In an article published in Sky & Telescope back in December 1998, I made the prediction that the photometric detection of moons orbiting extrasolar planets was probably decades away (see “Habitable Moons”). After 16 years (just four years short of it being “decades”), I was beginning to think that I might have to eat those words with the possible detection of an “exomoon” orbiting a planet called Kepler 90g… and gladly so! While for many years I have been interested in the concept of habitable moons orbiting extrasolar planets, the discovery of any moon (habitable or otherwise) associated with an extrasolar planet would have a profound impact on our understanding of the planet formation process and the habitability of other worlds. The discovery in 2013 of a photometric “anomaly” associated with a transit of Kepler 90g offered the best candidate for the detection of a exomoon to date.
Background
Kepler 90 (also known as KIC 11442793 or KOI 351) is probably a late F or early G-type dwarf star with about 1.8 times the luminosity of the Sun some 2,500 light years away. In November 2013, a paper by Cabrera et al. was submitted for publication in The Astrophysical Journal describing the detection of a compact system of seven planets ranging from super-Earths to gas giants orbiting within about 1.1 AU of this star. One of the planets described by Cabrera et al. included the second outermost planet that was found, Kepler 90g.
Kepler 90g is in a 211-day orbit about 0.7 AU from its sun with a radius of about 8 times that of the Earth (or RE). While the mass of these planets has yet to be determined, dynamical arguments limit the maximum mass of Kepler 90g to be less than that of Saturn. If it were any more massive, this system of planets would be rendered unstable. Strong interactions between Kepler 90g and h (the outermost known planet with a radius of 11 RE) generated observed transit timing variations (TTV) of as great at 25.7 hours in the former leaving open the future prospect of deriving the masses of these outer two planets by the TTV technique when more data become available and if radial velocity measurements prove to be insufficiently sensitive. Since Kepler 90g receives about 3½ times the stellar flux of the Earth, it is unlikely to be habitable regardless of its mass or composition.
Kepler’s light curve for the third observed transit of Kepler 90g. The solid curve is the expected 12.6 hour long transit signature without transit timing variations (TTV). The arrow to the right of the main transit event indicates the apparent dimming of Kepler 90 that could have been the result of a distantly orbiting moon of Kepler 90g. (Cabrera et al.)
A total of six transits of Kepler 90g were observed during Kepler’s primary mission with one expected transit missed due to an interruption in its photometric data. In addition to the TTV affecting the timing of the transit, a dimming was noted about 21.5 hours after the third observed transit of Kepler 90g that was not noticed during the other five opportunities. Cabrera et al. very cautiously speculated that this dimming might have been caused by a moon of Kepler 90g in a distant orbit near the edge of the Hill radius – the maximum distance a moon can orbit stably around its primary. Despite the issues, this one unexplained photometric “anomaly” was the best candidate for an exomoon detection to date and deserved additional attention.
New Results
The Harvard University-based project, The Hunt for Exomoons with Kepler (HEK), has been set up to search Kepler’s photometric data base for the telltale signs of moons orbiting extrasolar planets. A group led by HEK principal investigator, David Kipping (a Donald Menzel fellow at the Harvard College Observatory), examined the case for a moon orbiting Kepler 90g in order to determine if it was the bona fide exomoon candidate or the result of some other effect.
Kipping et al. found that based of the depth of the transit that this potential moon, designated Kepler 90g.01, was about 1.9 RE making it a super-Earth in size (and, based on analysis of other Kepler data, quite likely a mini-Neptune in composition – see “Habitable Planet Reality Check: Terrestrial Planet Size Limit”). The semimajor axis of the orbit of Kepler 90g.01 must be about 3.5 million kilometers with an orbital period greater than about ten days. In order for Kepler 90g.01 to have a stable orbit around Kepler 90g with such a large semimajor axis, the latter must have a mass of at least twice that of Neptune. Corresponding to a mean density of 0.39 g/cm3, this is a physically plausible lower limit value for such a planet.
Kipping et al. then subjected the photometric data of Kepler 90g to a number of conventional diagnostic tests and the candidate exomoon transit event passed them all. However, an examination of the pixel-level light curves using a new technique the HEK team dubs “transit centroid analysis” found that the purported exomoon transit signal was spread evenly among nearly all of the target’s pixels instead of being concentrated in the “flux centroid” (i.e. the center of the signal weighed by the signal-to-noise ratio or SNR). This lock-step change in brightness in all the target pixels independent of the SNR and the deviation of this observed behavior from simulations strongly suggests that the signal is spurious resulting from an instrument effect. Kipping et al. suggest that perhaps the culprit is a Sudden Pixel Sensitivity Dropout (SPSD) – a short-lived decrease in the sensitivity of groups of pixels in Kepler’s detector resulting from an interaction with a cosmic ray. Regardless of the exact cause of the issue, Kipping et al. conclude that the observed event is a false positive and that Kepler 90g.01 does not exist.
One of the lessons from this episode is that great care must be taken in establishing the existence of an exomoon. As the case of Kepler 90g.01 shows, even a candidate event that passes all of the conventional tests of a bona fide transit can turn out to be the a false positive resulting from a subtle instrumental effect. Despite the extreme difficulty facing exomoon hunters, Kipping et al. conclude in their paper “despite the mine-field of false positives, a confirmed detection is surely inevitable”. Nothing would please me more than if this prediction comes true within the next couple of years in defiance of my prediction of 1998.
Follow Drew Ex Machina on Facebook.
Related Reading
“Habitable Moons”, Sky & Telescope, Vol. 96, No. 6, pp. 50-56, December 1998 [On line version]
“Habitable Moons: Background and Prospects”, Centauri Dreams, September 19, 2014 [Post]
“Habitable Moons: A New Frontier for Exobiology”, SETIQuest, Vol. 3, No. 1, pp. 8-16, First Quarter 1997 [Article]
“Habitable Planet Reality Check: Terrestrial Planet Size Limit”, Drew Ex Machina, July 24, 2014 [Post]
General References
J. Cabrera et al., “The Planetary System to KIC 11442793: A Compact Analogue to the Solar System”, The Astrophysical Journal, Vol. 781, No. 1, Article id. 18, January 2014
David M. Kipping et al., “The Possible Moon of Kepler 90g is a False Positive”, arXiv: 1411.7028 (Accepted for publication in The Astrophysical Journal Letters), November 25, 2014 [Preprint]
Hi Drew, (from one LePage to another).
This was really interesting. Great write up!
I guess one could imagine detecting a moon through a planetary wobble when you have enough transits, even if the moon itself did not transit the star (or did not transit separately from the planet).
Cheers, Mike