High Energy Processes in Wolf-Rayet Stars

arXiv:1812.03954v1 [astro-ph.SR] 10 Dec 2018 1 Eal [email protected] *Email: Correspondence 4 3 2 1 2.1 high have stars WR Galaxy. the in and known evolved temperatures effective are most few the are a stars only WO (WO). (WC), oxygen-type on carbon-type and based (WN), subtypes nitrogen-type three spectra: into optical their classified broadly are stars WR Skinner* L. Stephen Stars Wolf-Rayet in Processes Energy High TYPE ARTICLE xxx/xxxx DOI: 2018; December 6 Received uioiiso Nsassa ag faottoodr of orders two about of range a X-ray span stars. stars WC WN from of not luminosities surprisingly but stars, WO and (non-binary) WN single putatively from detected been have X-rays yia asls rates mass-loss typical pesfrW n CsasaetpclyV typically are stars WC and WN for speeds Rbnre r oeta ore of sources binaries. potential WR are some binaries in WR detected been has emission continuum radio (synchrotron) Nonthermal properties. with X-ray observed compatible the not cases some in are predictions model but shock winds, supersonic their with usually associated is shocks to emission X-ray attributed Their sources. X-ray strong are stars ms km 2 acceleration. ray cosmic Galactic for ugra cd c. oa Bulgaria Sofia, Sci., Acad. Obs., Bulgarian Astron. Natl. & Astron. of Inst. Austria Vienna, Vienna, of Univ. Astrophysics, of Dept. Switzerland Dorf, Davos WRC, PMOD, USA 80309-0389, CO Boulder, Colorado, of Univ. CASA, −1 X-RAYS OVERVIEW STARS: WOLF-RAYET n vnhge o Osas ot u o l,WR all, not but Most, stars. WO for higher even and , igeW Stars WR Single > eie:0 ot 0000 Month 00 Revised: 000Kadteedu id with winds tremendous and K 20,000 1 M ̇ enrSchmutz Werner | ∼ 10 −5 M

⊙ ryeiso n sites and emission -ray yr tr:Wl-ae;sas niiul(QCp;Xry:sta X-rays: Cep); (CQ individual stars: Wolf-Rayet; stars: KEYWORDS: emission X-ray test stringently which orbit full a covering e eut from X-ray their results on new mainly focusing b stars, We WR in stars. processes of energy generations future for medium material interstellar raw the providing enrich end winds the powerful approaching Their rapidly novae. cores, their in burning nuclear ofRyt(R tr r asv ( massive are stars (WR) Wolf-Rayet −1 cetd 0Mnh0000 Month 00 Accepted: emnlwind Terminal . ∞ ∼ 2 00-2500 - 1000 aulGüdel Manuel | Chandra antd ihtpclvle o L log values typical with magnitude trsetaadi h pcrmo h Osa R12 is 142, WR far so star is WO origin unexplained. Its the models. shock of radiative from spectrum anticipated the not in and WN deter- spectra single in be star prominent to is which remains plasma, hotter speeds the But wind mined. much and with stars rates WR mass-loss to of relevance higher emission their X-ray but soft stars, the O-type explaining of some in result success mod- a had Such have as 1980). els White wind & the Lucy (e.g. in instabilities form line-driven that shocks X-ray from posit which production models, shock wind radiative of predictions temperatures. distributed of is range plasma the a X-ray reflect over fully the not since do conditions models physical 2T true Such 2012). 2010; al. et ner otrcmoeta kT at component hotter hc ie o L log WC gives a single which from a is for far obtained so limit yet star upper not stringent responsi- reasons most partially for The be ble. may quiet absorption X-ray 2018). wind or al. strong but et faint known, Skinner X-ray 2010; are al. stars et WC (Sokal kpc 1.74 of distance e ta.21) h nyglci Osa eetds a in far so L detected log star with WO 142 galactic WR only is X-rays The 2012). al. et ner olcmoeta kT at component with cool plasma a optically-thin (2T) two-temperature a as modeled 2006). al. et (Skinner kpc 2.11 of h eprtr fteco opnn scnitn with consistent is component cool the of temperature The h -a pcr fsnl Nsascnb acceptably be can stars WN single of spectra X-ray The bevtoso h cisn WR eclipsing the of observations 3 vtzrZhekov Svetozar | ≥ 0M 10 x ≤ Chandra ⊙ 99 rss ergs 29.99 rs vle tr negigadvanced undergoing stars evolved ) 2 1 ≈ ≈ e [T keV 5 - 2 . . e [T keV 0.7 - 0.3 x bevto fW 3 (WC8) 135 WR of observation ≈ 4 13ess ers 31.3 −1 models. x tits at ifl umrz high- summarize riefly fterlvsa super- as lives their of ∼ ihhayelements, heavy with msin epresent We emission. 2 + 10 ≈ iayC Cep CQ binary O AADR2 GAIA 32±1 −1 1 0-6 K (Skin- MK] 60 - 20 ≈ tits at K n a and MK] 8 - 3 rss ergs GAIA −1 distance (Skin- DR2 2

2.2 WR Binaries TABLE 1 Chandra Observations of CQ Cep Massive WR binaries are typically strong X-ray sources. This includes WC systems such as 2 Vel(=WR11;WC8+O7) and ObsId StartDate Duration Phase Rate Pvar −1 WN systems such as WR 147 (WN8+B0.5V). In some cases, (ks) () (cks ) the X-ray emission is bright enough to obtain high-resolution 145382013Mar19 85.6 (−0.10, 0.50) 24.7±5.6† 0.22 X-ray grating spectra, allowing individual emission lines to 17734 2017Feb27 18.3 (−0.07, 0.06) 20.9±4.0 0.16 be identified and studied (e.g. 2 Vel, Skinner et al. 2001; 20016 2017Mar4 16.4 (0.66,0.78) 20.3±3.9 0.07 WR 140, Pollock et al. 2005; WR48a, Zhekov et al. 2014). 20017 2017Mar5 33.9 (0.43,0.67) 20.5±4.1 0.05 Line information, when available, constrains the plasma tem- 20018 2017Mar5 19.2 (0.79,0.93) 20.6±4.7 0.04 perature distribution, metal abundances, and distance from the O star is in front at = 0. P is the probability of variable star(s) where the line forms. var count rate based on the Chandra glvary statistical test. The X-ray emission of WR binaries is potentially an admix- †ACIS-S effective area was higher in 2013. ture of multiple components including that of the individual stars (and their winds) and colliding wind (CW) shock emission originating between the stars, or near the surface of the O star passing in front of the WR star at = 0 (Skinner et al. star with the weaker wind (Usov 1992). In most cases, these 2015). The second half was observed in four exposures during different components cannot be spatially-resolved with current Feb.-March 2017, again capturing the O star in front as well as generation X-ray telescopes. the WR star in front. Simultaneous optical light curves were The maximum CW shock temperature for an adiabatic obtained using the Chandra Aspect Camera Assembly (ACA). 2 shock is kTcw ≈ 1.96[V⟂/1000 km/s] keV. Here, is the The main objective was to search for X-ray variability during mean atomic weight (amu) in the wind ( ≈ 4/3 for He- eclipses, which is expected if the hottest X-ray plasma is con- dominated WN winds) and V⟂ is the wind velocity component fined to the region on or near the line-of-centers between the perpendicular to the shock front. The hottest plasma is pre- two stars. dicted to lie on or near the line-of-centers where V⟂ ≈ V∞, Both the primary and secondary visual eclipses are clearly if the winds have reached terminal speeds. This latter condi- seen in the optical light curves (Fig. 1-top panels). Comparison tion will be satisfied in wide binaries (Porb ∼ years) but not in of the overlapping ACA light curves indicates that the system close binaries (Porb ∼ days). For typical WR wind speeds V∞ was slightly brighter in 2013 than in 2017. Optical variability ≈ 1000 - 2500 km/s, maximum shock temperatures kTcw,max ≈ is also apparent from significant scatter in the high-precision 2 - 12 keV are expected. Temperatures in this range are indeed GAIA photometry at similar phases but different epochs. The observed in some WR binaries, but also in some (apparently) times of minimum ACA optical brightness at =0 in 2013and single WR stars. 2017 give an accurate orbital period P = 1.641239 (±8.0e−07) d. Analysis of historical data suggests that the period may be variable (Koenigsberger et al. 2017). 3 THE ECLIPSING WR BINARY CQ CEP In contrast, the phased X-ray light curves (Fig. 1-bottom CQ Cep (= WR 155) is an eclipsing WN6+O9 binary system panels) show only low-level fluctuations. No statistically sig- in a near-circular high inclination 1.64 d orbit (Demircan et al. nificant variability is present and the mean count rates of the 1997). The masses and radii of the two stars are nearly equal observations agree to within ±1 (Table 1). ACIS-S effective (M∗ ≈ 21 M⊙,R∗ ≈ 8 R⊙). Their separation is D ≈ 20 R⊙, area declined during 2013-2017 due to contaminant buildup, placing the two stars nearly in contact (Demircan et al. 1997). resulting in lower count rates in 2017. At such close separation, the winds will not have reached ter- Phase-resolved X-ray spectra obtained in 2017 are shown in minal speeds before colliding. The higher momentum of the Figure 2-top. All four spectra are similar and spectral fits show WR wind will overpower the O star wind and the CW shock no significant difference in their observed X-ray fluxes. The will form at or near the O star surface. CQ Cep is a superb sys- median energies of the photons detected in the observations tem for testing CW model predictions at close spacing where are nearly identical, i.e. E50 = 1.88, 1.90, 1.87, and 1.87 keV the winds will be at subterminal speeds and radiative cooling (background is negligible). The spectra are overlaid in Figure may be important (Stevens et al. 1992). 2-bottom. Several blended emission lines are detected includ- We have observed CQ Cep with the Chandra X-ray Obser- ing Ne X, Mg XI, Si XIII, S XV, Ar XVII and possibly Ne vatory using the ACIS-S CCD imaging spectrometer over a IX and Ca XIX. Faint Fe K (Fe XXV) emission was detected full orbit (Table 1). The first half of the orbit was observed in a in the 2013 observation (Skinner et al. 2015), tracing very hot single uninterrupted observation in March 2013, capturing the plasma at T ∼ 63 MK. 3

Hipparcos mag | | 9.2 9 8.8 3−5−2017 φ : (0.43, 0.67) CQ Cep GAIA | |

| 0.01 0.02 0.03 0.04 | |

GAIA G mag 3−4−2017 | φ : (0.66, 0.78) 8.8 8.6 8.4

−0.2 0 0.2 0.4 0.6 0.8 1 Orbital Phase 0.01 0.02 0.03 0.04 CQ Cep (Chandra ACA) 3−5−2017 φ : (0.79, 0.93)

O in front WR in front 0.01 0.02 0.03 0.04

9.2 9| 8.8 8.6 | |

−0.2 0 0.2 0.4 0.6 0.8 1 Orbital Phase 0.3−8 keV (binsize = 1800 s) S XV

| |

0.01 0.02 0.03 0.04 0.05 2017. Top: Separated by observation (Table 1). Bottom: All

−0.2 0 0.2 0.4 0.6 0.8 1 Orbital Phase four spectra overlaid. 2−8 keV (binsize = 1800 s) CQ Cep (Chandra ACIS)

O in front WR in front The intrinsic (unabsorbed) X-ray luminosity of CQ Cep at its GAIA distance (Table 2) is at the high end of the range for WR stars. The predicted luminosity for a WR wind shocking onto ACIS−S rate (Cts/s) 0.01 0.015 0.02 0.025

−3 the surface of an O-type companion of radius R at separation 2 D is Lx,cw = 0.125(R∕D) Lwind,wrF . The WR wind kinetic 0 5×10 −0.2 0 0.2 0.4 0.6 0.8 1 ̇ 2 Orbital Phase luminosity is Lwind,wr = (1/2)MwrV∞,wr. The correction factor F < 1 accounts for changes in the direction and magnitude of FIGURE 1 Phased CQ Cep light curves. Top: Optical (Hip- the WR wind velocity vector across the O-star surface, which parcos epoch 1990.0-1993.1, GAIA G mag epoch 2013.3- must be considered in closely-spaced binaries where the stel- 2014.6; 1 errors). 2nd row: Optical (Chandra ACA), color lar radii are nearly equal, as in CQ Cep. Inserting appropriate coded by observation date (Table 1). 3rd row: Chandra X-ray values for CQ Cep, one obtains log L = 35.11 - 35.27 ergs broad (0.3-8 keV). Bottom: Chandra X-ray hard (2-8 keV). x,cw s−1 (Skinner et al. 2015), which is about fifty times greater than observed. Similar discrepancies have been noted in other Spectral fits with non-solar abundance 2T models are able close WR binaries and explanations involving inhomogeneous to reproduce the spectra. The inferred absorption (NH) and winds have been proposed (Zhekov 2012). plasma temperatures (kT) are somewhat model dependent, as The absence of significant X-ray variability during eclipses is the intrinsic (unabsorbed) X-ray luminosity (Lx). Fits of the is remarkable in comparison to the dramatic optical variabil- 2013 spectrum were discussed in Skinner et al. (2015). Table 2 ity. This lack of variability, even in the high 2-8 keV energy compares fits of the 2013 and 2017 spectra using a a 2T plane- range, indicates that the hottest X-ray plasma occupies a region parallel shock model. The temperature of the hot component much larger than the scale of the binary system and is not

(kT2 ≈ 3.4-3.6 keV) is about twice the predicted value for the localized near the line-of-centers. If the hot plasma does form WN6 wind shocking onto the surface of the O star companion in a CW shock, then orbital motion may have distorted the at subterminal speed (Skinner et al. 2015). However, the pre- shape of the interaction region and displaced it away from the dicted shock temperature depends on the assumed WR wind line-of-centers (Skinner et al. 2015, and references therein). acceleration proﬁle in the region between the stars, which is not well-constrained observationally. 4

TABLE 2 CQ Cep Spectral Fits (shock model) of -ray telescopes will be needed to quantify the properties of WR binaries in this very high energy (∼1 - 100 GeV) regime.

Year NH,1, NH,2 kT1, kT2 Fx† log Lx‡ (1021 cm−2) (keV) (erg s−1) 6 CONCLUSIONS 2013 4.4[1.7],0.3[0.1] 0.6[0.1],3.4[0.6] 2.09 33.43 2017 6.5[2.7],0.3[0.1] 1.0[0.3],3.6[0.9] 2.20 33.53 Wolf-Rayet stars exhibit a diverse range of high-energy phe- Fits are based on a 2T plane-parallel shock model (XSPEC nomena, largely attributed to their powerful shocked winds. vpshock). The 2017 values are from a simultaneous fit of all But details of the underlying emission mechanisms are still not well-understood. Hot X-ray plasma is prevalent in WN 4 spectra. The absorption NH of each temperature component and WO stars but temperatures and luminosities are not fully was allowed to vary independently. The tabulated NH values do in accord with shock model predictions. Other factors such not include interstellar absorption, held fixed at NH = 4.4e21 cm−2. Solar abundances were used for the cool component and as magnetic fields may be involved in WR X-ray production. generic WN abundances for the hot component (Skinner et al. Mysteriously, carbon-rich WC stars remain undetected in X- 2015). Brackets enclose 1 errors. rays. Nonthermal (synchrotron) radio emission has been traced †Absorbed flux (0.3-8 keV). Units: 10−13 ergs cm−2 s−1. to the colliding wind region in WR binaries, but the mech- ‡Unabsorbed X-ray luminosity (d=3.32 kpc; GAIA DR2). anism(s) which accelerate electrons to relativistic energies is still debated. Gamma-ray emission has been reported from the nearest WR binary 2 Vel. Future improvements in sensitivity 4 NONTHERMAL RADIO EMISSION and angular resolution will be needed to confirm this intriguing discovery and search for -rays in more distant WR systems. Thermal (free-free) radio continuum emission is produced by the strong photoionized winds of WR stars. In addition, radio continuum emission characterized by negative or flat spectral ACKNOWLEDGMENTS indices ≤ 0 (flux density varies with frequency as F ∝ ) interpreted as synchrotron emission has been detected in sev- This work was supported by SAO/CXC award GO6-17009X. eral WR+OB binaries (Abbott et al. 1986). It is believed that the nonthermal radio emission is produced in the CW region between the stars where electrons are accelerated to relativis- REFERENCES tic energies by shocks, but magnetic fields may also play a role (Kissmann et al. 2016). Support for the origin of nonthermal Abbott, D. e. a. 1986, ApJ, 303, 239. radio emission in the CW region comes from spatially-resolved Benaglia, P., & Romero, G. 2003, A&A, 399, 1121. Churchwell, E. e. a. 1992, ApJ, 393, 329. radio observations of WR 147 (Churchwell et al. 1992). The Demircan, O. e. a. 1997, AN, 318, 267. detectability of any nonthermal radio emission is affected by Kissmann, R. e. a. 2016, ApJ, 831, 121. absorption and viewing geometry. As a result, nonthermal Koenigsberger, G. e. a. 2017, A&A, 601, A121. emission is not seen in all WR binaries. Lucy, L., & White, R. 1980, ApJ, 241, 300. Pollock, A. e. a. 2005, ApJ, 629, 482. Pshirkov, M. 2016, MNRAS, 457, 99. Schmutz, W. e. a. 1996, A&A, 311, L25. 5 GAMMA-RAYS Skinner, S. e. a. 2006, Ap&SS, 304, 97. Skinner, S. e. a. 2010, AJ, 139, 825. Skinner, S. e. a. 2012, AJ, 143, 116. The formation of a WR+BH binary may be accompanied by Skinner, S. e. a. 2015, ApJ, 799, 124. a -ray burst (Tutukov & Cherepashchuk 2003). The best can- Skinner, S. e. a. 2018, ApJ, in press. didate for a WR+BH system to date is Cyg X-3 (Schmutz Sokal, K. e. a. 2010, ApJ, 715, 1327. et al. 1996). Some models predict that electrons in the CW Stevens, I. e. a. 1992, ApJ, 386, 265. region of WR binaries will be accelerated up to -ray ener- Tutukov, A., & Cherepashchuk, A. 2003, Ast. Rep., 47, 386. Usov, V. 1992, ApJ, 389, 635. gies by shocks and inverse Compton scattering (e.g. Benaglia Zhekov, S. 2012, MNRAS, 422, 1332. & Romero 2003). Initial support for this idea comes from the Zhekov, S. e. a. 2014, ApJ, 785, 8. reported Fermi-LAT -ray detection of the WR binary 2 Vel by Pshirkov (2016). Although this report is encouraging, the detection significance (6) is relatively low and other WR binaries in the Pshirkov (2016) sample were undetected by Fermi. Improvements in the sensitivity and angular resolution