Variable Very Low Frequency Emission from the Gamma-Ray Binary LS I

Variable Very Low Frequency Emission from the Gamma-Ray Binary LS I +61 33 Benito Marcote In collaboration with M Ribó, J M Paredes (UB), C H Ishwara-Chandra (GMRT), J Swinbank, J Broderick, S Markoff, R Fender, R Wijers (LOFAR) July 8, 214

Outline 1 Introduction Radio emission from gamma-ray binaries The gamma-ray binary LS I +61 33 Goals of the project 2 Radio observations (GMRT & LOFAR) 3 Results 4 Conclusions 2 / Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 2/27 27

Gamma-Ray Binaries Binary systems which host What a are compact -ray binaries? object orbiting a high mass star that have the non-thermal maximum of the Spectral Energy Distribution in γ-rays (Paredes et al 213, Dubus 213) System Main star P/ days Cygnus X-3 WR 2 Cygnus X-1?? O97Iab 56 MWC 656?? Be 64 PSR B1259 63 95Ve 12367 HESS J632+57 B Vpe 315 LS I +61 33 B Ve 265 1FGL J1186 5856 O6V 166 LS 539 O65V 39 Green: X-ray binaries with gamma-ray emission Red: known gamma-ray binaries Binary systems Are systems that comprise a compact object orbiting a companion the non-thermal maximum of the Spectral Energy Distribution (S Figure 3: SED of the X Paredes-Fortuny (UB) Introduction -ray binary LS I +61 33 FromMoldón( SED of the gamma-ray binary LS I +61 33 Sidoli et al (26) 3 / Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 3/27 27

Gamma-Ray Binaries Young non-accreting pulsar scenario Strong shock between both winds: Relativistic pair plasma wind from the pulsar Stellar wind from the massive companion star Originally proposed by Maraschi & Treves (1981), re-proposed by Dubus (26) Radio flux dominated by the synchrotron emission Orbital motion Wind standoff Star Pulsar V Zabalza et al: GeV-TeV em Coriolis turnover Shocked pulsar wind Shocked stellar wind Model for LS 539 from Zabalza et al (212) Fig 1 Sketch of the proposed scenario at the periastron of a close-orbit system similar to LS539 The region of the CD where significant turbulence, and therefore wind mixing, are expected to take place is indicated with a wavy line The red regions indicate the two proposed emitter locations Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 4/27 4 / 27 o t w e 2 a p r d e s t t e o m t t e o w b i a l u b l

Extended radio emission Using VLBI observations, we resolve the radio emission for all the gamma-ray binaries which have been explored The Astrophysical Journal Letters,732:L1(6pp),211May1 8 Run A 6 τ+1 8 27 July 28 Run B 6 τ+21 8 27 August 17 Run C 6 τ+315 Moldón et al 28 June 6 4 4 4 MilliARC SEC 2-2 -4 2a 2= 144 AU MilliARC SEC 2-2 -4 MilliARC SEC 2-2 -4-6 -6-6 MilliARC SEC 8 6 4 2-2 -4-6 -8 LS I +61 33 Adapted from Dhawan et al (26) -8-8 -8 8 6 4 2-2 -4-6 -8 8 6 4 2-2 -4-6 -8 MilliARC SEC MilliARC SEC Figure 1 LBA images of PSR B1259 63 at 23 GHz North is up and east is to the left The dates and the days after the periastron passage (τ) arequotedatthetop of each panel The synthesized beam is displayed in the rectangle on the bottom-right corner of each image The red crosses mark the region where the pulsar should be contained in each run (see the text) As a reference, the size of the major axis of the orbit of PSR B1259 63/LS 2883 is shown in the first panel For each image, the displayed contours start at 3σ and increase by factors of 2 1/2 ;the1σ rms close to the source in each image from left to right is 3, 66, and 15 mjy beam 1 (A color version of this figure is available in the online journal) Table 2 Observational Parameters for Each Run PSR B1259 63 Moldón et al (211) Run MJD On-source Time No of Antennas τ+days a Orbital Phase a ( ) True Anomaly a ( ) θ a ( ) A 54394 543946 465 5 11(2) 15(2) 87(14) 12(14) 91(14) 125(14) 191 B 5432897 543294 375 5 21(2) 214(2) 1698(14) 1733(14) 983(3) 991(3) 279 C 5462329 5462367 42 4 3153(2) 3157(2) 25496(14) 25527(14) 16597(1) 16599(1) 346 5 / Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 5/27 27 Notes θ is the mean true anomaly plus 18 a Periastron passage at τ = MJD 5437971(1), from the fourth orbital solution obtained in Wang et al (24)

The gamma-ray binary LS I +61 33 LS I +61 33 B Ve star (125 ± 25 M ) d = 2 ± 2 kpc e = 72 ± 15 P orb = 26496 ± 3 d P super = 1667 ± 8 d Variability at all frequencies X-ray TeV: correlated(?) Radio TeV: correlated Optical Radio: correlated GeV TeV: anticorrelated Frail & Hjellming (1991), Casares et al (25), Gregory (22) 6 / Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 6/27 27

Radio emission of LS I +61 33 LS I +61 33 exhibits a large variability at radio frequencies Emission orbitally modulated (P orb 265 d) Ray et al (1997) The outbursts are not equal from cycle to cycle Radio pulsar searches have been conducted without success (McSwain et al 211, Cañellas et al 212) Therefore, the observed radio emission is not pulsed 7 / Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 7/27 27

Radio emission of LS I +61 33 Focusing on one single outburst No 1, 1998 RELATIONSHIP BETWEEN 2CG 135]1 AND LSI ]61 33 424 423 STRICKMAN ET AL Vol 497 OSSE range than was observed The OSSE integral Ñux 1 from 5È3 kev during VP 325 is (43 ^ 11) ] 1~4 photons cm~2 s~1, while the integral of the ASCA best-ðt 1 9 June 94 (97) 26 June 94 (62) power-law models extrapolated to the same energy range are (1 ^ 4) 1 ] 1~4 and (5 ^ 2) ] 1~4 photons cm~2 s~1 for radio phases 2 and 42 respectively These are inconsistent1 11 June 94 (6) 28 June 94 (7) at the 28 p and 34 p levels, assuming that the spectrum 1 does not change (in particular, does not harden signiðcantly) from 1 to 3 kev Since the relatively 1 12 June large 94 (9) OSSE Ðeld of view 1 July does 94 (79) not exclude the nearby QSO 241]622 as a possible source of 1 emission, we have also plotted the result of EXOSAT (2È1 kev) observations of the QSO (Turner & Pounds 1989), 1 15 June 94 (19) 2 July 94 (84) corrected for its position in the OSSE collimator response Note that the 17 1photon index power-law spectrum used to model the EXOSAT data is consistent with the OSSE result 1 17 June 94 (28) 3 July 94 (88) When we look 1at an overall view of the spectra from this region, no clear picture emerges However, the fact that OSSE and COMPTEL 1 18 June results 94 (31) fall signiðcantly 5 July above 94 (95) an extrapolation of the ASCA spectra and that the OSSE and 1 COMPTEL spectra are consistent with the X-ray spectrum of QSO 241]622, lead to the likelihood that at least some, 1 19 June 94 (35) 7 July 94 (2) and perhaps a signiðcant amount, of the Ñux observed by both OSSE and 1COMPTEL might come from the QSO rather than LSI ]61 33 Cycle-to-cycle variations in the X-ray emission 1 21 June 94 (42) 8 July 94 (7) may also contribute to the observed discrepancy More contemporaneous 1 1 X-ray 1 and1 c-ray data 1 are 1 required to test for the existence of cycle to cycle variability Frequency (GHz) of the LSI ]61 33 system and to better study the X-ray FIG 4ÈMultifrequency radio light curves of LSI ]61 33 observed Strickman FIG period 5ÈTime et recently evolution al of the (1998) reported LSI ]61 33 nonthermal by Paredes radio spectrum et al during (1997a) the outburst of 1994 JuneÈJuly Spectra are labeled according to the date and radio phase of the observation with the VLA corresponding to the radio outburst of 1994 JuneÈJuly Similarly, the low signiðcance of the OSSE light curves Error bars not shown are smaller than the symbol size a function shownof inorbital Figure phase 2 does Ultimately notthe allow high-energy us to usewith temporal wind optical signa- depth e ects dictating the time delay shock emissivity will depend on the geometrical and radiative tures characteristics to associate of the pulsar LSI cavity ]61 33 as the pulsar with orbits the OSSE the radiosource emission We 8 between the injection of the electrons and the visibility of around notethethat primary thelong Ñuxtimescale detected modulations by OSSE in the is relatively We shall nowhigher consider (at least qualitatively) / whether Benito Marcote bmarcote@amubes BeXRB emission from the system are explained by the inñuence of a each of the proposed models can explain the observed feaduring 214 - Valencia 1994 April observation, in apparent coincidence 8/27 27 Flux Density (mjy)

Superorbital modulation LS I +61 33 also exhibits a periodic modulation of the amplitude and the phases of these outbursts The period of this superorbital modulation is 1667 d ( 44 yr) 199A&A232377P Paredes et al (199) 9 / Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 9/27 27

Superorbital modulation LS I +61 33 also exhibits a periodic modulation of the amplitude and the phases of these outbursts The period of this superorbital modulation is 1667 d ( 44 yr) GORY Vol 575 Residuals days Flux Density mjy 8 6 c 4 4 2 2 mjy 1 / Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 36 38 4 42 44 46 48 5 1/27 27 ctral Index 1 a 5 5 1 2 4 6 8 1 Time days 35 3 25 2 15 1 5 b 2 4 6 8 Time days Orbital Phase Orbital Phase 1 8 6 4 2 36 38 4 42 44 46 48 5 P 2 Cycles 1 8 6 Gregory (22) 432 GREGO a b 25 mjy 28

Superorbital modulation 428 GREGORY Vol 575 Flux mjy P2 PHASE 6 84 175 15 225 GHz 83 GHz 125 1 75 5 25 5 1 15 2 25 2 15 1 5 5 1 15 2 25 12 1 8 6 P2 PHASE 84 15 P2 PHASE 15 6 225 GHz 83 GHz 225 GHz 83 GHz Spectral Index 4 2 2 4 4 2 2 4 4 2 2 P2 PHASE 6 84 5 1 15 2 25 P2 PHASE 84 15 5 1 15 2 25 P2 PHASE 15 6 4 4 5 1 15 2 25 5 1 15 2 25 Time days Time days Gregory (22) Fig 1 Left: Showsradiooutburstcross-correlationtemplatesat22and83GHzforthreeP2 phase bins Right: Showsthe22 83GHzspectralindex templates derived from the cross-correlation templates 11 / 27 Benito Marcote intercompared bmarcote@amubes before averaging to eliminate one BeXRB of the 214 - each Valencia P2 phase bin was 27 The templates shown in Figure 11/27

Goals At ν 1 GHz LS I +61 33 and all the gamma-ray binaries have been poorly explored We expect Absorption mechanisms arising from the spectrum (Bosch-Ramon 29) A large extended structure, nebula at arcsec-arcmin scales (Bosch-Ramon et al 211) A more steady emission along the time (Durant et al 211) Goals of this work Determine the low-frequency spectrum of LS I +61 33 Study its variability Search for absorption mechanisms Detect for first time the predicted extended emission 12 / Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 12/27 27

GMRT: Giant Metrewave Radio Telescope Located near Pune, India 3 45-m dishes Separated up to 25 km Y-configuration Observing frequencies: 154, 235, 325, 61, 142 MHz Resolution: 2 6 arcsec 13 / Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 13/27 27

LOFAR: The Low Frequency Array New generation (digital) radio interferometer The Netherlands (core), France, Germany, Sweden, United Kingdom, Poland 24 core, 14 remote and 8 international stations 48 HBAs and 96 LBAs each LBA: 3 8 MHz & HBA: 11 25 MHz Baselines: 1 m 15 km Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 14 / 14/2727

LOFAR: The Low Frequency Array We started a collaboration with the Transient Key Project (TKP) of LOFAR We have access to the following data for LS I +61 33 at 15 MHz: One deep observation during the commissioning phase (211) Testing of the current status of LOFAR Still high noise (rms = 4 mjy) Regular short observations with the Radio Sky Monitor (RSM) 6 pointings in 213 4 are already analyzed More pointings in the coming months 15 / Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 15/27 27

Studying LS I +61 33 at ν < 1 GHz We have analyzed 3 unpublished archival GMRT observations at 235/61 MHz in 25 28 and one observation at 15 MHz in 28 6 5 º= MHz 4 3 GMRT 2 1 LOFAR 535 54 545 55 555 56 565 MJD 16 / Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 16/27 27

Results from the GMRT observations GMRT data from Nov 25 to Feb 26 1 : :5 Á 61 MHz 235 MHz 8 S º = mjy 6 4 2 537 5372 5374 5376 5378 MJD 17 / Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 17/27 27

Results from the GMRT observations GMRT data from Nov 25 to Feb 26 1 61 MHz 235 MHz 8 S º = mjy 6 4 2 : :5 1: 1:5 2: Á 18 / Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 18/27 27

Results from the GMRT observations GMRT data from Nov 25 to Feb 26 2:5 2: (235 61 MHz) 1:5 1: :5 : :5 : :5 1: 1:5 2: Á 19 / Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 19/27 27

Results at very low frequencies With the GMRT observation at 154 MHz we detected for first time a gamma-ray binary at these very low frequencies ν = 154 MHz Bandwidth 32 MHz Beam: 3 14 arcsec 3 24 18 (J2) 12 Point-like source S ν = 37 ± 2 mjy 6 +61 43 m 42 m 41 m 4 m (J2) 39 m 2 h 38 m 2 / Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 2/27 27

Results at very low frequencies 6 RSM pointings to LS I +61 33 with LOFAR up to now ν = 149 MHz Bandwidth 78 MHz Beam: 27 15 arcsec g 4 analyzed obs 3 detections > 3σ Point-like source Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 21 / 21/2727

Results at very low frequencies Preliminary results from the RSM LOFAR observations Detection of LS I +61 33 in three of four analyzed observations The behavior of the source is variable along the time 1 5 φ LOFAR 15 MHz 8 Sν / mjy 6 4 2 5635 564 5645 MJD 22 / Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 22/27 27

Discussion Unresolved questions Variability at 15 MHz, orbitally modulated? We observe a large variability at 61 MHz but a more constant emission at 235 MHz What is the origin of this behavior? Outbursts absorbed at 235 MHz? Emitting region becomes optically thick at the very low frequencies? Changes in the electron density of the surrounding plasma with produces different absorption mechanisms? Changes in the emitting region? lk 18 lk, B 75 B 23 / Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 23/27 27

Consistency of the results Comparing these results with the ones from Strickman et al (1998) We need to take into account the superorbital period: ϕ super Strickman 7 ϕsuper GMRT 2 ϕsuper LOFAR 8 S 22 175 mjy S 22 12 mjy S 22 175 mjy No 1, 1998 RELATIONSHIP BETWEEN 2CG 135]1 AND LSI ]61 33 423 OSSE range than was observed The OSSE integral Ñux from1 5È3 kev during VP 325 is (43 ^ 11) ] 1~4 photons cm~2 s~1, while the integral of the ASCA best-ðt power-law models extrapolated to the same energy range are (1 ^ 4) ] 1~4 and (5 ^ 2) ] 1~4 photons cm~2 s~1 for radio phases 2 and 42 respectively These are inconsistent 8 at the 28 p and 34 p levels, assuming that the spectrum does not change (in particular, does not harden signiðcantly) from 1 to 3 kev Since the relatively large OSSE Ðeld of view does not exclude the nearby QSO 241]622 as a possible source of emission, 6 we have also plotted the result of EXOSAT (2È1 kev) observations of the QSO (Turner & Pounds 1989), corrected for its position in the OSSE collimator response Note that the 17 photon index power-law spectrum used to model the EXOSAT data is consistent with the OSSE 4 result When we look at an overall view of the spectra from this region, no clear picture emerges However, the fact that OSSE and COMPTEL results fall signiðcantly above an extrapolation 2 of the ASCA spectra and that the OSSE and COMPTEL spectra are consistent with the X-ray spectrum of QSO 241]622, lead to the likelihood that at least some, and perhaps a signiðcant amount, of the Ñux observed by both OSSE and COMPTEL might come from the QSO 61 MHz 235 MHz rather than LSI ]61 33 Cycle-to-cycle variations in the X-ray emission : may also contribute :5to the observed discrepancy 1: 1:5 2: More contemporaneous X-ray and c-ray data are required to test for the existence of cycle to cycle variability Á of the LSI ]61 33 system and to better study the X-ray FIG 4ÈMultifrequency radio light curves of LSI ]61 33 observed period recently reported by Paredes et al (1997a) with the VLA corresponding to the radio outburst of 1994 JuneÈJuly Similarly, the low signiðcance of the OSSE light curves Error Strickman Benito Marcote bars not shown are smaller et bmarcote@amubes than al the symbol (1998) size shown in Figure BeXRB2 214 does not - Valencia allow us to use temporal signa- 24/27 S º = mjy 24 / 27

Consistency of the results Considering the estimations from Gregory (22), we extrapolate the fluxes from Strickman et al (1998) to the ones expected at the superorbital phase observed in the GMRT data 2 1 5 Sν / mjy 2 1 5 Scaled φ = 8 Strickman et al (1998) 2 5 1 2 5 1 ν / GHz 25 / Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 25/27 27

Consistency of the results Considering the estimations from Gregory (22), we extrapolate the fluxes from Strickman et al (1998) to the ones expected at the superorbital phase observed in the GMRT data 2 1 Sν / mjy 5 2 1 φ = 8 φ φ 5 5 2 5 1 2 5 1 ν / GHz 26 / Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 26/27 27

Conclusions Detection for the first time of a gamma-ray binary at 15 MHz Claiming of variability at this frequency The outbursts are not present at 235 MHz but still visible at 61 MHz The origin of this behavior is still unclear The extended emission at very low frequency is not detected in the LOFAR and GMRT images Sensitivities 1 mjy and/or resolutions of arcsec are necessary More data at 15 MHz could resolve the origin of the observed variability The characterization of the behavior of LS I +61 33 at low frequencies still requires deeper studies in combination with high-frequency observations 27 / Benito Marcote bmarcote@amubes BeXRB 214 - Valencia 27/27 27