From CM Draconis to the Crowded Field BW3: Aspects of the Search for Extrasolar Planets Around Small Eclipsing Binaries

We review some apsects of the search for extrasolar planets around small-mass short-period eclipsing binaries, including lessons from the extensive search around CM Draconis (over 1,000 hours), and the extention of the search to small eclipsing binaries in the crowded field Baade’s Window 3 (BW3) in the Galactic plane. In addition to being of intrinsic interest to planet formation studies, due to a likely more complex angular momentum history than single stars, we outline several aspects of planet detection around eclipsing binaries that also make them attractive targets for ground-based searches. Advantages of Eclipsing Binary Searches CM Draconis is the smallest known eclipsing binary system consisting of two M4 main-sequence stars of total mass 0.444 Ms (solar mass), and a total area about 12% that of the solar disc (Lacy 1977). Immediately this made it a prime candidate for extensively testing the photometric transit detection method as its small cross-sectional area allows the highest differential attenuation for a transiting planet of a given size, while the small total binary mass allows the largest offset by a third-body giant planet by the eclipse timing method (Schneider and Doyle 1995, Deeg et. al. 2000). The advantage of choosing eclipsing binaries for such photometric transit searches include, of course, an expected higher probability of planetary orbital alignment with the observer's line-of-sight, with precessional damping of protoplanetary material— caused by the binary itself—further being expected to constrain the tendency for protoplanetary material to accrete in the binary orbital plane (Schneider and Doyle 1995). Transit events across eclipsing binaries are quasi-periodic as the planet is not so much crossing in front of a star (as in the single star case) as the binary stars are orbiting behind the planet as it moves through the “transit window” (i.e. across the region of the binary barycenter as seen along the observer’s line-of-sight). This quasi-periodicity can have two effects: to generally reduce the duration of individual transits, and to allow such transits to be clearly distinguished from rotational periodicities. Shorter individual transit events (50 minutes is typical for CM Draconis) take place at a lower noise region in the ground-based observational power spectrum than typical single star transits of 5 hours or more—the observational noise at a frequency of 4.8-day is almost 4 times the noise power at 28.8-day (Doyle et. al. 1996, Deeg et. al. 1998, Jenkins et. al. 2000). Close binaries (especially highly convective late-type dwarfs) will rapidly become tidally locked in rotation so that short-term periodicities associated with any starspot activity, for example, should be easily separable from planetary transit events (without the need for multiple-filter bandpass observations ; see Rosenblatt 1971; Borucki and Summers 1984). In addition, this quasi-periodicity lends itself to a modeling of possible transits that can be “matched” with the differential light curve (Jenkins et. al. 1996, Doyle et. al. 2000). By comparing (correlation matching) models of virtually all possible transit events that could have occurred within the light curve, transit events of individual candidates may be isolated and their detection confidence may be quantified, which we will discuss further below. (We note that such detections do not take place in frequency space as the total amplitude attenuations expected by infrequent transits would not have sufficient power in the spctrum to be detected.) Finally, in addition to transit detection, eclipsing binaries are a sort of “clock” in themselves—the binary pair, when offset around a second barycenter (caused by the giant planet), show a periodic delay or advance in the time of occurance of their eclipse minima. Such a second barycenter offset would have to result from a third massive body being in orbit around the binary pair (Doyle et. al. 1998). Thus a precise timing of eclipses can also allow the detection of jovian-mass planets around the stellar system provided the giant planets have sufficiently large semi-major axes (i.e. a few light seconds). The CM Draconis Search Observations and Modeling In order to detect extrasolar planetary transit events at or below the photometric observational noise level (the signal or intrinsic detectability, pi ), as well as to approach the likelihood of a transit event actually being in the data (the coverage or observational detectability, po) a great deal of data must be obtained. To this end we (the Transit of Extrasolar Planets, or TEP Network; Deeg et. al. 1998) observed CM Draconis over the observing seasons from 1994 to 1999. Requring the photometric precision to be better than 0.7% we obtained a total of 1,014 hours of (mostly R-band) photometry on the system. These observations resulted in a light curve (differential magnitude—CM Draconis minus the standard stars in the field—plotted agains Heliocentric Julian Day) with 26,042 points, including 82 eclipse events (Doyle et. al. 2000, Deeg et. al. 2000, Deeg and Doyle 2000). We then generated models for virtually all possible planetary transits that could have occurred across the discs of CM Draconis A and B using a program we call the transit detection algorithm (TDA). Due to the extensive coverage, over 570 million individual models of possible planets were required, each one with a different period and starting phase (the phase of CM Draconis already being known at each observation time). Optimizing the detection limit for the smallest possible planet size we reached a 90% detection confidence for planets of size 3 Re (Earth radii) with periods of 60 days or less (the confidence was 99% for 30 day periods), and a confidence of 50% for planets of size 2.5 Re (80% for periods of 10 days or less). Thus this was the confidence (in the signal detection sense) with which we would have expected to have seen such planet transits at the noise level in our observatons (see Doyle et. al. 2000, for full details). While no outstanding candidates appeared in this search, it is clear that when performing such a search near the observational noise (0.45% photometry translates to a single transit of a planet of size 2.6 Re around CM Draconis) predicted transit events with follow-up observational confirmations are essential. Of the nine top candidates we isolated with the TDA, only one has presently “survived” the follow-up observational tests. In addition, we have precisely determined times of minima of the 82 binary eclipses from 1994 through 1999 and a sub-set of the best of these should allow strong constraints to be placed on jovian-mass planets in orbit around CM Dracoins (essentially regardless of orbital inclination) for periods from about 100 days to over a decade (see Deeg et. al. 2000, for full details). Conclusions From CM Draconis of Bioastronomial Interest CM Draconis may be the first main-sequence system searched for, what may essentially be considered, terrestrial-sized planets. The significant transiting planet detection limit achieved was in the range 2.5 to 3 Re , or about 1% to 2% the size (volume) of Jupiter. The region essentially searched—from 7 to 60 days—was dictated by the nearest possible stable planet orbit (a binary-to-planet semimajor axis ratio of about 1 to 3 being required; Holman and Wiegert 1999) as well as our total data coverage (for multiple transits to be in the light curve). As the CM Draconis system itself has only about 1.03% the total luminosity of the Sun, this region—from 7 to about 35 days—constitutes the region that receives an equivalent amount of stellar energy as the terrestrial planets in our Solar System. (Interestingly enough, the surviving candidate from our 1999 observations has a period of about 22.6-days, giving it an equivalanet insolation about that of Earth. This will be observationally confirmed or ruled out next year.) The circumstellar habitable zone (CHZ) is defined as the region around a star that receives enough stellar flux to accommodate liquid water on the surface of a roughtly Earthlike planet. The inner boundary is conservatively limited by the locale where a moist runaway greenhouse would occur (water is photodissociated in the upper atmosphere) , while the outer boundary of the CHZ can be defined as the locale where CO2 condensation begins—i.e., not only losing its greenhouse warming properties but increasing the planetary albedo by snowing out of the atmosphere (Kasting et. al. 1993). By these criteria, the CHZ around virtually all M-dwarf stars is well within the tidal rotational locking limit for a planet (this is because the luminosity increases with the inverse square of the planet’s orbital semi-major axis while the tidal force increases as the inverse cube of this distance). What then, one may ask, would be the interest in main-sequence M-stars from the viewpoint of possible exobiology, as any planets within the CHZ would be synchronous locked to the orbital period and the atmosphere might be expected to boil off on the stellar side and freeze out on the dark side? Interestingly, however, recent 3-D modeling of synchonously rotating planets have revealed that only about 100 millibars of CO2 would be sufficient to lower the thermal gradient so that liquid water could exist on an Earth-like planet’s surface within the habitable zone of such a synchronously rotating planet (Haberle et. al. 1996). Further, M-stars might allow a proportionately extended inner boundary to their CHZ (perhaps as high as three times the solar inner CHZ flux) as there is essentially no UV radiation from M-stars to initiate the photodissociative escape of water in the upper atmosphere, and therefore no moist runaway greenhouse effect (Joshi et. al. 1998). Finally, as the result of a recent in-depth study of infrared photosynthesizing bacteria as the base of an M-star planetary ecosystem, it would appear that higher-plant habitability might be possible (Heath et. al. 1999). Given that such M-stars (at least in the solar neighborhood) constitute about three-quarters of all stars, such results could be rather encouraging for exobiological studies, as well as allowing CM Draconis itself to remain of relevant interest to the bioastronomical community. Extention to Larger Telescopes and Crowded Stellar Fields We conclude, from the work outlined above, that a successful search for both inner terrestrial-sized and outer jovian-mass planets around small eclipsing binaries may be undertaken with existing technology using the photometric transit method. For stars with larger disc areas, and also of larger masses, the photometric precision must be improved,and this can best be accomplished by going to larger telescopes. Figure 1 shows the improvments in detection one might expect, scaling from results on CM Draconis using the Crossley 0.9-meter. Figure 1. A comparison of times required to reach the stated detection confidence for 2.5 and 3.0 Earth radii transiting planets ( pi ), scaled for telescope aperature size from the Crossley results for CM Draconis. Scaling includes both Poisson statistics and improvement over scintillation with (telescope aperture) (see Dravins et. al. 1998, Eq. 10). Telescope size cannot, of course, mitigate the necessity for thorough period-phase observational coverage ( po ) . 1 10 10