We have a fairly good understanding of stars like the Sun: we can measure their compositions and ages, and we have models that explain their physical properties (mass, radius) and the courses their lives take. However, most of the stars in the Galaxy are M dwarfs, which are at most about half the mass of the Sun, and thousands of degrees cooler at their surfaces. Our understanding of these stars is much shakier, and is currently a vibrant area of research (which I am quite interested in!).
M dwarfs are different from the Sun in at least two important ways:
- Their cool atmospheres allow atoms to stick together to create molecules (e.g. TiO, VO) that can't form on the Sun. This makes M dwarf spectra complex, so it is difficult to discern their compositions.
- The coolest M dwarfs also have structures that are dominated by convective energy transport, instead of radiation like in the Sun. Combined with the fact that M dwarfs may have strong magnetic fields, this makes it difficult to construct models that can accurately predict M dwarf properties.
One well-studied M dwarf system is called CM Draconis (CM Dra). CM Dra is a close eclipsing binary system (1.3 day period) of two almost identical M4 dwarfs. These types of eclipsing binary systems are important checkpoints for development of stellar models, as they allow model-independent derivation of stellar parameters (e.g. I can measure the mass dynamically, without assuming anything about the star). CM Dra is one of only 5 known systems that provide such checkpoints at these low masses, so it is an exceptionally valuable system. It also stands out because it has a nearby white dwarf that appears to be associated with the system. If the white dwarf is part of the same system, it (or rather its progenitor) formed at the same time as the M dwarfs, and we can use its cooling curve (and our understanding of how long the progenitor lived) to independently constrain the entire system's age.
The problem for M dwarf models is that they haven't yet been able to hit most of the checkpoints (including CM Dra). For example: knowing the mass, composition, and age (and some reasonable guesses about internal structure) of an M dwarf, one should be able to plug these into their model to determine the star's radius. But CM Dra and other similar systems the stars appear to be "poofier" than expected, which is to say that their radii are too large, sometimes by as much as 10% (far larger than the uncertainties in the measurements provided by the eclipsing binary systems, which are closer to 1%). Popular explanations for this discrepancy are that current models cannot properly account for how magnetic fields in M dwarfs could alter the convection inside the star, or that starspots could have an effect.
Models for M dwarfs are sensitive to the composition of the star. A higher proportion of metals (metallicity) can cause a star to be poofier than a more metal-poor counterpart of the same mass. And the metallicity for CM Dra has until recently not been well constrained, since it only contains M dwarfs, whose spectra can be confusing. So in attempting to model this system, earlier works have assumed that some reasonable amount of the extra poofiness is actually due to the metallicity of the star, and have concluded that in this case CM Dra is only has about a radius discrepancy of ~3%.
We were in a particularly good position to help clarify the CM Dra metallicity situation. About a year ago, we developed a method of using strong features in medium-resolution M dwarf spectra to empirically estimate the stellar metallicity. This relation has proved quite successful for single stars, allowing us to get a handle on the metallicities of several hundred nearby M dwarfs and giving us a jumping-off point for more thorough modeling studies. CM Dra was visible during an observing run (last May) for our M dwarf metallicity program, so it would be great to use this relation on CM Dra! But recall that CM Dra is actually two stars, which are so close together that we can't resolve them. However, we were able to use a clever trick: since CM Dra undergoes almost total eclipses (this system just keeps getting better), if we observe at the right time we can isolate the light from only one star. So this is what we did! We observed it several times, at both primary and secondary eclipses, and out of eclipse as well (for good measure), and every time we looked, our calibration gave approximately the same answer: [Fe/H] = -0.3.
After talking with Greg Feiden (who works on the Dartmouth Stellar Evolution Program stellar models) at the Cool Stars 17 conference this summer, we were able to discern what this meant for models of CM Dra: previous studies likely overestimated its metallicity, thereby masking the true extent of the radius discrepancy, which is likely closer to 6-7%. This may sound like a small difference, but it excludes most explanations for the poofiness that involve simple modifications of the convective model or just starspots. It throws into relief the true extent to which we do not understand the structures of these low mass stars. It will be very exciting to see how these models develop as modelers work out ways to include the effects of magnetic fields. And it will also be important for observers to keep searching for valuable checkpoint systems like CM Dra in order to make sure the models are sufficiently constrained. Hopefully this study will help motivate work from both theoretical and observational groups as we work to understand low-mass stars.
(Above) Low mass stellar observations (points) and models (lines) for stellar mass vs radius. Our metallicity measurement (within a reasonable range of uncertainty, [Fe/H] = -0.30 +- 0.12 dex) constrains which models can be used for CM Dra, which are shown in black. The CM Dra points are significantly different from the model predictions. In fact, most other systems at these low masses deviate from model predictions as well; only KOI-126 matches with model predictions. (From Terrien et al. 2012b, ApJL accepted)
No comments:
Post a Comment