The Repeating Fast Radio Burst FRB 121102 as Seen on Milliarcsecond Angular Scales

4.8 MB·

by  B. Marcote

Draft version January 5, 2017
TypesetusingLATEXtwocolumnstyleinAASTeX61
THE REPEATING FAST RADIO BURST FRB 121102 AS SEEN ON MILLIARCSECOND ANGULAR SCALES
B. Marcote,1 Z. Paragi,1 J. W. T. Hessels,2,3 A. Keimpema,1 H. J. van Langevelde,1,4 Y. Huang,5,1 C. G. Bassa,2
S. Bogdanov,6 G. C. Bower,7 S. Burke-Spolaor,8,9,10 B. J. Butler,8 R. M. Campbell,1 S. Chatterjee,11
J. M. Cordes,11 P. Demorest,8 M. A. Garrett,12,4,2 T. Ghosh,13 V. M. Kaspi,14 C. J. Law,15 T. J. W. Lazio,16
M. A. McLaughlin,9,10 S. M. Ransom,17 C. J. Salter,13 P. Scholz,18 A. Seymour,13 A. Siemion,15,2,19
L. G. Spitler,20 S. P. Tendulkar,14 and R. S. Wharton11
7
1
0
1Joint Institute for VLBI ERIC, Postbus 2, 7990 AA Dwingeloo, The Netherlands
2
2ASTRON, Netherlands Institute for Radio Astronomy, Postbus 2, 7990 AA Dwingeloo, The Netherlands
n
3Anton Pannekoek Institute for Astronomy, University of Amsterdam, Science Park 904, 1098 XH Amsterdam, The Netherlands
a
J 4Sterrewacht Leiden, Leiden University, Postbus 9513, 2300 RA Leiden, The Netherlands
4 5Department of Physics and Astronomy, Carleton College, Northfield, MN 55057, USA
6Columbia Astrophysics Laboratory, Columbia University, New York, NY 10027, USA

] 7Academia Sinica Institute of Astronomy and Astrophysics, 645 N. A’ohoku Place, Hilo, HI 96720, USA
E
8National Radio Astronomy Observatory, Socorro, NM 87801, USA
H
9Department of Physics and Astronomy, West Virginia University, Morgantown, WV 26506, USA
h. 10Center for Gravitational Waves and Cosmology, West Virginia University, Chestnut Ridge Research Building, Morgantown, WV 26505,
p USA
- 11Cornell Center for Astrophysics and Planetary Science and Department of Astronomy, Cornell University, Ithaca, NY 14853, USA
o
r 12Jodrell Bank Centre for Astrophysics, University of Manchester, M13 9PL, UK
st 13Arecibo Observatory, HC3 Box 53995, Arecibo, PR 00612, USA
a 14Department of Physics and McGill Space Institute, McGill University, 3600 University St., Montreal, QC H3A 2T8, Canada
[
15Department of Astronomy and Radio Astronomy Lab, University of California, Berkeley, CA 94720, USA

1 16Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109, USA
v 17National Radio Astronomy Observatory, Charlottesville, VA 22903, USA
9 18NationalResearchCouncilofCanada,HerzbergAstronomyandAstrophysics,DominionRadioAstrophysicalObservatory,P.O.Box248,
9
Penticton, BC V2A 6J9, Canada
0
1 19Radboud University, 6525 HP Nijmegen, The Netherlands
0 20Max-Planck-Institut fu¨r Radioastronomie, Auf dem Hu¨gel 69, Bonn, D-53121, Germany
.
1
0 (Received 2016 December 10; Revised 2016 December 21; Accepted 2016 December 22)
7
Submitted to ApJ
1
:
v
ABSTRACT
i
X
The millisecond-duration radio flashes known as Fast Radio Bursts (FRBs) represent an enigmatic astrophysical
r phenomenon. Recently, the sub-arcsecond localization (∼ 100 mas precision) of FRB 121102 using the VLA has led
a
to its unambiguous association with persistent radio and optical counterparts, and to the identification of its host
galaxy. However,anevenmorepreciselocalizationisneededinordertoprobethedirectphysicalrelationshipbetween
the millisecond bursts themselves and the associated persistent emission. Here we report very-long-baseline radio
interferometricobservationsusingtheEuropeanVLBINetworkandthe305-mArecibotelescope,whichsimultaneously
detect both the bursts and the persistent radio emission at milliarcsecond angular scales and show that they are co-
located to within a projected linear separation of (cid:46) 40 pc ((cid:46) 12 mas angular separation, at 95% confidence). We
detect consistent angular broadening of the bursts and persistent radio source (∼ 2–4 mas at 1.7 GHz), which are
both similar to the expected Milky Way scattering contribution. The persistent radio source has a projected size
Correspondingauthor: J.W.T.Hessels
[email protected],[email protected]
2 Marcote et al.
constrained to be (cid:46)0.7 pc ((cid:46)0.2 mas angular extent at 5.0 GHz) and a lower limit for the brightness temperature of
T (cid:38)5×107 K. Together, these observations provide strong evidence for a direct physical link between FRB 121102
b
andthecompactpersistentradiosource. Wearguethataburstsourceassociatedwithalow-luminosityactivegalactic
nucleusorayoungneutronstarenergizingasupernovaremnantarethetwoscenariosforFRB121102thatbestmatch
the observed data.
Keywords: radio continuum: galaxies — radiation mechanisms: non-thermal — techniques: high
angular resolution — scattering
FRB 121102 as Seen on Milliarcsecond Angular Scales 3
1. INTRODUCTION nals using the EVN (Paragi 2016). This was made pos-
sible by the recently commissioned EVN Software Cor-
FastRadioBursts(FRBs)aretransientsourcesofun-
relator (SFXC; Keimpema et al. 2015) at the Joint In-
known physical origin, which are characterized by their
stitute for VLBI ERIC (JIVE; Dwingeloo, the Nether-
short(∼ms),highlydispersed,andbright(S ∼0.1–
peak
lands). HerewepresentjointAreciboandEVNobserva-
10 Jy) pulses. Thus far, 18 FRBs have been discov-
tions of FRB 121102 which simultaneously detect both
ered using single dish observations (e.g., Lorimer et al.
the persistent radio source as well as four bursts from
2007; Thornton et al. 2013; Petroff et al. 2016). Unam-
FRB121102,localizingbothtomilliarcsecondprecision.
biguousassociationswithmultiwavelengthcounterparts
In §2 we present the observations and data analysis. In
have been extremely limited by the poor localization
§3wedescribetheresultsandin§4wediscusstheprop-
that such telescopes provide (uncertainty regions of at
ertiesofthepersistentsourceanditsco-localizationwith
leastseveralsquarearcminutes). Keaneetal.(2016)re-
the source of the bursts. A discussion of the constraints
ported the first apparent localization of an FRB by as-
that these data place on the physical scenarios is also
sociating FRB 150418 with a pseudo-contemporaneous
provided. Finally, we present our conclusions in §5.
transient radio source in a galaxy at z ∼ 0.5. However,
further studies have shown that the transient source
2. OBSERVATIONS AND DATA ANALYSIS
continues to vary in brightness well after the initial
FRB 150418 burst detection, and can be explained by We have observed FRB 121102 using the EVN at
a scintillating, low-luminosity active galactic nucleus 1.7 GHz and 5 GHz central frequencies (with a maxi-
(AGN; e.g. Williams & Berger 2016; Giroletti et al. mum bandwidth of 128 MHz in both cases) in 8 observ-
2016; Bassa et al. 2016; Johnston et al. 2016), which ing sessions that span 2016 Feb 1 to Sep 21 (Table 1).
leaves limited grounds to claim an unambiguous associ- These observations included the 305-m William E. Gor-
ation with FRB 150418. don Telescope at the Arecibo Observatory (which pro-
Thusfar,FRB121102istheonlyknownFRBtohave vides raw sensitivity for high signal-to-noise burst de-
shown repeated bursts with consistent dispersion mea- tection) and the following regular EVN stations: Ef-
sure(DM)andskylocalization(Spitleretal.2014,2016; felsberg, Hartebeesthoek, Lovell Telescope or Mk2 in
Scholzetal.2016). Recently, usingfast-dumpinterfero- Jodrell Bank, Medicina, Noto, Onsala, Tianma, Torun´,
metricimagingwiththeKarlG.JanskyVeryLargeAr- Westerbork(singledish),andYebes. Oftheseantennas,
ray(VLA),FRB121102hasbeenlocalizedto∼100mil- Hartebeesthoek, Noto, Tianma, and Yebes only partici-
liarcsecond(mas)precision(Chatterjeeetal.2017). The pated in the single 5-GHz session.
preciselocalizationoftheseburstshasledtoassociations We simultaneously acquired both EVN VLBI data
with both persistent radio and optical sources, and the products (buffered baseband data and real-time corre-
identification of FRB 121102’s host galaxy (Chatterjee lations) as well as wideband, high-time-resolution data
etal.2017;Tendulkaretal.2017). EuropeanVLBINet- from Arecibo as a stand-alone telescope. The Arecibo
work (EVN) observations, confirmed by the Very Long single-dish data provide poor angular resolution (∼
BaselineArray(VLBA),haveshownthatthepersistent 3 arcmin at 1.7 GHz), but unparalleled sensitivity in
source is compact on milliarcsecond scales (Chatterjee order to search for faint millisecond bursts. By first de-
et al. 2017). Optical observations have identified a faint tecting bursts in the Arecibo single-dish data, we could
(m =25.1±0.1ABmag)andextended(0.6–0.8arcsec) then zoom-in on specific times in the multi-telescope
r(cid:48)
counterpart in Keck and Gemini data, located at a red- EVN data set where we could perform high-angular-
shiftz =0.19273±0.00008–i.e. ataluminositydistance resolution imaging of the bursts themselves.
of D ≈ 972 Mpc, and implying an angular diameter
L
distance of D ≈683 Mpc (Tendulkar et al. 2017). The 2.1. Arecibo Single Dish Data
A
centroidsofthepersistentopticalandradioemissionare
For the 1.7-GHz observations, Arecibo single-dish ob-
offsetfromeachotherby∼0.2arcsec,andtheobserved
servations used the Puerto-Rican Ultimate Pulsar Pro-
optical emission lines are dominated by star formation,
cessing Instrument (PUPPI) in combination with the
withanestimatedstarformationrateof∼0.4M yr−1
(cid:12) L-band Wide receiver, which provided ∼ 600 MHz
(Tendulkar et al. 2017). In X-rays, XMM-Newton and
of usable bandwidth between 1150–1730 MHz. The
Chandra observations provide a 5-σ upper limit in the
PUPPI data were coherently dedispersed to a DM =
0.5–10 keV band of LX < 5×1041 erg s−1 (Chatterjee 557pc cm−3,aspreviouslydonebyScholzetal.(2016).
et al. 2017).
Coherent dedispersion removes the dispersive smear-
In the past few years, significant efforts have been
ing of the burst width within each spectral channel.
made to detect and localize millisecond transient sig-
The time resolution of the data was 10.24µs, and we
4 Marcote et al.
Table 1. Properties of the persistent radio source and detected FRB 121102 bursts from the Arecibo+EVN ob-
servations. All positions are referred to the 5-GHz detection of the persistent source (RP026C epoch): α =
J2000
5h31m58.70159s, δ = 33◦8(cid:48)52.5501(cid:48)(cid:48). The observations conducted on 2016 Feb 1 (RP024A) and 2016 Sep 19
J2000
(RP026A) did not produce useful data, and are not included here (see main text). The arrival times of the bursts
are UTC topocentric at Arecibo at the top of the observing band (1690.49 MHz). All these bursts had gate widths
of2–3ms,andthequotedfluxdensitiesareaveragesoverthesetimewindows. Wenotethatthelargererroronthe
fluxdensityofBurst#2isduetothefactthattheimageisdynamic-rangelimitedbecauseoftheburst’sbrightness.
√
Thelastrowshowstheaveragepositionobtainedfromthefourburstsweightedbythedetectionstatisticξ=F/ w
(fluence divided by the square-root of the burst width).
Session Epoch ν ∆α ∆δ S ξ
ν
(YYYY-MM-DD) (GHz) (mas) (mas) (µJy) (Jy ms1/2)
RP024B 2016-02-10 1.7 1.5 ± 2 −2 ± 3 200 ± 20 —
RP024C 2016-02-11 1.7 −4 ± 2 −5 ± 3 175 ± 14 —
RP024D 2016-05-24 1.7 1 ± 3 −5 ± 4 220 ± 40 —
RP024E 2016-05-25 1.7 1 ± 3 2 ± 4 180 ± 40 —
RP026B 2016-09-20 1.7 1.9 ± 1.8 −0.4 ± 2.3 168 ± 11 —
RP026C 2016-09-21 5.0 0.0 ± 0.6 0.0 ± 0.7 123 ± 14 —
(YYYY-MM-DD hh:mm:ss.sss) (Jy)
Burst #1 2016-09-20 09:52:31.634 1.7 −14 ± 3 −1.4 ± 1.8 0.46 ± 0.02 ∼0.8
Burst #2 2016-09-20 10:02:44.716 1.7 −3.3 ± 2.5 4.3 ± 1.6 3.72 ± 0.12 ∼5
Burst #3 2016-09-20 10:03:29.590 1.7 −10 ± 5 0.8 ± 3 0.22 ± 0.03 ∼0.4
Burst #4 2016-09-20 10:50:57.695 1.7 3 ± 6 6 ± 4 0.17 ± 0.03 ∼0.2
Avg. burst pos. 2016-09-20 1.7 −5 ± 4 3.5 ± 2.2 — —
recordedfullStokesparameters. At5GHz, theArecibo expected DM of FRB 121102. This is required to
single-dish observations were recorded with the Mock separate astrophysical bursts from radio frequency in-
SpectrometersincombinationwiththeC-bandreceiver, terference (RFI). For each candidate burst found us-
which together provided spectral coverage from 4484– ing single pulse search.py (and grouping common
5554 MHz. The Mock data were recorded in 7 par- events across DM trials), the astrophysical nature was
tially overlapping subbands of 172 MHz, with 5.376- confirmedbyproducingafrequencyversustimediagram
MHz channels and 65.476 µs time resolution. In addi- to show that the signal is (relatively) broadband com-
tion to the PUPPI and Mock data, the autocorrelations pared to the narrow-band RFI signals that can some-
of the Arecibo data from the VLBI recording were also times masquerade as dispersed pulses.
available (these are restricted to only 64 MHz of band-
width, see below).
2.2. Arecibo+EVN Interferometric Data
The Arecibo single-dish data were analyzed using
tools from the PRESTO1 suite of pulsar software (Ran- EVNdatawereacquiredinrealtimeusingthee-EVN
som 2001), and searched for bursts using standard pro- setup, in which the data are transferred to the central
cedures (e.g., Scholz et al. 2016). The data were first processing center at JIVE via high-speed fibre networks
subbanded to 8× lower time and frequency resolution andcorrelatedusingtheSFXCsoftwarecorrelator. The
and were then dedispersed using prepsubband to trial high data rate of VLBI observations requires visibilities
DMs between 487–627 pccm−3 in order to search for to be typically averaged to 2-s intervals during corre-
pulses that peak in signal-to-noise ratio (S/N) at the lation, which is sufficient to study persistent compact
sources near the correlation phase center, like the per-
sistent radio counterpart to FRB 121102. However, we
1 Availableathttps://github.com/scottransom/presto also buffered the baseband EVN data to produce high-
time-resolutioncorrelationsafterwardsforspecifictimes
FRB 121102 as Seen on Milliarcsecond Angular Scales 5
when bursts have been identified in the Arecibo single- rival times. A pulse profile was then created for each
dish data. of the bursts by plotting the total power in the cross-
WeusedJ0529+3209asphasecalibratorinallsessions correlations as a function of time. We then used this
(1.1◦ away from FRB 121102). In the first five sessions pulse profile to determine the exact time window for
(conducted in Feb and May) we scheduled phase ref- whichthecorrelationfunctionwasaccumulated,i.e. the
erencing cycles of 8 min on the target and 1 min on ‘gate’. Wede-dispersedandcorrelatedtheEVNdatato
the phase calibrator. Whereas this setup maximized produce visibilities for windows covering only the times
the on-source time for burst searches, it provided less of detected bursts. We applied the previously described
accurateastrometryduetopoorerphasesolutions. The calibration to the single-pulse data and imaged them.
pulsar B0525+21 was also observed in one of these ses- The final images were produced using a Briggs robust
sionsfollowingthesamestrategy(phasereferencedusing weighting of zero (Briggs 1995) as it produced the most
J0521+2112), in order to perform an empirical analysis consistentresults(balancebetweenthelongestbaselines
ofthederivedastrometryininterferometricsingle-burst to Arecibo and the shorter, intra-European baselines).
imaging. In the following three sessions in September, Images with natural or uniform weighting did not pro-
however,weconducted5-mincycleswith3.5minonthe duce satisfactory results due to the sparse uv-coverage.
target and 1.5 min on the phase calibrator, improving The flux densities and positions for all datasets were
the phase referencing, and hence providing more accu- measured using Difmap and CASA3 by fitting a circu-
rate astrometry. Two sessions failed to produce useful lar Gaussian component to the detected source in the
calibrated data on the faint target, and are not listed uv-plane.
in Table 1. The first session (2016 Feb 1) was used
to explore different calibration approaches, whereas the
3. RESULTS
2016 Sep 19 session was unusable because the largest
3.1. Burst Detections
EVN stations were unavailable and the data could not
be properly calibrated without them. An extragalactic The EVN observations detect the compact and per-
∼2mJycompactsource(VLA2inKulkarnietal.2015), sistent source found by Chatterjee et al. (2017) with a
wasidentifiedinthesameprimarybeamasFRB121102 synthesized beam size (FWHM) of ≈21 mas×2 mas at
(with coordinates α = 5h31m53.92244s, δ = 1.7 GHz and ≈ 4 mas×1 mas at 5 GHz, with position
J2000 J2000
33◦10(cid:48)20.0739(cid:48)(cid:48)). This source has been used to acquire angle ≈−55◦ in both cases.
relative astrometry of FRB 121102 during all the ses- On 2016 Sep 20, we detected four individual bursts
sions and to provide a proper motion constraint. in the Arecibo single-dish PUPPI data that overlap
The 2-s integrated data were calibrated using stan- with EVN data acquisition (Table 1). No bursts were
dard VLBI procedures within AIPS2 and ParselTongue detected in the Arecibo PUPPI (1.7 GHz) or Mock
(Kettenis et al. 2006), including a priori amplitude cal- (5 GHz) data from other sessions in which there are
ibration using system temperatures and gain curves for simultaneous EVN observations that can be used for
eachantenna,antenna-baseddelaycorrectionandband- imaging the bursts. We formed images from the cali-
pass calibration. The phases were corrected by fringe- bratedvisibilitydataforeachburst,andmeasuredtheir
fittingthecalibrators. ThephasecalibratorJ0529+3209 positions with respect to the persistent radio source.
was then imaged and self-calibrated using the Caltech Figure 1 shows these positions together with the persis-
Difmap package (Shepherd et al. 1994). These correc- tent source at 1.7 and 5.0 GHz. The nominal positions
tions were interpolated and applied to FRB 121102, measuredforthefourburstsarespread(cid:46)15masaround
which was finally imaged in Difmap. thepositionofthepersistentsource,andwediscussthis
The arrival times of the bursts were first identified scatter in §3.2.
using Arecibo single-dish data, and then slightly re- Figures 2 and 3 show data corresponding to the
fined for application to the EVN data. First using co- strongest burst (Burst #2) – in the time-domain and
herentlydedispersedAreciboauto-correlationsfromthe in the image plane, respectively. We have characterized
√
EVN data, we performed a so-called gate search by cre- the bursts using the detection statistic ξ = F/ w (flu-
ating a large number of short integrations inside a 50- encedividedbythesquare-rootoftheburstwidth; e.g.,
ms window around the nominal Arecibo single-dish ar- Cordes & McLaughlin 2003; Trott et al. 2013). When
matched filtering is done to detect a pulse (as we have
2 TheAstronomicalImageProcessingSystem,AIPS,isasoft-
ware package produced and maintained by the National Radio 3 The Common Astronomy Software Applications, CASA, is
AstronomyObservatory(NRAO).
softwareproducedandmaintainedbytheNRAO.
6 Marcote et al.
done, starting with the single-dish PUPPI data), then
40pc
the S/N of the detection statistic, i.e. the output of 52.5700
the correlation, is proportional to ξ. Localization of the
source in an image (whether in the image or in the uv
domain)willtendtohavethesamescalingiftheuvdata
52.5600
are calculated with a tight gate (time window) around
the pulse so that it also scales as w. Using only flu-
0)
ence as a detection statistic is not appropriate because 0
20 52.5500
ahigh-fluence,verywideburstcanstillbeburiedinthe (J
δ
noise, whereas a narrower burst with equivalent fluence
is more easily discriminated from noise. Burst #2 was
52.5400
roughly an order-of-magnitude brighter than the other
3 bursts, and shows a detection statistic ξ that is also a
factor of >6 higher than the other bursts. This bright-
est burst is separated by only ∼ 7 mas from the cen- +33◦08052.5300
troid of the persistent source at the same epoch and is
58.703s 58.702s 58.701s 5h31m58.700s
positionally consistent at the ∼2-σ level. We thus find α(J2000)
no convincing evidence that there is a significant off-
Figure 1. EVN image of the persistent source at 1.7 GHz
set between the source of the bursts and the persistent
(white contours) together with the localization of the
source. Since Burst #2’s detection statistic, ξ, is signif-
strongest burst (red cross), the other three observed bursts
icantly larger than for any of the other three bursts, its (gray crosses), and the position obtained after averaging all
apparent position is least affected by noise in the image fourburstsdetectedon2016Sep20(blackcross). Contours
plane, as we explain in the following section, §3.2. As startata2-σ noiselevelof10µJyandincreasebyfactorsof
such, in principle it provides the most accurate position 21/2. Dashed contours represent negative levels. The color
scale shows the image at 5.0 GHz from 2016 Sep 21. The
of all 4 detected bursts, and the strongest constraint on
synthesized beam at 5.0 GHz is represented by the gray el-
the maximum offset between bursts and the compact,
lipse at the bottom left of the figure and for 1.7 GHz at the
persistent radio source.
bottom right. The lengths of the crosses represent the 1-
σ uncertainty in each direction. Crosses for each individual
3.2. Astrometric Accuracy burstreflectonlythestatisticalerrorsderivedfromtheirS/N
and the beam size. The size of the cross for the mean po-
The astrometric accuracy of full-track (horizon-to-
sition is determined from the spread of the individual burst
horizon observations) EVN phase-referencing is usually locations,weightedbyξ(seetext),andisconsistentwiththe
limited by systematic errors due to the poorly modeled centroid of the persistent source to within <2σ.
troposphere, ionosphere and other factors. These errors
arelessthanamilliarcsecondinidealcases(Pradeletal. dataandusingonlytheresidualdelays(delaymapping;
2006),butinpracticetheycanbeafewmilliarcseconds. Huang et al. 2017, in prep.). With this method we
Given the short duration of the bursts (a few millisec- have obtained an approximate position of α =
J2000
onds),ourinterferometricEVNdataonlycontainalim- 5h31m58.698s(+0.004), δ = 33◦8(cid:48)52.586(cid:48)(cid:48)(+0.040),
−0.006 J2000 −0.044
ited number of visibilities for each burst, which results where the quoted errors are at the 3-σ level. This
in a limited uv-coverage and thus very strong, nearly method provides additional confidence that the image-
equal-power sidelobes in the image plane (see Figure 3, plane detection of the bursts is genuine, since the posi-
bottompanel). Inthiscasewearenolongerlimitedonly tions obtained with the two methods are consistent at
bythelow-levelsystematicsdescribedabove. Theerrors the 3-σ level.
in the visibilities, either systematic or due to thermal Next, we carried out an empirical analysis of single-
noise, may lead to large and non-Gaussian uncertain- burst EVN astrometry by imaging 406 pulses recorded
ties in the position, especially for low S/N, because the from the pulsar B0525+21, which was used as a test
responsefunctionhasmanysidelobes. Itisnotstraight- source in the 2016 Feb 11 session. PSR B0525+21 has
forward to derive the astrometric errors for data with typical pulse widths of roughly 200 ms and peak flux
just a few-milliseconds integration. Therefore, we con- densities of ∼70–900 mJy. This corresponds to a range
ducted the following procedure to verify the validity of of measured detection statistics ξ ∼ 0.5–27 Jy ms1/2,
the observed positions and to estimate the errors. compared to the range ξ ∼ 0.2–5 Jy ms1/2 measured
First, we independently estimated the approximate for the 4 detected FRB 121102 bursts. Figure 4 shows
positionofthestrongestburstbyfringe-fittingtheburst
FRB 121102 as Seen on Milliarcsecond Angular Scales 7
10
8
1.68
y) 6
J
(
.17 4
S
z) 2
H1.66
G 0
(
ν
90
)
1.64 (◦ 0
θ
90
−
0.74 0.76 0.78 0.80 −1800 5 10 15 20 25 30 35
∆t(s) uv-distance(Mλ)
12 Ar-Ef Ar-Mc Ef-O8
40 3
10
20 2
y) 8 mas) 0 1 (Jy)
(J 6 δ( .17
S.17 4 ∆−20 0 S
2 40 1
− −
0 40 20 0 20 40 40 20 0 20 40
− − − −
∆α(mas) ∆α(mas)
4 2 0 2 4 4 2 0 2 4 4 2 0 2 4
− −∆t(ms) − −∆t(ms) − −∆t(ms) Figure 3. Top: Amplitudes and phases of the obtained
visibilities for the strongest burst observed on 2016 Sep 20
Figure 2. Top: Dynamic spectrum of the strongest burst (Burst#2inTable1)asafunctionoftheuv-distance. Bot-
detectedon2016Sep20(Burst#2inTable1)fromArecibo tom: Dirty (left) and cleaned (right) image for the same
autocorrelations,showingthedispersivesweepacrosstheob- burst. The cleaned image has been obtained by fitting the
serving band. Bottom: Coherently dedispersed and band- uv-data with a circular Gaussian component. The synthe-
integrated profiles of the same burst as observed in the sized beam is shown by the gray ellipse at the bottom right
cross-correlations for Arecibo-Effelsberg (Ar-Ef), Arecibo- of the figure. The coordinates are relative to the position of
Medicina(Ar-Mc),andEffelsberg-Onsala(Ef-O8)afteronly the persistent source obtained in the same epoch.
applying a priori calibration. The measured peak bright-
nesses are 11.9, 10.7, and 10.9 Jy, respectively, where the
position (separated ∼8 mas with respect to the persis-
error is typically 10−20% for a priori calibration. The rms
tent source) shows that the average burst position and
on each baseline is 12, 80, and 300 mJy, respectively.
the persistent source position are coincident within 2σ.
We therefore claim no significant positional offset be-
the obtained positions for the different PSR B0525+21
tween the persistent radio source and the source of the
pulsesalongwiththesynthesizedbeamFWHMforcom-
FRB 121102 bursts.
parison. Thisdemonstratesthatthepositionalaccuracy
Finally, we place limits on the angular separation be-
of the bursts increases for larger ξ. It shows that pulses
with ξ (cid:38) 5 Jy ms1/2 are typically offset by less than tween the source of the bursts and the persistent radio
source by sampling from Gaussian distributions with
the beam FWHM, whereas for ξ ∼ 0.5–1 Jy ms1/2 the
centers and widths given by the source positions and
scatter can be closer to 10 mas. This matches well with
uncertainties listed in Table 1 and deriving a numeri-
whatwehaveobservedinthefourdetectedFRB121102
cal distribution of offsets. Using the average burst po-
bursts (Table 1, Figure 1).
sition compared to that of the persistent source, this
While Burst #2 is thus expected to provide the most
results in a separation of (cid:46) 12 mas ((cid:46) 40 pc) at the
accurate position, a more conservative way to estimate
95% confidence level (or (cid:46) 50 pc at 99.5% confidence
the positional error on the burst source is to consider
level). Althoughthepositionaluncertaintiesonindivid-
the scatter in all four detections, which is ∼ 10 mas
ual bursts are likely underestimated and non-Gaussian,
around the average position. In Figure 1 we show the
as discussed previously, the effect of this should be mit-
averagepositionfromthefourobservedbursts,weighted
igated somewhat by using the average burst position,
by their detection statistic ξ (Table 1). This average
whichincludesanuncertaintydeterminedbythescatter
8 Marcote et al.
300
25
250
10
)
20 /12 200
s
m y)
mas) 0 15 /12(Jy− µS(J.17 150
( h 100
δ dt
∆ wi
10 e 50
c
n
e
u
Fl 0
10 5 57450 57500 57550 57600 57650
− MJD
250
10 0 10
−
∆α(mas) 200
Figure4. PulselocalizationsfromthepulsarB0525+21ob-
servedon2016Feb11at1.7GHz. Atotalof406pulseswere Jy) 150
µ
imaged. Systematic uncertainties inversely proportional to (
the detection statistic ξ are observed. We note that only S.17 100
pulses with ξ (cid:38) 5 Jy ms1/2 are robustly localized to within
the FWHM, and for pulses with ξ ∼ 0.5–1 Jy ms1/2 the 50
scatter can be closer to 10 mas.
0
intheseparateburstdetections,asalsoseeninFigure4 9:30 10:00 10:30 11:00
UT
for B0525+21. We note that nearly identical separation
limitsareobtainedifweconsiderinsteadthepositionof Figure 5. Top: Light curve of the persistent source at
only the strongest burst, Burst #2. 1.7 GHz during all the EVN epochs. The horizontal line
represents the average flux density value and its 1-σ stan-
3.3. Measured Properties dard deviation. Bottom: Light curve of the source within
the 2016 Sep 20 epoch (last epoch in the top figure). The
Fitting the uv-plane data with a circular Gaussian
vertical red lines represent the arrival times of the four de-
component shows that both the bursts and persistent tectedbursts. Wedonotdetectbrighteningofthepersistent
radio source appear to be slightly extended. We mea- source on these timescales after the bursts.
sure a source size of ∼2±1 mas at 1.7 GHz in the de-
tected bursts in the uv-plane. In the persistent source The different epochs at which the persistent radio
we measure a similar value of 2–4 mas at 1.7 GHz in source was observed allow us to obtain the light-curve
all sessions, whereas at 5.0 GHz we measure an angu- of the compact source. Figure 5 shows the flux densi-
lar size of ∼ 0.2–0.4 mas. Measurements in the image ties measured for the five sessions at 1.7 GHz, which
plane (after deconvolving the synthesized beam) result are compatible with an average flux density of S =
1.7
insimilarvalues. Themeasuredsourcesizesfortheper- 177±18µJy. Theonlysessionat5.0GHzshowsacom-
sistent source are consistent with the ones obtained in pact source with a flux density of S = 123±14 µJy.
5
Chatterjee et al. (2017). Assuming that the source exhibited a similar flux den-
Notably, Burst #2 shows two pulse components, sep- sity at 1.7 GHz compared with the day before, we in-
arated by approximately 1 ms (Figure 2). Complex fer a two-point spectral index α = −0.27±0.24, where
profile structure is commonly seen in the brightest S ∝ να, for the source. Table 1 summarizes the ob-
ν
FRB 121102 bursts observed to date, many of which tained results.
showtwoormorepulsecomponents(Spitleretal.2016,
4. DISCUSSION
Hessels et al., in prep.). Furthermore, as can be seen in
the dynamic spectrum of the burst, there is fine-scale Chatterjee et al. (2017) have shown that the persis-
frequency structure (∼ MHz) in the intensity, which tent radio source is associated with an optical coun-
in principle could be due to scintillation or self noise. terpart, which Tendulkar et al. (2017) show is a low-
Thiswillbeinvestigatedinmoredetailinaforthcoming metallicity, star forming dwarf galaxy at a redshift of
paper. z =0.19273±0.00008. Inthefollowing,weusethelumi-
FRB 121102 as Seen on Milliarcsecond Angular Scales 9
nosityandangulardiameterdistancesofD ≈972Mpc The compactness of the source at 5.0 GHz allows us
L
and D ≈ 683 Mpc, respectively, determined by Ten- to set an upper limit on its projected physical size of
A
dulkaretal.(2017). WeshowthattheVLBIdataalone (cid:46)0.7pc(1.4×105 AU,giventhedistanceofthesource).
provide further support to the extragalactic origin of The angular size measured for the source at 1.7 and
both radio sources. Furthermore, we argue that the 5.0 GHz (∼2 and ∼0.2 mas, respectively) is consistent
bursts and the persistent radio source must be phys- withtheangularbroadeningexpectedinthedirectionto
ically related because of their close proximity to each FRB 121102 for extragalactic sources due to local scat-
other. We assume such a direct physical link in the fol- tering (multi-path propagation) of the signal by the in-
lowing discussion. tervening Galactic material between the source and the
observer(predictedbytheNE2001modeltobe∼2mas
at 1.7 GHz; Cordes & Lazio 2002). Angular broaden-
ing scales as ∝ ν−2 and thus we would expect a size of
4.1. Persistent Source and Burst Properties
∼0.4 mas at 5.0 GHz, also roughly consistent with the
The results from all the EVN observations conducted
measured value at that frequency.
at 1.7 GHz show a compact source with a persistent
The obtained angular sizes are thus likely to be pro-
emission of ∼ 180 µJy, which is consistent with the
ducedbyangularbroadeningandnotbythefactthatwe
flux densities inferred at ∼ 100× larger angular scales
are resolving the source. The angular broadening mea-
with the VLA (Chatterjee et al. 2017). No significant,
sured in the bursts (∼ 2 mas) supports this statement
short-termchangesinthefluxdensityareobservedafter
as this broadening must be produced extrinsically to
the arrival of the bursts or otherwise (Figure 5). The
thebursts, giventhattheirmilliseconddurationimplies
average flux density of the persistent source implies a
thattheemittingregionmustbesmallerthanthelight-
radio luminosity of L ≈ 3×1038 erg s−1. The sin-
1.7 crossing time, i.e. (cid:46)1000 km, and thus must appear to
gle measured flux density at 5.0 GHz corresponds to a
bepoint-like. Acaveathereisthatthemeasuredsource
similar luminosity of L ≈7×1038 erg s−1 (νL , with
5.0 ν size depends strongly on the gain of the telescope pro-
a bandwidth of 128 MHz at both frequencies). Addi-
viding the longest baselines (Natarajan et al. 2016), in
tionally, the 5.0-GHz data allow us to set a constraint
ourcaseAreciboat1.7GHz. Weexcludethepossibility
on the brightness temperature of the persistent source
that the Arecibo baselines have a lower amplitude (due
of T (cid:38) 5 × 107 K. Considering the measured radio
b to a large gain error), because the measured persistent
luminosities and the current 5-σ X-ray upper limit in
source size agrees with that obtained with the VLBA
the 0.5–10 keV band of 5×10−15 erg s−1 cm−2 (Chat-
independently (Chatterjee et al. 2017). At 5.0 GHz the
terjee et al. 2017, which implies L <5×1041 erg s−1)
X presenceofthreetelescopeswithsimilarbaselinelengths
we infer a ratio between the 5.0-GHz radio and X-
(Arecibo, Hartebeesthoek and Tianma) also assures the
ray luminosities of logR > −2.4, where R =
X X consistency of the measured source size.
νL (5 GHz)/L (2−10 keV) as defined by Terashima
ν X Theintrinsicsourcesizeofthepersistentsourcecould
&Wilson(2003). Thestrongestobservedburstexhibits be as small as 7µas ν−5/4, the limit implied by syn-
a luminosity of ∼ 6×1042 erg s−1 at 1.7 GHz in the 1
chrotronself-absorptionforafrequencyofopticaldepth
2-ms integrated data. These values imply an energy
unity ν ∼ 1 GHz (Harris et al. 1970). Or if T (cid:46)
of ∼ 1040(∆Ω/4π) erg, where ∆Ω is the emission solid 1 b
1012 K by synchrotron self-Compton radiation, the size
angle.
is (cid:38) 20 µas ν−1. These angles are too small to resolve
With the EVN sessions spanning a period of ap-
withVLBIbutcouldbeprobedwithinterstellarscintil-
proximately 7 months, we derived a constraint on the
lations.
proper motion of the persistent source of −6.4 < µ <
α Wehaveconstrainedtheprojectedseparationbetween
1.4 mas yr−1, and −2.8 < µ < 6.2 mas yr−1 at a 3-σ
δ the source of the bursts and the persistent radio source
confidence level. These values have been obtained after
to be (cid:46)40 pc. Such a close proximity strongly suggests
removingtheoffsetsmeasuredinthein-beamcalibrator
that there is a direct physical link between the bursts
source(VLA2,see§2). Sincemostofourshortobserva-
and the persistent source, as we now discuss in more
tionswerenotidealforastrometry, thisisapreliminary
detail.
result, to be further improved on by follow-up observa-
tions, which will also have the advantage of spanning a
4.2. Possible Origins of FRB 121102
longerperiodoftime. Nonetheless,theseresultsalready
rule out the presence of parallax (cid:38)3 mas at a 3-σ con- Thedatapresentedhere,inadditiontotheresultspre-
fidence level, setting a distance for the persistent source sented by Chatterjee et al. (2017) and Tendulkar et al.
(cid:38)0.3 kpc. (2017), allow us to constrain the possible physical sce-
10 Marcote et al.
nariosfortheoriginofFRB121102. Whilethefactthat to be between 100 and 1000 yr old and the persistent
the bursts are located within (cid:46) 40 pc of the persistent radio source is powered by the spin down of a rapidly
radiosourcestronglysuggestsadirectphysicallink,the rotating pulsar or magnetar. In this case, the previ-
persistentradiosourceandthesourceoftheFRB121102 ouslyshownpersistentradiosourcevariability(Chatter-
burstsdon’tnecessarilyhavetobethesameobject. We jeeetal.2017,wherethehighercadenceofobservations
primarily consider two classes of models that could ex- allowed variability to be studied in more detail) could
plainFRB121102anditsmultiwavelengthcounterparts: be induced by scintillation, which is consistent with the
ahighlyenergetic,extragalacticneutronstarinayoung compact(sub-milliarcsecond)SNRsizeatthisage. Lun-
supernova remnant (SNR) or an active galactic nucleus nan et al. (2014) and Perley et al. (2016) show that su-
(AGN; or analogously a black hole related system with perluminoussupernovae(SLSNe)aretypicallyhostedby
a jet). low-metallicity,low-massgalaxies,andarepossiblypow-
eredbymillisecondmagnetars. Additionally,itisshown
4.2.1. Young neutron star and nebula thatSLSNeandlonggamma-raybursts(LGRBs)could
share similar environments. We note that these condi-
As previously shown by Spitler et al. (2016), the re-
tionsareinagreementwiththeopticalgalaxyassociated
peatability of FRB 121102 rules out an origin in a cat-
with FRB 121102 (Tendulkar et al. 2017).
aclysmic event that destroyed the progenitor source,
e.g. the collapse of a supramassive neutron star (Fal-
4.2.2. Active galactic nucleus / accreting black hole
cke & Rezzolla 2014). The repetition and energetics of
the bursts from FRB 121102 have been used to argue Models have been proposed in which the bursts are
that it comes from a young neutron star or magnetar duetostrongplasmaturbulenceexcitedbytherelativis-
(Cordes&Wasserman2016;Lyutikovetal.2016;Popov tic jet of an AGN (Romero et al. 2016) or due to syn-
& Pshirkov 2016). At birth, the rapid spin of such (po- chrotron maser activity from an AGN (Ghisellini 2016).
tentially highly magnetized) objects can power a lumi- It is also conceivable to have an extremely young and
nous nebula from the region evacuated by its SNR. energeticpulsarand/ormagnetarneartoanAGN(Pen
The measured luminosity for the persistent radio & Connor 2015; Cordes & Wasserman 2016) – either
source cannot be explained by a single SNR or a pulsar interacting or not.
wind nebula similar to those discovered thus far in our The persistent radio source is offset by ∼ 0.2 arcsec
Galaxy. A direct comparison with one of the brightest (0.7 kpc) from the apparent center of the optical emis-
SNRsknown,CasA(300yrold;Baarsetal.1977;Reed sionofthedwarfgalaxy(Tendulkaretal.2017). There-
et al. 1995), shows that we would expect an emission fore, it is not completely clear whether the radio source
which is ∼ 4 orders of magnitude fainter at the given can be associated with the galactic nucleus or not, but
distance of D ≈ 972 Mpc. In the case of the Crab an offset AGN is plausible as reported in other galaxies
L
Nebula, the expected flux density would be even fainter (Barth et al. 2008). If the persistent source is indeed
(∼0.5 nJy) if placed in the host galaxy of FRB 121102. an AGN in the right accretion state, we can infer the
Compact star forming regions, such as seen in Arp 220, massoftheblackholeassumingtheFundamentalPlane
have collections of SNRs that have a luminosity and T of black hole activity (Merloni et al. 2003; Falcke et al.
b
consistent with the persistent radio source. However, 2004; K¨ording et al. 2006; Plotkin et al. 2012; Miller-
neither the SFR of ∼240–1000 M yr−1 nor the size of Jonesetal.2012). Giventhemeasuredradioluminosity
(cid:12)
the region of 250–360 pc of Arp 220 (Anantharamaiah and the upper limit on the X-ray value, we estimate a
et al. 2000) agrees with the properties of the persistent lower limit on the mass of the putative black hole of
source associated with FRB 121102 (0.4 M yr−1 and (cid:38) 2×109 M . This value would be hard to reconcile
(cid:12) (cid:12)
(cid:46)0.7 pc). with the fact that the stellar mass of the host galaxy
Muraseetal.(2016)andPiro(2016)discusstheprop- is likely at least an order of magnitude less than that,
ertiesofayoung(<1000yr)SNRthatispoweredbythe and its optical spectrum shows no signatures of AGN
spin-down power of a neutron star or white dwarf. The activity (Tendulkar et al. 2017).
SNR expands into the surrounding medium and evacu- Alternatively, we could be witnessing a radio-loud,
ates an ionized region that can be seen as a luminous but otherwise low-luminosity AGN powered by a much
synchrotron nebula. This model is also constrained by less massive black hole that accretes at a very low rate.
the observation that radio bursts of FRB 121102 are Thispopulationispoorlyknown,butEVNobservations
not absorbed by the nebula and that its DM has not of the brightest low-luminosity AGNs (LLAGNs) in a
evolved significantly over the last few years. Consider- sample of Fundamental Plane outliers (i.e. radio-loud,
ingalltheseeffects,inthisscenarioFRB121102islikely with R ∼−2) show that some of these have extended
X