What kind of stars are most likely to have planets? Narrowing the search is crucial if the goal is to build a target list for space-based missions, especially when we’re looking for terrestrial worlds. So learning that planet-bearing stars have higher metal contents — the elements above hydrogen and helium, presumably as a relic of their protoplanetary disks — would winnow the target list nicely, at least among Sun-type stars.
And most of the extrasolar planets found thus far have been in orbit around stars of the spectral types F, G and K (our Sun is a G-class star). For various reasons, these are the usual targets for radial-velocity surveys, and they’re also stars that can be readily analyzed for metallicity. But M dwarfs are the most common stellar type. We need to go to work on their parameters too, especially in the case of dwarfs that have known planetary companions.
A stride in that direction is taken by Jacob Bean (University of Texas), Fritz Benedict and Michael Endl (McDonald Observatory, UT) in a new paper that analyzes the metallicities of three planet-bearing M dwarfs. Only one of these, GJ 876, is known to have Jupiter-mass planets around it; it’s also the only M dwarf with more than one known planet. GJ 436 and GJ 581 appear to have Neptune-mass planets in close orbits.
The scarcity of high-mass planets around M dwarfs suggests that smaller stars are less likely to harbor them, which is consistent with the core accretion model of planet formation. In that model, gas giants build up rocky cores through collisions in the protoplanetary disks around young stars, reaching a critical mass beyond which gas from the disk rapidly accumulates.
As to metallicity, the three dwarfs studied all have metallicities lower than our Sun’s, a result that defies the planet/metallicity link in higher-mass stars. From the paper:
This result raises some interesting questions. Are the metallicities for these stars representative of the metallicities of the M dwarfs on planet search programs and might that explain the lower detection rates of planets for the M dwarfs? If that were the case, are the solar neighborhood M dwarfs in general metal deficient relative to the other spectral type? Or, what is causing the selection effect to lower metallicity M dwarfs for the planet search programs?
By way of answering these questions, the researchers plan to apply their analysis to a larger sample of M dwarfs that may show clearer trends in mass and metallicity. The question is important, for the link between planets and high metallicity seems clear in the larger F, G and K stars. Why and how do these planet-bearing M dwarfs depart from the trend?
The paper is Bean, Benedict and Endl, “Metallicities of M Dwarf Planet Hosts from Spectral Synthesis,” accepted for publication in The Astrophysical Journal Letters and available as a preprint online.
With Gliese 876 having a listed metallicity of 0.76, I don’t know if we can really consider it to be metal deficient. That’s pretty close to the Solar value, after all, and far above the mid-40’s range cited for the other two stars.
This probably depends on where you are getting the listed metallicity – the McDonald group mentions a 2005 study (Bonfils et al) that arrived at different metallicity values than the ones presented in this paper. But as I go back through the paper, the main reference is simply to ‘sub-solar metallicity.’ So you may well be right, though with Gliese 876 having the highest metallicity and also being the one M dwarf with three known planets, the trend the paper establishes seems viable. It will be interesting to see what the follow-up shows, as this seems to be the most accurate metallicity study of M dwarfs available.
Interesting that, especially in the light of a previous paper (http://arxiv.org/abs/astro-ph/0503260) giving near-solar metallicities for these stars – in that, Gliese 876 was measured as slightly above solar metallicity, and Gliese 436 only slightly subsolar.
Yes, that’s the Bonfils paper I referenced above — they do get different values, though here’s what the McDonald group says about that:
“In contrast, Bon?ls et al. (2005a) presented metallicity measurements for these M dwarfs based on a lower precision photometric relationship. Our derived [M/H] values are lower by 0.09, 0.08, and 0.34 dex for GJ 876, GJ 581, and GJ 436 respectively than those determined by Bon?ls et al. (2005a). In the case of the ?rst two, the values are well within the overlapping errors for the two measurements (0.32 dex), while our measurement for GJ 436 is just outside this di?erential range.”
And from this, they conclude:
“Taken together, the results from our analysis and that of Bon?ls et al. (2005a) do appear to rule out super-solar metallicities for these M dwarf planet hosts.”
The 0.76 figure for Gliese 876 I found at Ken Croswell’s website.
At any rate, this will indeed be an interesting and insightful matter, especially when a greater sample is achieved.
I think we just don’t have enough data to make any conclusions one way or another. Even with the more voluminous data for heavier stars, we still have to contend with pretty serious detector selection effects which skew the results in ways that dramatically narrow the range of conclusions we can make (essentially down to making lower bar estimates on overall planetary abundance, for example). This is even worse for low-mass and low-brightness stars because essentially every existing planetary detection method is dependent on intrinsic stellar brightness. Radial velocity measurements depend on brightness heavily for adequate signal to noise ratio of the stellar spectrum. Transit methods depend on brightness equally as heavily as they measure variations in brightness directly (meaning that low brightness means less signal to noise ratio and thus less effective dynamic range from which to extract a signal). Direct detection measurements (e.g. brightness changes due to reflection or direct detection via occultation / deep-nulling of the parent star) are similarly heavily dependent on brightness. Astrometry is less dependent on brightness, but has so far proven very ineffective at planetary detection.
Because the “effective detection distance” of a particular planetary detection method scales inversely with the square root of the brightness, and because the number of targets available within a given distance scales roughly to r^3, that means that the difference in available targets scales roughly to the 3/2 power by intrinsic brightness. Given that the intrinsic brightness ranges we are talking about when comparing F, G, and K stars to M dwarfs is several magnitudes, this leaves M dwarf planetary systems at a distinct disadvantage in terms of sample size, which is reflected in the paltry data we have. When we have tens or hundreds of detected M dwarf planetary systems, then we might start to talk about statistics and trends, but until then this is just guess work with too little data.
That’s a very good point, Robin. The dataset here is still tiny, and for the reasons you sum up so succinctly. It may be a while before we have the kind of numbers you’re looking for but a couple of hundred M dwarf systems would be useful indeed for figuring out these trends.
Also, don’t red dwarfs last a lot longer than other stars? Those that formed when there were a lot less heavy elements would still be around.