1. September 28 th, 20068 th EVN Symposium, Torun Winds in collision: high energy particles in massive binary systems Sean M. Dougherty (NRC) In collaboration with: Julian M. Pittard (Leeds) Evan O’Connor (PEI) Nick Bolingbroke (Victoria) Perry M. Williams (IfA, Edinburgh) Tony Beasley (ALMA)

2. Observations of Massive stars Most massive stars (O or B-type) –positive spectra from IR to radio –brightness temperature ~ 10 4 K –Thermal emission – expected! in a few systems –“flat” or negative spectra in the radio –brightness temperature ~ >10 6 K –Non-thermal radio emission – where from? –Also thermal/non-thermal X-rays,  -rays(?)

3. September 28 th, 20068 th EVN Symposium, Torun Moran et al. 1989 Williams et al. 1997 WR147 High resolution observations - MERLIN @ 5GHz: 50 mas = 77AU @ 650pc The radio structure of a colliding-wind binary –two components - one thermal + one non-thermal IR obs resolve two stars –WR+ O/B type Position of NT emission: consistent with position of wind-wind collision region

4. September 28 th, 20068 th EVN Symposium, Torun D What is a wind-collision region? Two massive stars with stellar winds Contact discontinuity where ram pressures are equal Standing shocks on either side of the CD Thermal X-ray emission from shock-heated gas in collision region Particle acceleration in wind-collision region – at the shocks –and/or through reconnection at the CD -> non-thermal emission radio, X-ray etc. Relativistic particles Higher magnetic, particle & radiation densities than in SNR –Good particle acceleration laboratory

5. September 28 th, 20068 th EVN Symposium, Torun –radially symmetric, isothermal winds, collide at terminal velocity, axis-symmetric –constrained by radio spectrum and images Radiative transfer –Assume cylindrical symmetry, ideal gas, adiabatic index=5/3 –Treatment of non-thermal emission U rel =  U thermal tangled magnetic field –assumption of shock acceleration power-law energy distribution at the shocks p = 2 electron momentum spectrum accelerated at shocks –Electron energy spectrum evolves downstream due to IC cooling. –Thermal/non-thermal emission & absorption - determined from 2D hydro grid Modelling radio emission from CWB systems Early models of CWB systems tended to be simple. –Point source non-thermal emission, radially symmetric winds – maintains analytic solutions –No consideration cooling mechanisms (e.g. Compton cooling – important - even for wide systems c.f. 146, 147) or other absorption. Hydro-modelling of CWBs

6. Hydro model of a CWB – spatial density distribution WR star – dense stellar wind O star – less dense wind Wind-collision region Wind-collision region hot plasma + NT particles

7. September 28 th, 20068 th EVN Symposium, Torun Intrinsic NT + IC cooling + Razin effect & ff absorption (  max =1000 ) Typical radio spectra

8. September 28 th, 20068 th EVN Symposium, Torun 1.6 GHz 22 GHz No IC coolingWith IC cooling

9. September 28 th, 20068 th EVN Symposium, Torun NT emission does not dominate Modelling the radio spectrum of WR147 Thermal flux NT flux – poor data constraints for modelling Total flux

10. September 28 th, 20068 th EVN Symposium, Torun Simulated MERLIN 4.8 GHz Spatial distribution in WR147

11. September 28 th, 20068 th EVN Symposium, Torun WR146 MERLIN obs - spatially resolved thermal and NT components (cf. WR147) Brightest radio CWB – NT emission dominates VLBI imaging => Excellent data constraints

12. September 28 th, 20068 th EVN Symposium, Torun WR146 (2) (O’Connor et al. 2005, 2007) Greyscale - EVN 5 GHz Contours – VLA+PT 43 GHz Crosses denote stellar positions - HST 43-GHz model 4.8-GHz EVN model (contours) EVN 5 GHz

13. September 28 th, 20068 th EVN Symposium, Torun NT emission in massive stars : binary required? In spatially resolved WR-systems, NT emission is from a wind-collision region c.f. WR146, 147, 140 Are all systems with NT emission “binary” systems? –25 WR stars - mixture of both single and binary that have measured radio spectra –11 systems have spectra identifying non-thermal emission WR 11, 48, 98a, 104, 105, 112, 125, 137, 140, 146, 147 10 of these WR stars have OB-binary companions ARE BINARY WR stars with NT radio emission ARE BINARY

14. September 28 th, 20068 th EVN Symposium, Torun And O+O star systems?: Cyg OB2 #5 Eclipsing O6+O6 binary – 6-day period! VLA 5GHz obs –thermal + non-thermal sources Binary coincident with thermal emission B-type star & NT emission (Contreras et al. 1999) Wind-collision region? Radio-detected O-stars –60% exhibit non-thermal emission –“large” fraction are binary VLA 5 GHz ARE BINARY? O stars with NT radio emission ARE BINARY?

15. State of Play: Wind-collision regions are laboratories for investigating particle acceleration Non-thermal emission in massive stars required a binary/companion –Certainly true for WR stars –Starting to look like the case for O stars Successful models of the both radio spectrum and spatial distribution of emission

16. September 28 th, 20068 th EVN Symposium, Torun A major reason why non-thermal emission is clearly seen in WR147 + WR146 –the systems are very wide –free-free opacity along l.o.s. to the wind-collision zone is small But --- “static” systems –families of satisfactory models –Ill-defined system parmeters = ill-constrained models Shorter period, eccentric systems –possibility of well-specified orbit parameters –variable radiation density – IC cooling  variable high energy emission –variable ion density  variable circumstellar ff opacity to WCR WR 140 is the best studied WR+OB binary –WR + O in a 7.9 year, eccentric (e=0.88) orbit - orbit size ~ 15 AU –Radio-bright – dramatic variations in radio emission as orbit progresses –WCR resolved by VLBI -> good data constraints. –IC cooling important –Flow time ~ R OB /v WR ~ 100 hrs –IC Cooling t IC ~12 hrs @apastron @periastron ~250 times shorter! –considerably shorter than flow time –at all radio frequencies under consideration –High eccentricity + good data  excellent lab for studying wind-collision phenomena WR 140 - the CWB laboratory

17. September 28 th, 20068 th EVN Symposium, Torun Cartoon of the colliding-wind region in WR140 Orbit parameters from Williams et al. 1990 - interaction region based on Eichler & Usov 1993

18. September 28 th, 20068 th EVN Symposium, Torun The radio light curve of WR140 8 years of VLA observations (White & Becker 1995) + WSRT data (Williams p.c.) 2cm 6cm 21cm

19. September 28 th, 20068 th EVN Symposium, Torun VLBA imaging of WR140 23 epochs @ 3.6 cm phase~ 0.74 -> 0.93 (from Jan 1999 to Nov 2000) Resolution ~ 2 mas Linear res ~ 4 AU Non-thermal emission (T b ~10 7 K) Resolved – “curved” emission region => wind-collision region Observe rotation & pm of emission region –Full orbit definition – particularly inclination –Distance independent of stellar parameters => Much needed modelling constraints

20. September 28 th, 20068 th EVN Symposium, Torun

21. September 28 th, 20068 th EVN Symposium, Torun “resolved” the binary components –12.7 mas @ 151.7 degrees at phase 0.297 Combined with other known orbit parms  families of solutions for a,  Orbit definition could wait for more IOTA observation, but in the meanwhile….. IOTA observation – Monnier et al. 2004

22. September 28 th, 20068 th EVN Symposium, Torun VLBA obs –assume axis of symmetry along line-of-centres –Rotation of WCR as orbit progresses => O star moves from SE to E of WR star during observations => derive inclination. Orbit inclination

23. September 28 th, 20068 th EVN Symposium, Torun Orbit solution –a = 9.0+/-0.5 mas;  = 353+/- 3 degrees ;  =122+/-5 degrees Orbit & distance of WR 140 Distance – NOT from stellar parameters! –a sin i = 14.10 +/- 0.5 AU => a = 16.6 +/- 1.1 AU for i = 122 deg. –a = 9.0 +/- 0.5 mas  Distance = 1.85 +/- 0.16 kpc O supergiant All important system parms now defined!!! –Stellar types –Distance –All orbit parameters (including inclination) –ALL VERY IMPORTANT to modelling

24. September 28 th, 20068 th EVN Symposium, Torun Modelling the spectra A:  =0.22,  e =1.38x10 -3,  B =0.05 B:  =0.02,  e =5.36x10 -3,  B =0.05 Using these orbit parms… Constrain  (& mass-loss) with thermal X-ray observations - independent of wind-clumping. Successfully model individual orbit phases – good! Most importantly, establish a value for B, the magnetic field strength Phase 0.837

25. September 28 th, 20068 th EVN Symposium, Torun Possible to constrain models with VLBI obs Modelling 8 GHz VLBI observations of WR140 - demands good observations

26. September 28 th, 20068 th EVN Symposium, Torun 1.6 5 8.3 15 22/43 Radiometry New multi-frequency VLA observations Repeat fluxes from previous orbit(s) –Suggests that emission arises from a “well-behaved” process –Similar behavior seen in O+O star binary systems

27. September 28 th, 20068 th EVN Symposium, Torun First stab at modelling WR140 Looking good But… Relationship from one to another is UNCLEAR – bad Continues as a work in progress

28. September 28 th, 20068 th EVN Symposium, Torun Modelling the spectra Models give all radio emission components Most important -- intrinsic L syn, the non-thermal radio power Now have estimate of B and intrinsic L syn And why are these so important? Phase 0.837 Thermal stellar wind L syn

29. September 28 th, 20068 th EVN Symposium, Torun WR140 lies within the error box of 3EG J2022+4317 EGRET (100MeV – 20 GeV) From Benaglia & Romero (2003) Is WR140 a gamma-ray source?

30. September 28 th, 20068 th EVN Symposium, Torun IC scattering Calculation of high energy spectrum Typical models: U B /U ph only at stagnation point Use L obs rather than L intrinsic for L sync In our models, spatial variations for: U B /U ph L sync n e, n i (thermal/nt) Determined by fits to radio spectra  high energy spectra  0 -decay - from hadronic collisions involving NT ions (evidence of relativistic ion production)  p+p   0 +X,  0   +  Assumed  rel,ions /  rel,electrons ~ 10-100 (Constrained by energy budget considerations) Relativistic bremsstrahlung from interactions of NT electrons with thermal particles

31. September 28 th, 20068 th EVN Symposium, Torun NT bremsstrahlung Pion decay Emission from WR140 at phase 0.8 Photon pair production opacity Inverse Compton Radio ASCA INTEGRALGLASTVERITAS

32. September 28 th, 20068 th EVN Symposium, Torun Predicted luminosities and fluxes at phase 0.8 Predicted EGRET flux is 2.5x lower than 3EG J022+4317 Predicted observing time for a 5  detection is 350 ksec (~4days) GLAST 5  sensitivity at E > 100MeV for a 2yr all-sky survey is 1.6 x 10 -9 ph s -1 cm -2 (should detect with GLAST) In turn, use high-energy observations to then constrain the models

33. September 28 th, 20068 th EVN Symposium, Torun Looks like a duck, quacks like a duck – it’s a duck! Cyg OB2 #9 Not a spectroscopic binary –Apparently single! VLBA obs –looks like a WCR Other evidence of companion? e.g. WCR rotate on plane-of-sky? Variable radio emission –2.4-yr period Radio obs => binary Is there a WCR?

34. September 28 th, 20068 th EVN Symposium, Torun Summary Colliding winds in early-type binaries are useful laboratories for investigating particle acceleration –New insights into particle acceleration – at higher mass, B-field, and energy densities than in SNRs Excellent data on a number of systems –Radiometry and imaging – WR140 and WR146 – more recently Cyg OB2 #9 –WR140 has well-constrained system parms from high-resolution imaging – very important for modelling –WR140 and Cyg OB2 #9 – similar flux orbit-to-orbit - emission arises from well-behaved process(es) Hydro models of plasma distribution –Successful models of spectrum and spatial distribution of emission. –Some issues revealed in models of WR146 – better data constraints high-frequency spectrum & spatial extent of emission Models lead to intrinsic synchrotron radio emission and magnetic energy density –used to estimate the non-thermal X-ray and  -ray emission Insight into particle (ions & electrons) acceleration efficiencies, and the B-field Exciting period with respect to new data from INTEGRAL, GLAST, HESS, VERITAS, etc. –Constrain models (e.g. pion decay signature of relativistic ion production). Dougherty, Beasley, Claussen, Zauderer, Bolingbroke, 2005, ApJ 623, 447 Pittard, Dougherty, Coker, O’Connor, Bolingbroke, 2006, A&A 446,1001 Pittard & Dougherty, 2006, MNRAS, in press O’Connor, Dougherty, Pittard, Williams, 2007, in prep