KIC 8462852: Whence The Fading Star

As of this morning we have discovered over 1,500 exoplanets, planets orbiting stars other than the Sun.
In addition to the confirmed discoveries, we have over 3,000 candidate exoplanets, most discovered by the NASA Kepler Mission through transits, where we see the shadow of a planet as it crosses in front of its parent star.
We know, statistically, that most of the candidate planets are real, but a few % are weird misalignments of something else that makes us think there might be a planet there. Filtering those out is hard.

We also see other things, weird and wonderful things...

Little over year ago, I got an email telling me to check out KIC 8462852 - the Kepler Input Catalog entry star number 8426852 - the Planet Hunters, the crowd sourced amateurs who eyeball all the Kepler data, had tagged this star as interesting, and Dr Tabetha Boyajian at Yale and her colleagues were quietly floating what they'd seen to various people because it was really interesting.

Light Curve for KIC8462852 from Boyajian et al  2015 Light Curve for KIC8462852 from Boyajian et al 2015 (click to embiggen)

Interesting it was. The star, an F3 main sequence star, apparently a quiet middle aged normal star, somewhat more luminous than the Sun, is dimmed, but a large amount, 10-20%, for many days, in irregular patterns and in asymmetric ways.

For comparison, the shadows of planets dim stars typically by somewhere between 1% and 0.0001%, or less, and for hours or a day or two, not weeks.

Now, we do see things somewhat like this around very young stars, from huge dusty debris disks from which planets are being made, but those often completely obscure the star, and we see their signature in infrared observations as a huge heat glow, as the disk radiates absorbed star light.
The WISE satellite had observed this star somewhat earlier, and it had no measurable infrared glow.
So no big obscuring disk. It is also not young by any age indicator we know of. (Or is it?)

This is a puzzle, because whatever is obscuring the star has to be local to the star; it is moving slow, but not very slow (probably orbiting between 1 and 10 AU from the star); and it has to be BIG - to obscure the star to that extent we are either talking about opaque objects that are several hundred thousand kilometers across - several times bigger than Jupiter - or we are talking translucent clouds that are bigger than the Star itself, more than a million km across.
Really nothing else will work.
Planets don't come that big.
A star that big would be very easily visible.

Most anything else is ruled out by various other observational constraints.

This is fun, it is what most scientists live for - the unexpected mystery, the open puzzle, something new!

What made this particularly interesting is that Jason Wright and collaborators were at the same time working on the Ĝ - G-hat project exploring astronomical signatures of alien civilizations, and were working on a paper on transit signatures of alien megastructures.
Section 4 of the paper discusses KIC8426852 as an example of the sort of thing you might see if there were alien megastructures orbiting stars out in the galaxy, and suggests that currently KIC8426852 is the best known candidate. Which it is.
Which does not mean that the dimming of the star is caused by alien megastructures.

The first transit, looks like a backward comet transit - a comet leading with its tail - a very very big comet; but maybe that is a hint, and we can play with some numbers and try to find testable observables to check.

If the transits are cometary, then they are not actual comets, they have to be disrupted debris clouds from multiple cometary bodies.

The optimal scenario has the comets shattered into perfect 0.1-1 μm dust grains, doesn't matter much whether it is ice, carbon or silicates.
Opacity of such grains is maximal and robust for optical, and fairly flat - though there is still a prediction if the dimming is from optically thin (translucent) dust, then the amount of dimming is different depending on the colour of the light, Kepler only looked in white light so we haven't seen that yet.

So how much dust do we need?
The opacity is ~ 10-21 /NH cm2 - for standard solar dust/H ratio.
There is no H here, but that is ok, we can ignore that.

1) If the cloud is opaque, it is few 100,000 km across and some 'rithmetic suggest a total absolute minimum dust mass of ~ 1016 gm.
This corresponds to about 2 km radius rocky/icy body totally shattered into micron dust grains.

2) If the cloud is translucent and ~ 1,000,000 km across, then you need more dust, but can still do with ~ 3 km radius rock. As an absolute minimum, if ground up perfectly into micron dust. Which it won't be.

Ok that is surprisingly doable.
NB: for realistic dust from comets you need a much bigger comet, since actual comets rarely get ground up evenly into perfectly distributed 1 micron dust grains - which is a problem.

But, these dust clouds are not bound.
They ought to be spreading apart with speeds ~ 0.1 km/sec or so.
So they only last ~ few days-to-months!
This is not much longer than their transit time.

And they had to cross right in front of our line of sight at just the right time. Odds of them doing that were about 1/1,000.

But, Kepler did look at over 100,000 stars!

But, most of them were boring stars unlike this one...

There is one thing about this star - it has a probable close companion.

KIC8462852 candidate companion KIC8462852 candidate companion

There is an apparent M star projected about 1,000 AU from the primary star. In just the right place to be in an orbit that would send it careening through the Kuiper belt of the primary star every million years or so, triggering a comet shower onto the star!

We know that the distribution of comet close approaches ~ 1/a where a is the distance of closest approach. So roughly one comet in 100 comes within 1 AU of star if they start at 100 AU.

We have to keep up the comet showers every million years for a billion years without running out of comets.
So, how many comets do we need?

The answer is well over a billion.

That, is not impossible.

So is that the answer?
Well, maybe - if it is, then we won't see those transits again, because the comets will be on near hyperbolic orbits that won't come around and we saw many different objects on adjacent orbit.
We'd also expect the dust clouds to dissipate.

The infrared glow from such clouds is ~ 10-6 the luminosity of the star, maybe 0.1% or less of the infrared glow of the star (depending on their exact distance and temperature), which is very hard to pick out.

However, the real question would then be, what broke the comets? It could be a fast very close passage to the star, or a very close passage to a planet, or a very unlikely comet-comet collision, but none of those seem likely to produce the sort of dust clouds required to explain the dimming we see.

There is an alternative, non-alien scenario, for the next post...

In the mean time, contemplate this: we just learned that normal middle aged stars can naturally have stuff come and dim them by 10-20% for multiple days.

Imagine that happening to the Sun.

More like this


By Derek Fox (not verified) on 19 Oct 2015 #permalink

Is it at all plausible that the star could be an intrinsic variable? Say, with the internal dynamo doing strange things like this?

The lack of an infrared signature is a dealbreaker for quite a few otherwise plausible models. Including one in which putative aliens are using a piece of megastructure like a partial Dyson sphere--they have to dissipate heat somehow, and for surfaces with temperatures of a few hundred kelvin (as anything involving alien engineering would have to be), it would be dissipated in infrared. Suppressing that would involve such tremendous amounts of liquid helium that they'd have to devote some large fraction of the collected energy to re-liquefying the helium, so I'm not sure that would be worthwhile.

By Eric Lund (not verified) on 19 Oct 2015 #permalink

No, there is no way the star can dim like that on these time scales intrinsically.

The problem with many natural phenomena is you'd expect the structure to be circumstellar or radially extended, which is precluded by the absence of mid-IR emission in the WISE data.

So, any IR has to be either transient, having appeared post-WISE observations; or, be low enough to be below WISE upper limits, which requires fairly small large patches of obscuration; or, all the re-emission is in far-IR outside WISE's sensitivity range.
The latter two are broadly consistent with alien megastructure conjectures.

The real tests will be whether the transits (if they repeat) are achromatic; whether the shapes are time variable; if there is any mid/far IR (now); whether the star is really young (GAIA will nail that down); and what the transits look like in detail over time. if they repeat - there is some high S/N data possible on this.

It is really hard to fit anything plausible to these things. Lots of clever people have been trying for over a year. Requires either lots of arbitary free parameters, or amazingly freakish luck to have caught a rare phenomenon in the act; or we are very wrong about something.

If the objects are orbiting, wouldn't there be positions from which starlight would be reflected rather than occulted ? The graphs only appear to show dips.

In general, yes, you can get forward scattering and reflections possible. We see them in some cases for transiting planets.
They depend on the albedo of the object transiting.

If you look carefully at the lights curve, there are in fact a few places where the light curve seem to be above the baseline at bit less than 1% level - that is a huge signal for Kepler, and may be real.

Suddenly a weird newspaper headline became way more interesting.

By Miguel V. (not verified) on 20 Oct 2015 #permalink