Cumulative  Deviation of data & model scaled  to 0.3  99%  90%  95% HD 36861J (rp200200a01) Probability of Variability A Large ROSAT Survey of X-Ray Time Variability in O Stars Allison S. Adelman 1, David H. Cohen 2 (1) Bryn Mawr College, (2) Swarthmore College THEORY O star X-ray emission comes from shock-heated gas present in their stellar winds; for B stars, the situation is more uncertain, and their X- rays may be related to magnetic fields, at least in some cases. The same may be true for certain O stars too. Unstable mass flow driven by the radiation field of the star produces strong shocks as faster wind material impacts on slower material (Owocki, Castor, & Rybicki 1988). Shocks occur stochastically, so X-ray output should be time variable. X-ray variability should be due to evolution of individual shock structures and due to time variability in the number of individual shock zones in the wind.  The X-ray variability properties (timescales and spectral properties, as well as amplitude of variability) should tell us about the properties and underlying physics of wind shocks. At the simplest level, the amplitude of X-ray variability is proportional to the square root of the number of individual shocks. Two Sample Images of ROSAT fields Analysis Procedure Data A pointed observation of O star HD57060 O star HD57061 is also present, but overlapping another point source, smeared out near the edge of the image The support structure of the telescope is clearly visible ObsID rp200066a02 HD HD HD A pointed observation of a young open cluster of O stars (NGC 6231) 3 O stars are present in the central cluster; however many were too indistinguishable to extract data 2 other O stars with good data are present in the field, as well as 1 more that is smeared out at the edge of the image Sample Light Curves of O Stars HD A Representative Non-Variable Star HD A Representative Variable O Star Background O Star X-Rays Project We searched the ROSAT PSPC archives of pointed observations for O stars, observed both intentionally and serendipitously Our sample includes 60 O stars in 86 separate observations, including many O stars not previously reported on in the X-ray literature. We extracted source counts for each O star, and performed several types of time variability analyses on each object HD mean = bins = 12 Reduced  2 = 0.80 P= 36% st. dev. = K-S Probability = HD36861J mean = bins = 20 Reduced  2 = 0.93 P = 45% st. dev. = K-S Probability = Examples of K-S Test Results  Cumulative  Deviation of data & model scaled to 0.3  99%  90%  95% HD (rp200199a00) Probability of Variability We ran  2 tests of the null hypothesis (of a constant count rate). And we also calculated the one-sided K-S statistic. For longer pointings, which were broken up into several observations, we calculated the K-S statistic for each observation. If the K-S test gave a positive result (>90% probability of variability) we tested the hard (E>0.5keV) and soft (E<0.5keV) spectral ranges separately. OBSERVATIONS Very few large, systematic studies of X-ray variability among O stars. It is generally accepted that O star X-ray emission is not variable, but this has not been thoroughly quantified. A few individual cases of X-ray variability have been detected:  1 Ori C: periodic modulation: a young magnetic rotator.   Pup: Very long observations have turned up very low level (~2%) periodic (P=18h) variability correlated with H  variability.  Ori: one-time brightening of ~15%: possibly also magnetic in origin.  Ori and Cyg OB2-8 detected with Einstein; also some interacting binaries.  No systematic survey of O star X-ray variability has been carried out using the largest, high-sensitivity database of X-ray observations: The ROSAT archive. There are numerous sites of X- ray emission on the sun. Each one evolves in time, as does the overall distribution, leading to significant X-ray variability. UV observations of O star winds show small- and large-scale variability; often periodic but also stochastic. ObsID rp200112n00 HD57061 HD57060 Linear Fits to Variable Stars O StarK-S Probability K-S Prob. for Soft Energy Channels K-S Prob. for Hard Energy Channels HD HD HD HD HD HD HD HD37742J (rp200198a00) HD37742J (rp900386n00) HD HD Hard and Soft Energy Channel K-S Probability Results Bright stars in the spectral range earlier than about B3 are soft X-ray sources, with L X ~ L Bol Note: K-S test is generally more sensitive than  2 (can see in lightcurve at the bottom of the previous column too) refers to linear fits 11 of the 17 observations showing variability have non-zero slope of their count rates Conclusions Nearly 30% of the O stars in the sample are variable No dramatic variability (e.g. no flares) Much of it is ‘long’ timescale variability (  > few ksec) Significant amount of hard variability wavelength - velocity Time (days) Numerical radiation hydrodynamic simulations (snapshot at left) show highly time-dependent structure, with shock waves advecting through the wind. Flow timescales and cooling timescales are of order 1000s of seconds. The ROSAT PSPC is a gas proportional counter with a two- degree field of view and some very modest energy resolution (each photon is tagged with an approximate energy, as well as a position and arrival time in the detector) Note: the spatial resolution (FWHM~5”) degrades rapidly off-axis.