Planet Hunters Talk

KOI 3204

  • Shellface by Shellface

    So last Friday, this paper went up on arxiv. It's a good read, but isn't exactly groundbreaking. However, one of the almost-validated KOIs caught my attention:

    KOI 3204: This is a hot (7338 K) dwarf star with a single planet candidate in a 0.57-day orbital period. Speckle imaging shows this KOI to be a single star, helping confirm the hot planet candidate’s properties, which are calculated here to be Rp = 1.01R⊕ and Teq = 3268 K. The planet is validated at a level of 98.5% based on the speckle images.

    Teff = 7338 K is A9-F0, which makes this (to my knowledge) the hottest star with an ultra-short period (USP) planet candidate. As you should know from cases such as Kepler-10 and Kepler-78 these are very valuable as, due to their extreme equilibrium temperatures, they have large out-of-transit phase variations and secondary transits.

    Anyway, I downloaded the Kepler data to see what the phase curve contains. A quick look shows that it is pretty much flat, which is what you would expect for such an early-type star as they are mostly devoid of solar-type oscillations and spots. This makes analysing the planet candidate easier, at least. I phased the lightcurve and then binned it into 200 parts, and this was the result:

    There are a few things that can be immediately gleaned from this;

    • The candidate spends about 30% of its orbital period in transit, which is around what would be expected (R* = 1.6 R☉, P = 0.573 d)
    • The secondary transit is not large, and is at most comparable in size to the errors (~13 ppm)
    • Phase variations are present, but are ~150° off expectations

    This strikes me as odd. If, as one would expect, the phase variations are due to reflected light, then they should be at maximum at secondary transit, but they are close to minimum here. Assuming a planetary model, this implies that the brightest point on the planet is on the night side, which does not make much sense.

    However, we assume that the phase variations are due to more "normal" reflected light, then the body they reflect from is small (due to the small dip at minimum phase variation) but emits a large amount of light (due to the large dip at maximum phase variation), which is vaguely similar to the characteristics of a white dwarf. Indeed, this phase curve is somewhat similar to that of the two stars seen in van Kerkwijk et al. (2010), which are both A-dwarfs transited by and eclipsing hot white dwarfs. The problem with this is that all signals involved here are rather small (~10 times smaller than the referenced cases) which would require the invocation of blended light, but the very paper that made me interested in this system did not detect any stellar companions.

    So in all, I'm confused. This lightcurve shows promising indications (of not being planet-induced), but the implied solution is problematic itself. Can anybody add to this mess?

    Posted

  • ajamyajax by ajamyajax

    Re: KIC 11456279 KOI 3204

    Well I don't know, but maybe "just a complex periodic variable star" can't be ruled out here. Based on the pattern I get in my unflattened fold anyway (Epoch=121.3356, Period=0.5721963, Duration=2.136 hours). Perhaps the slanted look in the LC here is a sign of subtle contamination (from a nearly aligned background EB of a similar spectral type), but only tried a few different options there. Also looks like TTV/ETV could be in the transits if real, which might support a white dwarf if that is microlensing as I understand it anyway. But also looks like undersampled data here which complicates pinning those values down, in my opinion.

    Interesting KOI indeed though. Thank you for posting your expert thoughts here.

    p1

    Posted

  • ajamyajax by ajamyajax

    p.s. I also read that same paper with interest because they looked at KOI 2311 which I also worked on (my ph link here). I was pleased about their blending review, but also a bit frustrated because they ignored a possible third object -- that was on the TCE list, now on the KOI list -- as others did.

    Really no big deal, just think this object makes a difference there. Have a good one.

    Posted

  • Shellface by Shellface

    I notice you are illustrating the SAP_FLUX of the star. Looking at the normalised raw lightcurve, the dominant variation appears to repeat in nature every 4 seasons, which makes me believe it is an instrumental effect (I remember it's something to do with telescope rolls?). What I downloaded was PDCSAP_FLUX, which (when normalised) is far smoother. Whether this is representative of the actual photometry is not something I can say, but this lightcurve is more useful here.

    Anyway, I've thought some more about this. Running with the companion WD idea, the issue that the amplitude of variation is ~10 times too small implies that the star being transited contributes ~10% of the system flux, which in turn implies that the star being transited is ~1 magnitude fainter than the bright component. This is problematic because the aforementioned paper precludes such bright companions at seperations beyond ~0.05". However, as the system lies ~800 parsecs away, such bright companions are possible at projected separations below 40 AU, which is vaguely physically viable.

    ΔV ~ 1 mag for an F0V primary indicates that the star being transited has a spectral type of F5V, assuming it is on the main sequence. This strikes me as peculiar for a star with a close white dwarf companion (most such stars seem to be M-dwarfs or A-type), though I expect that this is not particularly meaningful.

    Ultimately, I expect a spectrum of the visible star will be able to determine the nature of this system. If it is a ~F0V/F5V binary (where the secondary is orbited by a white dwarf), then the two bright components should be discernable because stars of such spectral types have very different rotational velocities. This would thus give a composite spectrum with a "broad" feature, and a "narrow" feature that varies with a period of 0.57 days. If something like this is not observed, then I am thoroughly stumped as to the nature of this system!

    "Interesting KOI indeed though. Thank you for posting your expert thoughts here."

    I certainly wouldn't say I'm an expert - I'm simply loosely familiar with the relevant physics for this situation.

    Posted

  • ajamyajax by ajamyajax

    Well I'm just a PH'er, but saw this comment on cfop (3204) and then found a paper quote that might support your WD theory:

    bieryla
    2013-06-15 14:06:58
    "The classification shows that this is a very rapid rotator with Vrot=140km/s." ...

    and from this study mentioning a white dwarf companion to Regulus:

    "It has been known for a long time that accretion from a companion star can result in very high rotation rates for the accretor."

    from:
    "The Past and Future History of Regulus"

    http://arxiv.org/pdf/0904.0395v1.pdf

    p.s. and maybe a thin accretion disc could explain the increased flux seen during some of these possible transit eclipses?

    Posted

  • Shellface by Shellface

    Ah, now that is interesting - stars of spectral type ~F0 typically have rotational velocities considerably less than 140 km/s! For, say, v sin i = 140 ± 5 km/s and R* = 1.59 ± 0.20 R☉ (or so), the equation:

    Protmax = R/(0.0198*vsini)

    gives the rotational period of the star to be ≤0.57 ± 0.10 days, where the highest value is for i = 90° (and, thus, approximately aligned with the orbital inclination of the transiting object). This strongly implies the primary is the object being orbited, and evidences against the necessity of another component.

    I expect that the source of this v sin i was a single spectrum. This pretty much disproves my previous expected result, but doesn't help to constrain the parameters of the companion much. For a minimum sensible WD mass of ~0.1 M☉, the minimum RV amplitude should be 15-20 km/s, which is probably significantly detectable even though the star is unfavourable (faint, hot, extremely fast rotator). Such measurements would probably be enough to characterise the entire system.

    As for the nature of the out-of-transit variations, their relation to the rotational period of the star must now be considered. Though rotational synchronisation does facilitate mass transfer, it also requires one of the stars (here, the primary) to be overflowing its roche lobe, and thus requires it to be evolved. This is certainly possible given the errors on the photometric log g, but it is not the only possibility and not one I can comment much further on as I have little knowledge of accretion disks.

    • If the "deep" event is the WD transit, then the phase variations may respond to a brightening of the stellar disk due to the companion. I'm… not sure what that means.

    • If the "shallow (unobserved) event is the WD transit, then the phase variations may respond to a darkening of the stellar disk due to the companion. This is more or less consistent with a starspot that is trailing the WD's orbit.

    Both scenarios indicate the star's "response" occurs approximately 1 hour after the companion passes above that part of the disk, and also implies that the companion's orbit is prograde. Note that this is ignoring any other contributions to out-of-transit variations, particularly those caused by the companion.

    This KOI does seem like it deserves follow-up, even if it is a planetary false positive. The model described here is at least partially satisfactory, with the most glaring issue being that the ratio of primary/secondary transit depths is peculiar regardless of which is due to the actual crossing of the WD. Perhaps this can be addressed with a more numerical model.

    Posted