ),and ter.
10
8 Tendulkar et al.
axes (a = 0(cid:48).(cid:48)44, b/a = 0.68) from the i(cid:48) and r(cid:48) images. (cid:18) EM (cid:19)1/2
× , (5)
In the source frame (denoted below by the subscript, 600 pc cm−6
‘s’), the surface density is
where (cid:15) ≤ 1 is the fractional variation inside discrete
S(Hα) =(1+z)4S(Hα)=243Rayleigh. (2) clouds due to turbulent-like density variations and ζ ≥
s
1 defines cloud-to-cloud density variations in the ion-
For a temperature T =104T K, we express the emis-
4 ized region of depth L in kpc. Here we have used
sion measure (EM =(cid:82) n2eds) given by Reynolds (1977) EM =600 pc cm−6 ankdpcassumed 100% cloud-to-cloud
s
in the galaxy’s frame
variations(ζ =2)andfullymodulatedelectrondensities
(cid:20) S(Hα) (cid:21) inside clouds ((cid:15)=1).
EM(Hα) =2.75pccm−6T0.9 s ,
s 4 Rayleigh The host contribution to the measured DM is a factor
(1+z)−1 smaller than the source frame DM3. Also, the
≈670pccm−6T0.9. (3)
4 line of sight to the FRB source may sample only a frac-
We get a smaller value from the extinction-corrected tion of D(cid:100)Ms depending on if it is embedded in or offset
Hβ flux, EM(Hβ)s ≈ 530pccm−6. For the calcu- from the Hα-emitting gas. For an effective path length
lations below, we proceed with a combined estimate, through the ionized gas L ≤L, we then have
FRB
EM ≈600pccm−6.
s (cid:18) (cid:19)
This value is fairly large compared to measurements D(cid:100)M(FRB)= D(cid:100)Ms LFRB
of the local Galactic disk. The WHAM Hα survey, for 1+z L
example, givesvaluesoftensofpccm−6 intheGalactic (cid:18)L (cid:19)(cid:20) 4L f (cid:21)1/2
plane and about 1 pc cm−6 looking out of the plane ≈324 pc cm−3 FRB kpc f . (6)
L ζ(1+(cid:15)2)
(Hill et al. 2008). However, lines of sight to distant
pulsars and studies of other galaxies give EM values in This estimate can be compared with empirical con-
the hundreds (Reynolds 1977; Haffner et al. 2009). straints discussed in Chatterjee et al. (2017) on contri-
The estimate for EM is sensitive to the inferred solid butions from the host and the intergalactic medium
s
angle of the galaxy and emitting regions. Ongoing ob- (IGM) to the total DM made by subtracting the
servations with the Hubble Space Telescope will better NE2001 model’s DM contribution from the Milky
resolvetheHαemittingstructuresandimproveourcon- Way (Cordes & Lazio 2002) (DMMW = 188 pc cm−3)
straint on the EM with respect to the location of the and the Milky Way halo (DMMWhalo = 30 pc cm−3)
burst. from the total DM = 558pccm−3. This gives
The implied optical depth for free-free absorption at DMIGM + DMhost = 340pccm−3. The Milky Way
an observation frequency ν (in GHz) is contributions have uncertain errors but are likely of or-
der 20%. The measured redshift implies a mean IGM
τff≈3.3×10−6[(1+z)νGHz]−2.1T4−1.35EMs contribution DMIGM ≈ 200 pc cm−3 (Ioka 2003; In-
≈1.4×10−3ν−2.1T−0.45. (4) oue 2004) but can vary by about ±85 pc cm−3 (Mc-
GHz 4
Quinn 2014). This yields a range of possible values
Free-freeabsorptionforFRB121102isthereforenegligi-
for DM : 55 (cid:46) DM (cid:46) 225pccm−3 that further
ble even at 100MHz. This suggests that the radio spec- host host
implies 0.09 (cid:46) (L /L)(cid:2)L f /ζ(1+(cid:15)2)(cid:3)1/2 (cid:46) 0.35.
tra of the bursts and possibly the persistent source are FRB kpc f
The ionized region therefore must have some degree of
unaffected by absorption and are inherent to the emis-
clumpiness or the effective path length is significantly
sion process or to propagation effects near the sources,
smaller than the size of the ionized region.
confirming the inference made by (Scholz et al. 2016)
Radio pulsars in the Large and Small Magellanic
basedonthewidelyvaryingspectralshapesofthebursts
Clouds have DMs spanning the range 45–273pccm−3
alone.
and 70–200 pccm−3, respectively (Manchester et al.
4.3.1. Implied DM from Hα-emitting Gas 2005). This empirically demonstrates that the free elec-
The EM implies a DM value sometimes given by tron content of star-forming dwarf galaxies is of the or-
DM=(EMf L)1/2, where f is the volume filling factor der we estimate. The relatively large DM contribution
f f
of ionized clouds in a region of total size L (Reynolds fromthehostgalaxy(asinferredfromtheHαemission)
1977). As summarized in Appendix B of Cordes et al. implies that any contributions from the vicinity of the
(2016), additionalfluctuationsdecreasetheDMderived
from EM, giving a source-frame value,
3Thefactorof(1+z)−1isacombinationofthephotonredshift,
(cid:20) f (cid:21)1/2 timedilationandthefrequency−2dependenceofcoldplasmadis-
D(cid:100)Ms≈387 pc cm−3L1k/p2c ζ(1+f(cid:15)2)/4 persion.
The host of FRB121102 9
FRB source itself are probably quite small. This may lines that we observe in the host of FRB121102 has no
ruleoutaveryyoung(<100yr)supernovaremnant(e.g. analogue in any known galaxy to the best of our knowl-
Piro 2016). edge.
The high star formation rate is consistent with the
presence of a young SNR or a cluster of young massive
4.4. Implications for Source Models
stars (i.e. an OB association), which would naturally
Chatterjee et al. (2017) reported the locations of the
link FRBs to neutron stars which are the favored pro-
radiobursts,theopticalandvariableradiocounterparts
genitor models.
and the absence of millimeter-wave and X-ray emission.
Marcoteetal.(2017)haveshownthattheburstsandthe
4.4.1. Relation to Dwarf Galaxies
persistent radio source are colocated to within a linear
It is interesting to note that the only FRB host di-
projected separation of 40pc, suggesting that the two
rectly identified so far is a low metallicity dwarf galaxy
emission sources should be physically related, though
rather than, say, an extremely high-star-formation-rate
not necessarily the same source. The radio sourceprop-
galaxy such as Arp220 or a galaxy with a very power-
erties are consistent with a low luminosity AGN or a
ful AGN or some other extreme characteristics. Dwarf
young(<1000yr)supernovaremnant(SNR)poweredby
galaxies are also a small fraction of the stellar mass
an energetic neutron star (e.g. Murase et al. 2016).
in the Universe (Papastergis et al. 2012). Ravi et al.
Theopticalpropertiesofthegalaxiesreportedheredo
(2016) also suggested that the extremely low scatter-
not add support to the AGN interpretation although it
ing of FRB150807 compared to its DM may be linked
cannot be conclusively ruled out. The BPT diagnostics
to its origin from a low-mass (<109M ) galaxy. How-
forthehost(Figure2)shownoindicationofAGNactiv- (cid:12)
ever,thestrongpolarizationandscatteringpropertiesof
ity. However,thismaynotbeconclusiveasthemajority
FRB110523 do suggest the presence of turbulent mag-
of radio-loud AGN show no optical signatures of activ-
netized plasma around the source (Masui et al. 2015),
ity (Mauch & Sadler 2007). This is further supported
suggesting that individual FRB environments may be
by five low luminosity AGN with no optical signatures
quite diverse.
have also recently been discovered (Park et al. 2016).
If FRBs are indeed more commonly hosted by dwarf
However, these objects are almost exclusively hosted in
galaxies with much larger stellar masses (∼ 1010M ). galaxies in the low redshift Universe, they would share
(cid:12)
this preference with two other classes of high-energy
Wealsonotethattheradiosourceisoffsetfromtheop-
transients — long duration gamma-ray bursts and su-
tical center of the galaxy by 170–300mas, correspond-
perluminoussupernovae,bothofwhichpreferlow-mass,
ing to a transverse linear distance of 0.5–1kpc, nearly
low-metallicity, andhighstarformationratehosts(e.g.,
a quarter to half of the radial extent, which is not con-
Fruchter et al. 2006; Perley et al. 2013; Vergani et al.
sistent with a central AGN, but such offsets have been
2015; Perley et al. 2016, and other works). Indeed,
seen before in dwarf galaxies, e.g. Henize 2-10 (Reines
superluminous supernovae are prefentially hosted by
et al. 2011).
EELGs (Leloudas et al. 2015). If this relation is true, it
Theassociationofanoptical/X-rayAGNwithadwarf
may point to a link between FRBs and extremely mas-
galaxy is also extremely rare. A search of emission-line
dwarf galaxies (108.5 (cid:46)M (cid:46)109.5M ) using BPT line sive progenitor stars, possibly extending to magnetars
∗ (cid:12)
thathavebeenassociatedwithmassiveprogenitorstars
diagnostics identified an AGN rate of ∼ 0.5 % (Reines
(e.g. Olausen & Kaspi 2014).
et al. 2013), with an additional 0.05 % of dwarf galax-
ies searched exhibiting narrow emission lines consistent
4.5. Future Optical Follow-Up of FRBs
with star formation band broad Hα consistent with an
AGN.Similarly,anX-raysurveyofz <1dwarfgalaxies A link between FRBs and dwarf galaxies will impact
reportedanAGNrateof0.6–3%(Pardoetal.2016). Of future multi-wavelength follow-up plans. Without the
thedwarfgalaxiesknowntohostAGN,onlytwoexhibit precise localization for FRB121102 (Chatterjee et al.
nuclear radio emission that appears to originate from a 2017), the host galaxy is scarcely distinguishable from
black-hole jet, Henize 2-10 and Mrk 709 (Reines et al. other objects in the deep Gemini images.
2011, 2014). Both have strong nuclear X-ray emission Due to the trade-off between field of view and local-
that originatesfrom the AGN but opticalemissionlines ization precision, FRB search projects that have a large
that are dominated by star-formation processes. The FRBdetectionratesuchasCHIME(KaspiV.M.etal,.
combination of a compact radio source, absent nuclear 2017, in preparation), UTMOST (Caleb et al. 2016),
X-ray emission, strong star-formation optical emission and HIRAX (Newburgh et al. 2016) will localize high
lines, and weak or non-existent broad optical emission signal to noise detections to only sub-arcmin precision.
10 Tendulkar et al.
IfFRBhostsarestar-forminggalaxieswithstrongemis- S.P.T acknowledges support from a McGill Astro-
sionlines, slitlessobjectiveprismspectroscopycouldef- physics postdoctoral fellowship. The research lead-
ficiently distinguish these objects from a field of stars ing to these results has received funding from the Eu-
and elliptical galaxies, leading to putative host identifi- ropean Research Council (ERC) under the European
cations without very precise localization. However, this Union’s Seventh Framework Programme (FP7/2007-
strongly depends on the link between FRBs and their 2013). C.G.B. and J.W.T.H. gratefully acknowledge
host properties and the homogeneity of FRBs — which fundingforthisworkfromERCStartingGrantDRAG-
willfirsthavetobeconfirmedwithmoreinterferometric NET under contract number 337062. J.M.C., R.S.W.,
localizations. and S.C. acknowledge prior support from the National
We note, of course, that our above discussion regard- Science Foundation through grants AST-1104617 and
ing the possible relationship between FRBs and dwarf AST-1008213. Thisworkwaspartiallysupportedbythe
galaxies in general is based on a single data point of a University of California Lab Fees program under award
repeating FRB, which may not be representative of the numberLF-12-237863. Theresearchleadingtothesere-
broader FRB population (see Spitler et al. 2016; Scholz sults has received funding from the European Research
et al. 2016, for more details). Council (ERC) under the European Unions Seventh
Framework Programme (FP7/2007-2013). J.W.T.H. is
an NWO Vidi Fellow. V.M.K. holds the Lorne Trottier
We are very grateful to the staff of the Gemini Ob- andaCanadaResearchChairandreceivessupportfrom
servatory for their help and flexibility throughout this an NSERC Discovery Grant and Accelerator Supple-
program. We also thank R. F. Trainor and A. Delahaye ment,fromaR.HowardWebsterFoundationFellowship
for helpful discussions. fromtheCanadianInstituteforAdvancedResearch(CI-
Our work is based on observations obtained at the FAR),andfromtheFRQNTCentredeRechercheenAs-
GeminiObservatory(programGN-2016B-DD-2), which trophysique du Quebec. B.M. acknowledges support by
is operated by the Association of Universities for Re- the Spanish Ministerio de Econom´ıa y Competitividad
search in Astronomy, Inc., under a cooperative agree- (MINECO/FEDER,UE)undergrantsAYA2013-47447-
ment with the NSF on behalf of the Gemini partner- C3-1-P, AYA2016-76012-C3-1-P, and MDM-2014-0369
ship: the National Science Foundation (United States), of ICCUB (Unidad de Excelencia ‘Mar´ıa de Maeztu’).
the National Research Council (Canada), CONICYT L.G.S.gratefullyacknowledgefinancialsupportfromthe
(Chile), Ministerio de Ciencia, Tecnolog´ıa e Innovaci´on ERC Starting Grant BEACON under contract number
Productiva(Argentina),andMinist´eriodaCiˆencia,Tec- 279702 and the Max Planck Society. Part of this re-
nologia e Inova¸c˜ao (Brazil). searchwascarriedoutattheJetPropulsionLaboratory,
This work has made use of data from the Euro- CaliforniaInstituteofTechnology,underacontractwith
pean Space Agency (ESA) mission Gaia (http://www. the National Aeronautics and Space Administration.
cosmos.esa.int/gaia), processed by the Gaia Data E.A.K.A. is supported by TOP1EW.14.105, which is fi-
Processing and Analysis Consortium (DPAC, http:// nanced by the Netherlands Organisation for Scientific
www.cosmos.esa.int/web/gaia/dpac/consortium). Research (NWO). M.A.M. is supported by NSF award
FundingfortheDPAChasbeenprovidedbynationalin- #1458952. S.B.SisaJanskyFellowoftheNationalRa-
stitutions, in particular the institutions participating in dio Astronomy Observatory. P.S. is a Covington Fellow
the Gaia Multilateral Agreement. This research made at the Dominion Radio Astrophysical Observatory.
use of Astropy, a community-developed core Python
Facility: Gemini:Gillett (GMOS)
package for Astronomy (Astropy Collaboration, 2013,
http://www.astropy.org). Software: ESO-MIDAS, astro-py, galfit, SExtractor
REFERENCES
Alam, S., Albareti, F. D., Allende Prieto, C., et al. 2015, Barentsen, G., Farnhill, H. J., Drew, J. E., et al. 2014,
ApJS, 219, 12 MNRAS, 444, 3230
Astropy Collaboration, Robitaille, T. P., Tollerud, E. J.,
Bassa, C. G., Beswick, R., Tingay, S. J., et al. 2016,
et al. 2013, A&A, 558, A33
MNRAS, 463, L36
Atek, H., Siana, B., Scarlata, C., et al. 2011, ApJ, 743, 121
Bertin, E., & Arnouts, S. 1996, A&AS, 117, 393
Baldwin, J. A., Phillips, M. M., & Terlevich, R. 1981,
PASP, 93, 5 Burke-Spolaor, S., & Bannister, K. W. 2014, ApJ, 792, 19