Документ взят из кэша поисковой машины. Адрес оригинального документа : http://star.arm.ac.uk/preprints/2015/671.pdf Дата изменения: Tue Jul 7 12:59:05 2015 Дата индексирования: Sun Apr 10 02:37:06 2016 Кодировка: Поисковые слова: http astrokuban.info astrokuban

Mon. Not. R. Astron. Soc. 000, 1­?? (2002)

Printed 2 July 2015

A (MN L TEX style file v2.2)

K2 and MAXI observations of Sco X-1 - Evidence for disc p re c e s s i o n ?

arXiv:1507.00149v1 [astro-ph.HE] 1 Jul 2015

Pasi Hakala1 , Gavin Ramsay,2, Thomas Barclay3,4, Phil Charles5 1
2 3 4 5

Finnish Centre for Astronomy with ESO (FINCA), VЁisalЁntie 20, University Of Turku, FIN-21500 PiikkiЁ, Finland. a Ёa o Armagh Observatory, Col lege Hil l, Armagh, BT61 9DG, UK NASA Ames Research Center, M/S 244-40, Moffett Field, CA 94035, USA Bay Area Environmental Research Institute, Inc., 560 Third St. West, Sonoma, CA 95476, USA School of Physics and Astronomy, University of Southampton, Highfield, Southampton SO17 1BJ, UK

ABSTRACT

Sco X-1 is the archetypal low mass X-ray binary (LMXB) and the brightest persistent extra-solar X-ray source in the sky. It was included in the K2 Campaign 2 field and was observed continuously for 71 days with 1 minute time resolution. In this paper we report these results and underline the potential of K2 for similar observations of other accreting compact binaries. We reconfirm that Sco X-1 shows a bimodal distribution of optical "high" and "low" states and rapid transitions between them on timescales less than 3 hours (or 0.15 orbits). We also find evidence that this behaviour has a typical systemic timescale of 4.8 days, which we interpret as a possible disc precession period in the system. Finally, we confirm the complex optical vs. X-ray correlation/anticorrelation behaviour for "high" and "low" optical states respectively. Key words: accretion, accretion discs ­ X-ray binaries - X-rays: individual: Sco X-1

1

INTRODUCTION

Sco X-1 is the brightest persistent extra-solar X-ray source in the sky (Morrison 196Ї ) and is a low mass X-ray binary 7 (LMXB), where a Roche-lob e-filling secondary star is losing matter that is accreted by a neutron star via an accretion disc (Charles & Coe 2006). There are numerous studies on this 0.787 d orbital p eriod system over the last fifty years covering almost the entire electromagnetic sp ectrum. It is also predicted to b e a strong source of gravitational waves (e.g. Aasi et al. 2014). As the prototyp e LMXB, Sco X-1 has b een studied for decades at all wavelengths, but esp ecially at optical and Xray wavelengths. Ilovaisky et al. (1980) found that the optical and X-ray flux was well correlated, esp ecially when in a bright state, whilst a longer series of optical observations showed that Sco X-1 changes from a high to low state on a timescale of a few hours (Hiltner & Mook, 1967). McNamara et al. (2003) presented a year-long optical study of Sco X-1 and concluded that the optical observations can b e accounted for by variations in the mass accretion rate. Sco X-1 is unique amongst LMXBs, b ecause to our knowledge, it is the only one showing clearly bimodal optical states (Hiltner & Mook, 1967).

The optical light curves of LMXBs are complex and changes are seen on short time-scales (see for instance Homer et al. 2001 and Hakala et al. 2009). The interpretation of these light curves is made more difficult b ecause of data gaps, the limited duration and cadence of the observations. Much time and resources have b een sp ent in this endeavour (e.g. Shih, Charles & Cornelisse 2011 to name but one). The p ossibility of uninterrupted photometric observations of sources by Kepler has therefore led to dramatic discoveries and results in the field of exo-planets, asteroseismology and accretion physics. Whilst the original Kepler field contained several dozen accreting cataclysmic variables (e.g. Howell et al 2013), it did not contain any known LMXB. Since the loss of two of its reaction wheels, Kepler has b een re-purp osed into K2, and is now making a series of observations of fields along the ecliptic plane, where each field is observed for 70 days (Howell et al. 2014). The fact that these fields go through the Galactic plane allows the study of typ es of ob jects which were not present in the Kepler field. In the Campaign 2 field, Sco X-1 was included as one of the target sources. This pap er presents the K2 observations of Sco X-1 and also simultaneous X-ray data using MAXI. The unprecedented set of optical observations allows us to characterise the optical b ehaviour of Sco X-1 in a way not previously p ossible.



E-mail:pahakala@utu.fi

c 2002 RAS


2

Figure 1. The K2 optical (top panel) and MAXI X-ray (lower panel) light curves of Sco X-1. The data have been normalized to unity and the mean has been subtracted. The error on the optical photometry is negligible and the error on the X-ray points are typically 1 percent.

2 2.1

OBSERVATIONS K2 observations

K2 observations were made of the Campaign field 2 b etween 2014 Aug 23 and 2014 Nov 10 (MJD = 56892­56971). Both Long Cadence (LC) data (30 min) and Short Cadence (SC) data (1 min) were obtained from a 14в12 pixel array centered on Sco X-1. Because of a p ointing adjustment early on in the Campaign, data from the first 2.5 days were omitted, giving a timeline of 71 days. A light curve was made using PyKe software (Still & Barclay 2012)1 which was develop ed for the Kepler and K2 mission by the Guest Observer Office. Although the results shown in this work use 1 min time resolution data i.e. SC data, we find that the 30 min resolution data confirm these results. Because the Kepler satellite lost two of its four reaction wheels, thrusters are used to p eriodically re-saturate the reactions wheels which can then b e used to correct for the drift in the satellite p ointing. This results in a significant movement in the targets p osition on the array on a timescale of 6 hr and 2 days. Since these systematic effects apply to all of the extracted light curves in the K2 field they can b e removed using the method outlined in Vanderburg & Johnson (2014). (See Fig 5 of this pap er to see an example of

removing this correlated noise from a light curve). sulting optical (4370-8360°) light curve of Sco X-1 A is unprecedented in the field of optical monitoring binaries. Given the brightness of Sco X-1 (R 12.4) tometric error on each K2 p oint is negligible. 2.2 MAXI observations

The re(Fig. 1) of X-ray the pho-

The MAXI all-sky monitor on the International Space Station allows for the detection and monitoring of bright X-ray sources over the entire sky in the 2­20 keV band (Matsuoka et al. 2009). Data covering the time interval of the K2 observations was downloaded from the the MAXI archive2 giving a total of 750 photometric p oints. Thus, on the average, we obtained one p oint approximately every 100 mins (Fig.1). The typical error on an individual MAXI p oint is less than 1%.

3 3.1

DATA ANALYSIS AND DISCUSSION Optical light curve

The K2 light curve (Fig. 1) immediately reveals that the system has effectively a bimodal optical brightness distribu2

1

http://keplergo.arc.nasa.gov/PyKE.shtml

http://maxi.riken.jp/top c 2002 RAS, MNRAS 000, 1­??


K2 and MAXI observations of Sco X-1 - Evidence for disc precession?

3

6000 5000 LS Power 4000 3000 2000 1000 0 20 LS Power 15 10 5 0 0.1 1.0 Period (d) 10.0
Optical (K2) power spectrum

X-ray (MAXI) power spectrum

Figure 3. The Lomb-Scargle power spectra of the K2 (optical) and MAXI (X-ray) light curves.

Figure 2. The histograms of optical and X-ray fluxes showing the bimodal "low"/"high" state behaviour in the optical.

tion, which we refer to as "low" and "high" states. This is clearly demonstrated by the light curve, as well as by the histogram of optical fluxes (Fig. 2). Earlier studies of the optical flux distribution (eg. McNamara 2003 and the references therein) have yielded more varied results, which we b elieve, has b een due to the uneven coverage of the variety of datasets employed. It is worth noting though, that the 89 h observing campaign of Hiltner & Mook (1967) produced an almost identical brightness histogram that presented here (Fig.2). K2, however, provides us with an unbiased view of the source b ehaviour over 71 days. We find that the source sp ends almost exactly equal amounts of time in b oth "low" and "high" states. If we adopt the mean flux (0.0) as the dividing line b etween the states, then the system sp ends 51 and 49 ±0.3% in "low" and "high" states resp ectively. We have carried out the p eriod analysis of the K2 data using the Lomb-Scargle p ower sp ectrum method (Scargle 1982) since there are some gaps in the data and the MAXI X-ray data is not equally spaced in time. The resulting p ower sp ectra are plotted in Fig 3. The optical p ower sp ectrum shows a clear signal at 0.787 d, which agrees with the rep orted orbital p eriod of the system (Hynes & Britt, 2012). There are other p eaks in the p ower sp ectrum worth commenting on. Firstly, there is another p eak at ab out 4.8 d and a third one at around 20d. Since the length of the observation is 71 d, we find that the 20 d p eak could easily b e produced by red noise effects, even though there is some tentative evidence in the light curve (Fig 1.) that extended p eriods of "high" state might b e separated by 20 days (see the extended 10d long p eriods of "high" state b eginning approximately at times 5, 25 and 45d). However, the next extended "high" state which would have occured at 65d is missing. Although the signature of the 4.8 d quasi p eriod is visible in the K2 light curve of Sco X-1 (Fig. 1) we have examined and derived p ower sp ectra of several dozen sources
c 2002 RAS, MNRAS 000, 1­??

which were observed in the same module and chip as Sco X-1 and had a similar brightness. None show any indication of p eriod around 4.8 d. We conclude that the 4.8 d quasip eriod which we detect in the K2 photometry is intrinsic to Sco X-1. Returning to the 4.8 d p eriod, we can see evidence for transitions b etween the "low" and "high" states on this timescale all through the K2 observation. These transitions are fairly rapid, since they can take place in less than 3 h, considerably less than Sco X-1's 18.9 h orbital p eriod. We interpret the 4.8 d p eak in the p ower sp ectrum as the main duty cycle for these state changes. If we fold the light curve on the orbital p eriod of 0.787 d separately for the "low" and "high" state data (Fig 4.), it is clear that the shap e of the orbital modulation changes. In "high" state the orbital light curve is almost sinusoidal, whilst in the "low" state it b ecomes distinctly non-sinusoidal in app earance. Furthermore, the amplitude of modulation is somewhat reduced in the "low" state. This would imply that the "high" state curve could b e produced simply by the X-ray heating of the inner face of the secondary star. However, the "low" state light curve requires a variable contribution from an accretion disc, either by means of phase dep endent absorption, emission or changing pro jected area of the disc. It is also p ossible that if the disc is warp ed out of the orbital plane, it could partly shield the secondary star from the X-ray irradiation, thereby diminishing and/or skewing the X-ray heating effects. 3.2 The Optical/X-ray correlation

McNamara et al. (2003, 2005) demonstrated that the optical and X-ray emission in Sco X-1 is anti-correlated when the system is in the "low" state and correlated when in the "high" state. They also showed that the accretion rate and B magnitude of Sco X-1 are closely related for most of the optical variability range. Our analysis of K2 and MAXI data confirms this. We have binned the optical and X-ray p oints into 200 common time bins and show our correlation results in Fig 5. Whilst the correlation b ehaviour does not app ear


4

Figure 4. The over the orbital clarity (the zero proximately 900

"high" and "low" state data folded separately period into 50 phase bins and plotted twice for phase is arbitrary). Each phase bin contains apdata points.

to b e strong, the rank correlation analysis reveals (carried out separately for the "low" and "high" state p oints) that the -0.304 anticorrelation (Kendall's tau) for the "low" state p oints has a chance probability of 9.4в10-6 . Similarly, for the "high" state p oints, we obtain Kendall's tau of 0.305 and a chance probability of 5.4в10-6 . We do not see any evidence for the optical and X-ray fluxes b eing correlated at the very lowest level of optical emission as suggested by McNamara et al. (2003). It is clear from Fig. 1 that the X-ray emission is increased and more stable during most of the optical minima. In order to demonstrate this further, we have plotted the data from the last minimum on a larger scale in Fig. 6. Evidently the optical data also show much larger variability than the X-ray data during the minimum. For some reason, the level of X-ray emission seems to b e the same during several optical minima (Fig. 1) and whenever the X-ray emission reaches a level higher than this, it b ecomes more unstable and starts flaring. It is therefore p ossible that the X-ray flux during the optical minima could represent the Eddington limit i.e. when the system passes from normal branch to the flaring branch in the X-ray colour-colour Z-diagram (van del Klis 1989). This is, however, not in agreement with McNamara et al. (2005), where they show that the accretion rate should b e a linear function of B magnitude.

Figure 5. The Optical vs. X-ray correlations. All data (top), "low" state (middle) and "high" state (bottom). We show the data binned into 200 time bins covering the K2 and MAXI observation.

3.3

The Origin of the 4.8d cycle

There are several plausible explanations for the origin of the 4.8 d modulation, some of which we now discuss in detail. It has b een shown (McNamara et al. 2003 and this work) that when the system is in the "high" state, the X-ray and optical flux correlate. This, together with the sinusoidal orbital modulation of the optical light curve, strongly suggests that in the "high" state the disc is flat and probably does not precess considerably. This would mean that the optical orbital modulation is due to the heated face of the secondary (with a constant emission comp onent from the accretion disc). This is supp orted by the SPH simulations of accretion discs in intermediate mass ratio q 0.3 systems (Murray et al. 2000). The mass ratio in Sco X-1 is estimated to b e 0.3 (Steeghs & Casares 2002), which should make the system stable against the 3:1 resonance when it is accreting steadily. However, if
c 2002 RAS, MNRAS 000, 1­??


K2 and MAXI observations of Sco X-1 - Evidence for disc precession?

5

cantly hotter than the surrounding disc in a system like Sco X-1, where the disc is predominantly X-ray heated. If the 4.8 d p eriod is indeed due to the disc precession, it is still not clear how this could drive the "low"/"high" state b ehaviour with such short transition times (less than 3h) from one state to another.

4

CONCLUSIONS

Figure 6. We show the last 5 days of the data shown in Fig. 1 which includes the last optical minimum and highlights the optical X-ray anticorrelation in the "low" optical state.

the accretion rate at L1 drops, it is plausible that the disc might then start precessing (Murray et al. 2000). It has b een suggested that the state changes in Sco X-1 are accretion rate related (Vrtilek et al. 1991, McNamara et al. 2005), in which case, the app earance of the "low" state folded orbital light curves could b e explained as a result of reduced accretion rate plus precession (either through changing disc area or changing shadowing of the secondary from the X-rays). However, this raises two serious complications. Firstly, what drives the accretion rate to change on a time-scale less than 3 hours, as shown by our data? Furthermore, the large scale disc structure should not b e able to react dramatically to the accretion rate changes in less than 1/6 of an orbital cycle. Secondly, echo mapping of Sco X-1 has revealed that the accretion disc itself, not the heated face of the secondary, is the prime source of optical continuum emission, whilst Bowen blend emission lines are reprocessed on the secondary (Munoz-Darias et al. 2007). ~ In general, the optical and UV emission lines can b e produced b oth in the disc and in the heated face of the secondary (Steeghs & Casares 2002, Boroson, Vrtilek & Raymond 2014). The cross-correlation analysis of HST UV data with RXTE data (Kallman et al. 1998) produced somewhat mixed results with the UV continuum and lines displaying inverse b ehaviour in relation to the X-ray data. The second p ossibility for the 4.8 d p eriod is disc precession. Assuming that 4.8 d is indeed the disc precession p eriod, and given the known 0.787 d orbital p eriod, then this would imply a b eat/sup erhump p eriod of 0.94 d, assuming prograde precession of the disc. This would further imply a sup erhump p eriod excess =0.20 which, although rather large, is still broadly compatible with that obtained from the SPH simulations for q =0.3 (=0.16, Murray et al. 2000). However, there is no clear sign of the exp ected 0.94 d p eriod in the p ower sp ectrum. One explanation for this could b e that the sup erhump p eriod is thought to arise in CVs as a result of the changing free fall length/p otential well depth of the stream and disc impact p oint (Whitehurst & King 1991). Whilst this is the case in CVs where the disc is viscously heated, the impact p oint might not b e signific 2002 RAS, MNRAS 000, 1­??

The K2 observations of Sco X-1 have given the X-ray binary community a virtually uninterrupted and unprecedented optical light curve of a LMXB covering more than 70 days. It shows the detailed investigation of the complex relationship b etween the optical and X-ray flux in a way not previously p ossible. Furthermore, K2 data has clearly demonstrated that Sco X-1 has bimodal optical states with rapid (<3 h) transitions from state to state. There is also evidence that these transitions p ossibly occur p eriodically with a p eriod of 4.8 d. As earlier studies have linked the optical brightness strongly with the accretion rate (McNamara et al. 2005), we conclude that the accretion rate could vary with a p eriod of 4.8 d. This, in turn, could b e due to the precession of the accretion disc at such a p eriod. Further observations of the accretion disc structure (eg. Doppler mapping) over this 4.8 d cycle are encouraged to verify the p ossible p eriodic changes in the disc geometry.

5

ACKNOWLEDGMENTS

Funding for the K2 spacecraft is provided by the NASA Science Mission Directorate. The data presented in this pap er were obtained from the Mikulski Archive for Space Telescop es (MAST). This research has made use of the MAXI data provided by RIKEN, JAXA and the MAXI team. Our work has made use of PyKE, a software package for the reduction and analysis of Kepler and K2 data. This op en source software pro ject is develop ed and distributed by the NASA Kepler Guest Observer Office. Armagh Observatory is supp orted by the Northern Ireland Government through the Dept Culture, Arts and Leisure.

REFERENCES
Aasi, J., et al., 2014, PhRvD, 91, f2008 Boroson B, Vrtiled S.D., Raymond J., 2014, ApJ, 793, 17. Charles P.A., Coe M., 2006, in "Compact Stellar X-ray Sources" eds. Lewin & van de Klis, CUP, p215. Hakala P., Hjalmarsdotter L., Hannikainen D.C., Muhli P., 2009, MNRAS, 394, 892. Hiltner W.,A., Mook D.E., 1967, ApJ, 150, 851. Homer L., Charles P.A., Hakala P., Muhli P., Shih I.-C., Smale A.P., Ramsay G., 2001, MNRAS, 322, 827. Howell, S. B., et al., 2013, AJ, 145, 109 Howell, S. B., et al., 2014, PASP, 126, 398 Hynes R.I., Britt C.T., 2012, ApJ, 755, 66. Kallman T., Boroson B., Vrtilek S.D., 1998, ApJ, 502, 441. Ilovaisky, S. A., Chevalier, C., White, N. E., Mason, K. O., Sanford, P. W., Delvaille, J. P., Schnopper, H. W., 1980, MNRAS, 191, 81 Matsuoka M., et al., 2009, PASJ, 61, 999. McNamara B. J., et al., 2003, AJ, 125, 1437.


6
McNamara B. J., Norwood J., Harrison T. E., Holtzman J., 2005, ApJ, 623, 1070. Morrison, P., 1967, ARAA, 5, 325 Munoz-Darias T., Martnez-Pais I. G., Casares J., Dhillon V. S., ~ Marsh T. R., Cornelisse R., Steeghs D., Charles P. A., 2007, MNRAS, 379, 1637. Murray J.R., Warner B., Wickramasinghe D.T., 2000, MNRAS, 315, 707. Scargle J.D., 1982, ApJ, 263, 835. Shih, I. C., Charles, P. A., Cornelisse, R., 2011, MNRAS, 412, 120 Steeghs D., Casares J., 2002, ApJ, 568, 273. Still, M., Barclay, T. , 2012, Astrophysics Source Code Library, record ascl:1208.004, http://adsabs.harvard.edu/abs/2012ascl.soft08004S Vanderburg, A., Johnson, J. A., 2014, PASP, 126, 948. van der Klis M., 1989, in "ESA, The 23rd ESLAB Symposium on Two Topics in X Ray Astronomy. Volume 1: X Ray Binaries", p203. Vrtilek S.D., Penninx W., Raymond J.C., Verbunt F., Hertz P., Wood K., Lewin W.H.G., Mitsuda K., 1991, ApJ, 376, 278. Whitehurst R., King A.R., 1991, MNRAS, 249, 25.

c 2002 RAS, MNRAS 000, 1­??