A Holy Grail for stellar wind A Holy Grail for stellar wind analysis analysis zeta Puppis seen by zeta Puppis seen by

XMMXMM

Y. Nazé (FNRS-ULg), G. Rauw (ULg), Y. Nazé (FNRS-ULg), G. Rauw (ULg), L. Oskinova (U. Potsdam), E. Gosset (FNRS-L. Oskinova (U. Potsdam), E. Gosset (FNRS-

ULg), A. Hervé (ULg), C.A. Flores (U. ULg), A. Hervé (ULg), C.A. Flores (U. Guanajuato)Guanajuato)

Zeta PuppisZeta Puppis

One of the closest One of the closest (335pc), earliest (O4I), (335pc), earliest (O4I), and brightest massive and brightest massive starsstars

Many intriguing Many intriguing properties : runaway star, properties : runaway star, chemical enrichment, fast chemical enrichment, fast rotation (post RLOF+SN ? rotation (post RLOF+SN ? Ejection from cluster ?)Ejection from cluster ?)

The first one observed by The first one observed by Chandra & XMM Chandra & XMM (Kahn et al. (Kahn et al. 2001, Cassinelli et al. 2011)2001, Cassinelli et al. 2011)

A decade of XMM A decade of XMM observationsobservations

18 observations18 observations

Data reductionData reduction

The best dataset available for a massive star The best dataset available for a massive star (~1/2 Ms for EPIC, 3/4 Ms for RGS)(~1/2 Ms for EPIC, 3/4 Ms for RGS)

18 observations taken in different modes (timing, full 18 observations taken in different modes (timing, full frame, large window, small window), with different frame, large window, small window), with different filters (medium, thick), sometimes off-axisfilters (medium, thick), sometimes off-axis

Bright Bright slight pile-up (~limit of large window mode) slight pile-up (~limit of large window mode)

Extraction with pattern=0, keeping the same circular Extraction with pattern=0, keeping the same circular regions for source and bkgd (regions for source and bkgd (NB: annular source = NB: annular source = KO!KO!))

New RGS pipeline (SAS 10) solved the flux/shift issuesNew RGS pipeline (SAS 10) solved the flux/shift issues

Variability of zeta PuppisVariability of zeta Puppis

In optical : In optical : long-term changes long-term changes (Conti & Niemela 1976)(Conti & Niemela 1976) 5d variations 5d variations (e.g. Moffat & Michaud 1981)(e.g. Moffat & Michaud 1981) a few h pulsations a few h pulsations (e.g. Reid & Howarth 1996)(e.g. Reid & Howarth 1996)

In X-rays : In X-rays : Einstein - nothingEinstein - nothing ROSAT revealed a small modulation (2% ROSAT revealed a small modulation (2%

amplitude) of 17h period in 0.9-2 keV band amplitude) of 17h period in 0.9-2 keV band (Berghöfer et al. 1996)(Berghöfer et al. 1996)

Chandra, XMM (1 dataset) – nothing Chandra, XMM (1 dataset) – nothing (Kahn et al. 2001, Oskinova et al. 2001)(Kahn et al. 2001, Oskinova et al. 2001)

Variability: Variability: the XMM the XMM

viewview Several energy bandsSeveral energy bands Several time bins (200s Several time bins (200s

to 5ks EPIC, 500s to 10ks to 5ks EPIC, 500s to 10ks RGS)RGS)

Chi-2 tests (constant, Chi-2 tests (constant, line, parabola); Fourier; line, parabola); Fourier; AutocorrelationAutocorrelation

Results :Results : Background is variableBackground is variable Instruments do not agreeInstruments do not agree

Variability: short & mid-Variability: short & mid-termterm

The longer you The longer you observe or the observe or the longer the time bin, longer the time bin, the more variable it the more variable it is is no obvious short no obvious short term variations but term variations but mid-term ones exist mid-term ones exist (with timescales > (with timescales > Texp : rotation ?)Texp : rotation ?)

Variability: Variability: short and mid-short and mid-

termterm

For the best data For the best data (small window, thick filter)(small window, thick filter) FourierFourier

0.3-0.4/d + ~1/d ?0.3-0.4/d + ~1/d ? Rotation (5d) ????Rotation (5d) ????

AutocorrelationAutocorrelation >0 if T<20ks, <0 @ 55ks >0 if T<20ks, <0 @ 55ks wind flow time ~5kswind flow time ~5ks

Not very significant anywayNot very significant anyway

Best case : pn dataBest case : pn data

Variability: linesVariability: lines

RGS dataRGS data TVS : nothingTVS : nothing Count rates and ratios : Count rates and ratios :

nothingnothing Comparison with average Comparison with average

spectrum by eye : non-spectrum by eye : non-significant variations may significant variations may exist but similar to exist but similar to optical…optical…

VariabilitVariability: long-y: long-

termterm

EPIC, RGS : EPIC, RGS : decrease !decrease !

NB : Fourier, NB : Fourier, autocorrelation and autocorrelation and relative dispersions relative dispersions calculated after calculated after detrendingdetrending

Variability: long-termVariability: long-term

EPIC spectra : fitted by EPIC spectra : fitted by tbabs (ISM, fixed) tbabs (ISM, fixed) * sum of 4 thermal comp. * sum of 4 thermal comp. (vphabs*vapec – Nh and (vphabs*vapec – Nh and kT fixed)kT fixed)

Pile-up affects all data Pile-up affects all data taken with medium filtertaken with medium filter

Formally unacceptable Formally unacceptable fitting but missing physics fitting but missing physics and disagreement and disagreement between instrumentsbetween instruments

Flux appears quite Flux appears quite constant (a few % constant (a few % decrease?) decrease?) count rate variations count rate variations come from detector come from detector sensitivity changessensitivity changes

A simple modelA simple model

Features Features (Oskinova et al. (Oskinova et al. 2004)2004) smooth wind or smooth wind or

random absorbersrandom absorbers random emittersrandom emitters solid angle conserved, solid angle conserved,

outward motion (beta outward motion (beta law)law)

LCs less variable:LCs less variable: at high Eat high E for smooth windfor smooth wind For more For more

emitting/absorbing emitting/absorbing clumpsclumps

How to compare with How to compare with data?data?

Relative dispersions calculated for each Relative dispersions calculated for each observation observation for full LCsfor full LCs for resampled LCsfor resampled LCs

Poisson noise !Poisson noise ! Relative dispersions in both cases ~ Poisson Relative dispersions in both cases ~ Poisson

statistics !statistics !

If additional variability exists, it is hidden in If additional variability exists, it is hidden in noise… hence its amplitude is small, and noise… hence its amplitude is small, and emitting/absorbing clumps are many (>10emitting/absorbing clumps are many (>1055) !) !

Global spectral fitting : Global spectral fitting : preliminary resultspreliminary results

a a (Kahn et al. 2001, (Kahn et al. 2001, Cassinelli et al. 2011)Cassinelli et al. 2011)

Line fitting : preliminary Line fitting : preliminary resultsresults

Line profile Line profile fitting using fitting using Owocki & Cohen Owocki & Cohen models : models : variation of tau variation of tau with wavelength with wavelength (cf. Cohen et al. 2010, in (cf. Cohen et al. 2010, in green – NB with resonance green – NB with resonance scattering in red)scattering in red)

BUT /!\ uniqueness BUT /!\ uniqueness of solution…of solution…

ConclusionsConclusions

A decade of XMM observations = best dataset !A decade of XMM observations = best dataset ! VariabilityVariability

Only noise on short-termOnly noise on short-term Trends on mid-term (DACs ? But no link with rotation, cf. Fourier)Trends on mid-term (DACs ? But no link with rotation, cf. Fourier) Long-term decrease due to detector degradationLong-term decrease due to detector degradation Comparison with models : a lot of wind parcels needed !Comparison with models : a lot of wind parcels needed !

Lines Lines Multi-temperature neededMulti-temperature needed Typical optical depth varies with wavelengthTypical optical depth varies with wavelength

For the future…For the future… Follow the star over its rotation periodFollow the star over its rotation period Observe it with more sensitive detectors to decrease Poisson Observe it with more sensitive detectors to decrease Poisson

noisenoise Develop more detailed modelsDevelop more detailed models