Andrew Mann

Spectro-Thermometry of M dwarfs and their planets
Testing the Metal of Late-Type Planet Hosts
Prospecting in Late-Type Dwarfs
They Might be Giants
Dark Matter in Merging Galaxy Clusters
Snapshot Photometry
The Invisible Majority

Spectro-Thermometry of M dwarfs and their planets

More coming soon!

You can find the paper here.
You can also download the machine readable Tables for stellar and planet parameters.
We released all the spectra taken for this study, which you can find here.

Testing the Metal of Late-Type Planet Hosts

It's well known that the frequency of Giant (~Jupiter-sized) planets is correlated to the metallicity of the host star. However, it's a bit less clear what role metallicity has on the smaller planets. Recent studies indicate that Neptune-sized and smaller planets can form around Sun-like stars for a range of metallicities (about -0.5<[Fe/H]<+0.4).

However, there are reasons to think this might not be the case for late K and M dwarfs. These stars have lower masses, and therefore like have lower-mass disks. This is the reason that late-type dwarfs have fewer giant planets than their Sun-like counterparts (independent of metallicity). Thus one could argue that Neptune-sized planets around small stars are the equivalent of giant planets around. This argument suggests that Neptune-sized planets around small stars will show a similar planet-metallicity correlation as giant planets around Sun-like stars.

In this paper, we measure the metallicity of late-type Kepler planet hosts using previously developed spectroscopic techniques (see below). We also observe well as a control sample of Kepler target stars with no detected planets in order to establish the overall metallicity of the late-type Kepler target population. Many of the stars without detected planets probably have planets, but aren't detected because they are not transiting or are too small to measure their transit. So this sample has some 'dilution' that must be taken into account. In fact, probably all or almost all of the stars have small (terrestrial-sized) planets. This is unlikely to be the case for the Neptune-sized planets that are more rare around M dwarfs, so it's possible to account for the dilution.

metalDistribution of metallicities for late-type Kepler target stars. KOIs are stars that host a candidate planet. For multi-planet systems, only the largest (detecteD) planet is considered. Note that many/most non-KOIs probably have undetected planets.

The Figure to the right shows the distribution of metallicities of late-type stars with no detected planet (control, black), terrestrial-sized planets (dashed pink-grey), Neptune-sized planets (blue), and giant planets (red). The median of each sample is marked with an arrow on top. There are only 3 giant planets (and 1 might be a false positive), but they are noticeably more metal-rich than the rest of the sample. Interestingly, however, the Neptune host sample is indistinguishable from the control sample.

This suggests that planet formation must be very efficient at building Neptune-sized objects, since these objects should be difficult to build around these small stars with the limited disk material available. It also suggests that planets might be far more common than originally thought, since these planets are able to form around even metal-poor stars (e.g., thick disk stars), where we previously thought such planets would be less common.

Check out the full paper for a lot more information on our findings.

Prospecting in Late-Type Dwarfs:
improved methods to measure M dwarf metallicities

metalTop: Sample M dwarf spectrum. Bottom: Correlation (measured by the adjusted coefficient of determination) between a given feature and the metallicity of the star (drawn from the primary star metallicity).

Decades of studying the Sun enabled us to determine the metallicity of stars similar to the Sun (late F, G, and early K dwarfs) with relative ease. This is not the case for M dwarfs, where the line lists are incomplete, and the formation of complex molecules creates line confusion. In this paper we use a clever trick to get around this: wide binaries. Since wide binaries likely formed from the same molecular cloud, we can assume they have the same age and metallicity. In the case of a F, G, or K dwarf primary and an M dwarf companion we can measure the metallicity of the companion M dwarf using the primary star. With this information we can develop empirical methods to measure M dwarf metallicities.

This technique has been used before, but one novel thing we do in this paper is to actually analyze what parts of M dwarf spectra change (and by how much) as metallicity changes. The Figure to the right shows a sample M dwarf spectrum (optical and NIR) on top. Below is the correlation between a given feature (or line) strength and the metallicity of the star. This information can then be used to generate improved line lists and help decide what are the 'best' metallicity indicators in spectra.

Click here to see the full paper and here for a pdf of my Cool Stars talk.
You can find a copy of the IDL program that uses our metallicity calibrations here.

They Might be Giants:
systematics in the late-type Kepler target sample

Description coming soon.

Click here to see the full paper.

BHOMS (but not the dangerous kind):

When galaxy clusters (collections of thousands of galaxies) collide, the stars, the gas, and the dark matter behave differently, enabling them to be studied separately. Intracluster gas from one cluster collides with gas from the other cluster, and is slowed down as a result. On the other hand, individual galaxies are like small compared to the megaparsec size of the interaction, and thus galaxies rarely collide with each other. Thus galaxies move more smoothly, slowed a little by gravity, but otherwise unimpaired. You can see a cartoon of this here.

bulletclusterThe Bullet Cluster: X-ray data (gas) is shown in red, and lensing data (mass) is shown in blue. In the absence of dark matter, and assuming simple Newtonian Gravity (or even some flavors of modified Newtonian gravity, MOND) then the blue and red regions should overlap.
Credit: APOD
X-ray: M.Markevitch et al.;
Lensing+Optical: D.Clowe et al.

So what happens with the dark matter? A good question. It turns out we can study the gas, galaxies, and total mass of the system separately. Galaxies are most easily detected in visible light, the hot gas can be traced by measuring the X-ray flux, and mass distribution of the galaxy cluster can be mapped using weak or strong gravitational lensing. If dark matter did not exist, then the the overall mass of the cluster should be well aligned with the gas (since there is more mass in the gas than in the galaxies). If dark matter does exist, and does not interact with itself (i.e., is not slowed significantly during the collision), then the mass will be better aligned with the galaxies. A textbook case is the Bullet Cluster, which shows an 8-sigma separation between the X-ray (gas) distribution and the lensing (mass) distribution, providing some of the strongest evidence to date for the existence of dark matter.

Of course, now we want to go deeper than that. With merging galaxy clusters like the Bullet Cluster we can set limits on the self-interaction cross section of dark matter. If dark matter does not self-interact then it will be perfectly aligned with the galaxies, but if it’s between the galaxy distribution and the gas distribution, it must have some non-zero self-interaction. This can also tell us a great deal about the nature of dark matter; some proposed dark matter theories predict that the dark matter particle will have a large self-interaction cross section, others predict a cross section of ~0.

In this paper we went looking for Bullet Cluster analogs. High-mass clusters provide better constraints, so we focused on galaxy clusters in the Massive Cluster Survey (MACS). We used available X-ray data from Chandra and optical data from SDSS, DSS, or the UH2.2m to produce color optical images with overlaid X-ray contours. Those with significantly misaligned X-ray and optical data are probably disturbed systems/mergers, and are potentially good systems for week lensing studies. See some examples below.

bhom1bhom2 VRI images of candidate merging clusters; overlaid are logarithmically spaced isodensity contours of the adaptively smoothed X-ray surface brightness as observed with the Chandra X-ray Observatory. Left: Abell 2744 Right: Abell 1578. In both cases the X-ray data shows multiple peaks, which are offset somewhat from the galaxy distribution.

As a nice bonus, since our sample covers a wide range of redshifts (0.15 < z < 0.7) we can study how the fraction of disturbed systems changes with time. As expected from cosmological models, we see a significant increase in the number of merging/disturbed systems with increasing redshift. In fact, >80% of the galaxy clusters at z>0.5 show some sign of a recent or ongoing merger!

Click here to see the full paper.

Snapshot Photometry:
a technique to hunt for transiting exoplanets

Description coming soon.

Click here to see the full paper.

The Invisible Majority?

The majority of the first 300-400 detected exoplanets are gas giants, with masses similar to, or greater than that of Saturn or Jupiter (gas giants). This, combined with the fact that our solar system contains 2 such objects, means that most work on planet formation has focused on systems containing gas giants.

However, the preponderance of evidence from doppler, transit, microlensing, and direct imaging surveys suggests that a substantial fraction (if not most) planetary systems have no giant planets. Instead, such planetary systems contain Neptune-mass and smaller objects. Less of these systems lacking giant planets are known, because smaller (less massive) planets are much more difficult to detect.

invis_fig1Planet mass vs. position, normalized by mass and position of the ice line. The IMPs form an almost vertical line on the left side of the Figure.

For this paper we investigated this ‘invisible majority’ of planetary systems, where the disk is able to form massive protoplanets (~1-5MEarth) near the ice line (where water condenses and the surface density of solids is substantially elevated), but is unable to form massive enough cores to form giant planets (M>10MEarth) prior to the dissipation of the gas. We run 230 simulations of such systems to investigate the evolution and final configuration of these protoplanets as a function of a variety of initial conditions (e.g., total mass in protoplanets, mass and position of protoplanets, position of the ice line, etc.).

The Figure to the left shows the final configuration of all simulations as a function of semi-major axis and mass (normalized by the ice line mass and position). Interestingly, independent of initial conditions, each simulation forms a 5-8 Earth-mass planet, which migrates inward to 0.3-0.5 times the distance to the ice line (the ice line is typically at ~ 3-5AU for a Sun-like star). IMPs are seen as an almost vertical line in the Figure to the left (the line is slanted somewhat because higher mass planets do not migrate as far). We nickname this planet the innermost migratory planet (IMP). Further, the inward migration of the IMP almost always accretes, scatters, or otherwise destroys any Earth-like planets we place in the inner solar systems. If it turns out that the IMP is too massive or too cold to be habitable, than this suggests that such systems will rarely have planets which fit into the traditional definition of habitability. However, given that IMPs have large quantities of water (since they formed right at the ice line), it then follows that IMP-like planets (or moons of IMP-like planets) may be the most numerous habitable planets in the universe.

invis_fig2Mass and radius of the majority of our simulated planets. IMPs can be seen bunched up around 1-2 AU. Regions are shaded based on their detectability using different techniques.

Most of these objects are too small and/or have orbital periods too large to detect with existing groud-based Doppler spectroscopy. Many are currently detectible by gravitational microlensing. In fact, such objects may have already been found (e.g., this one and this one). Many of the IMPs are just barely detectable by the Kepler spacecraft. Unfortunately, because IMPS have relatively very long orbital periods (~years), such objects will only be found at the tail end of the Kepler mission. We expect that Kepler will detect >100 of these objects with two or more transits by the end of the mission. Since Kepler traditionally requires 3+ transits to be considered a real detection, many will require followup to confirm their planetary nature. Stay tuned!

Click here to see the full paper.