≡ Menu

Toward a Census of Earth-Sized Worlds

While transit and radial velocity methods get most of the press when it comes to finding exoplanets, gravitational microlensing offers an independent alternative. Here a star passes in front of a far more distant object, causing the light from the source to be gravitationally ‘bent’ by the intervening star. The useful thing for exoplanet work is that if the ‘lensing’ star is orbited by one or more planets, they can leave their own signature in the microlensing event. And indeed, microlensing collaborations like MOA (Microlensing Observations in Astrophysics) and OGLE (Optical Gravitational Lensing Experiment) have made the method pay off in exoplanet discoveries.


Image: Gravitational microlensing relies on chance line-ups between an intervening star with planetary system and a more distant light source. Credit: California Institute of Technology.

Now researchers at the University of Auckland are proposing to measure low-mass planets, planets as small as the Earth, using these methods. Microlensing events vary in terms of the degree of magnification they provide, with both MOA and OGLE finding some events with extremely high magnifications. You would think the higher the magnification, the better, but this turns out to be not necessarily the case, as the university’s Phil Yock and colleagues explain in a new paper. Their simulations clarify the methods we can use to detect Earth-sized planets:

Our conclusion is that low-mass planets can be efficiently searched for in events with moderately high magnifications of order 50-200, and that the greater frequency of these events in comparison to those with higher magnifications lends advantage to monitoring them. However, to take full advantage of their discovery potential, telescopes with apertures in the range 1-2m would be needed to photometer them with sufficient precision to detect low-mass planets.

A network like this is, in fact, being deployed by the Las Cumbres Observatory Global Telescope Network, working in collaboration with SUPA/St Andrews (Scottish Universities Physics Alliance). These are 1- and 2-meter instruments, with three telescopes each in Chile, Australia and South Africa, and one each in Hawaii and Texas. Working with a telescope in the Canary Islands and supplementing its data with several other southern hemisphere instruments, the network, according to Yock and colleagues, ought to be able to help us not only detect Earth-sized planets but measure them at substantial distances from their star. Says Yock:

“Our proposal is to measure the number of Earth-mass planets orbiting stars at distances typically twice the Sun-Earth distance. Our planets will therefore be cooler than the Earth. By interpolating between the Kepler and MOA results, we should get a good estimate of the number of Earth-like, habitable planets in the Galaxy. We anticipate a number in the order of 100 billion.”

It’s an intriguing idea. The Kepler effort is all about developing a statistical read on how often Earth-class planets occur in the galaxy by looking at a huge number of stars. Yock’s work demonstrates that low-mass planets can come out of microlensing observations through continuous network monitoring. The paper notes that in recent times, about 10 events with magnifications over 200 have been detected by MOA and OGLE every year. Dropping the magnification requirement to 50 should quadruple the detection rate, allowing us to develop an independent check on measurements made by radial velocity and transit studies.

Consider how useful microlensing can be. We are not burdened, as with radial velocity and transits, with a measurement more sensitive to larger worlds closer to their host stars. The planets thus far detected through microlensing have had masses ranging from a few Earth masses to a few Jupiter masses, while their separation from their hosts has been on the order of several AU. This provides, says the paper, ‘a representative sample of the distribution of planets between us and the centre of the Galaxy that orbit a relatively unbiased sample of host stars.’

Thus we can supplement other ways of estimating the abundance of cool planets in the Milky Way by developing a statistically significant sample of detections through the kind of microlensing events Yock describes. The paper is Abe et al., “Extending the Planetary Mass Function to Earth Mass by Microlensing at Moderately High Magnification,” accepted for publication at Monthly Notices of the Royal Astronomical Society (preprint).


Comments on this entry are closed.

  • andy April 5, 2013, 12:50

    As I understand it, that ~2 AU value is for a star of about 10^5 Earth masses, i.e. ~0.3 solar masses, and microlensing is most efficient to find planets located close to the ice line.

  • Ron S April 5, 2013, 15:08

    The paper and the methodology are interesting. Except (as noted) as an independent method of planet detection by other means I doubt this will result in a reliable set of new detections. The big problem appears to be the “one shot” nature of the detection. You cannot reconfirm the lensing event on the next orbit since the background star is no longer there: it transited, once, and is forever gone. Even with a statistically-large deviation caused by the transit there will be doubts as to whether to assign the event to a planet.

    Also, since the AU separation will be a minimum (due to random distribution of orbit inclinations) this may be more relevant to detecting larger, outlying planets rather than small, rocky planets in the HZ.

  • Raffaele Antonio Tavani April 5, 2013, 18:12

    Segnalo, al creatore dell’interessante articolo, e agli altri lettori del “blog” che almeno uno dei link(il secondo, sotto la voce “RAS”) conduce ad una pagina “infetta”.
    (Almeno, è quello che mi segnala il mio “Norton”).
    E’ bene fare una verifica, perché temo che quella pagina sia “compromessa”.
    Non sono un “esperto” e quindi non sono in condizioni di comprendere il reale pericolo che possono incorrere i “navigatori” visitando quella pagina collegata al secondo link consigliato.
    (Il primo link, non l’ho verificato).
    Spero, di potervi essere stato utile.

    Via Google Translate:

    Let me point out to the creator of the interesting article and other readers of the “blog” that at least one of the links (the second, under the heading “RAS”) leads to a page labelled “infected”. (At least, that’s what tells me my “Norton”).
    It ‘good to do a test, because I fear that the page is “compromised.” I’m not an “expert” and therefore are not in a position to understand the real danger that “navigators” may incur when visiting that page connected to the second link recommended.
    (The first link, I have not checked).
    I hope to be able to be helpful.

    Greetings from Antonio Tavani

  • Daniel April 6, 2013, 2:40

    Talk About Exoplanets

    there is a good news, TESS Mission it’s been selected for launch in 2017


  • Paul Gilster April 6, 2013, 8:31

    Antonio Tavani writes:

    Let me point out to the creator of the interesting article and other readers of the “blog” that at least one of the links (the second, under the heading “RAS”) leads to a page labelled “infected”

    I suspect this is a mistake — in any case, I’m not getting any virus warnings here using Kaspersky and the page seems OK. I’m going to remove the link, though, just to be on the safe side.

  • Mike April 6, 2013, 13:29

    To Daniel. That’s great news about TESS being selected. M.I.T. has a good article on TESS at their website. web.mit.edu.

    I enjoyed the article on the Las Cumbres Observatory Global Telescope Network. What a great concept. I’m certain it will do very well.

    It will be interesting to see if planet hunting via micro-lensing observations proves to be a productive and reliable technique. Since micro-lensing maybe able to detect long period planets it could make detections that are useful for follow up direct imaging that require using expensive large telescopes that aren’t available for lengthy planet searches.

  • NS April 6, 2013, 18:27

    Glad to hear about the TESS mission being selected. An important follow-up to Kepler.

  • Marshall Eubanks April 7, 2013, 17:38

    @Ron S – Yes, it can hard to be sure what you have with any given microlensing detection. Note, however, that these are routinely observed by a suite of telescopes around the world (and even by spacecraft up to several AU away – see, e.g., http://www.ipac.caltech.edu/wfir2012/abstracts/Snodgrass.pdf ).

    There is thus no doubt that most of these events occur. The possible ambiguities occur in their interpretation. This enables the use of parallax, which helps to break degeneracies in a single measurement – see http://iopscience.iop.org/1538-4357/698/2/L147/pdf/1538-4357_698_2_L147.pdf . This is, however, not yet possible for all microlensing events.

    However, if you are trying to get statistics, even ambiguous measurements can be useful. (For a crude example, if you don’t see any signs of a planet in a large number of microlensing events by a particular type of star system, then the rate of planetary occurrence for that type of system is presumably small.)

  • Phil April 7, 2013, 22:52

    Is it just me – or are new exoplanet announcements way down this year (thus far) compared to recent years?


  • Ron S April 8, 2013, 9:43

    Marshall, your points are good ones. Perhaps I should have been less “qualitative” in my comment. What I wonder is, over many observed events, are we measuring the statistics of stellar anomalies or the statistics of exoplanets? The degree of uncertainty about stellar characteristics is key in this, and one I do not have a handle on.

    I did see the parallax section of the paper but I perhaps dismissed that too soon since I didn’t see how it could prove useful. Thank you for that link: I’ll have a look.

  • coolstar April 8, 2013, 15:41

    I agree with Ron S that the traditional weakness of micro-lensing experiments is that there is no follow-up observations that are possible. thus you need a really large sample of detections to try to counter-balance that in a survey. It’s an useful technique, but all those 1 – 2 meter class telescopes could possibly be more gainfully employed doing photometry and high precision spectroscopy of M stars, as in the Minerva project. Minerva will detect and characterize NEARBY planets, that can be followed up with the next generation of large ground based telescopes, and, perhaps in a few cases, with JWST. This paper may be of interest: http://labs.adsabs.harvard.edu/ui/abs/2013arXiv1303.3013M it’s by Morton and Swift and is the most recent attempt at characterizing the size distribution of planets around M stars from Kepler data. They find, convincingly in my opinion, a roughly 50% increase over even the very recent numbers of Dressing and Charbonneau.

  • andy April 9, 2013, 14:29

    How much of an issue is non-repeatability anyway? For example, you get a rich set of science coming out of things like supernovae and gamma-ray bursts.

    As for microlensing being “less useful” than other planet detection techniques, I wouldn’t be so fast to make that judgement – it does probe a rather different region of the parameter space than other methods.

    Also to note: there has been follow-up of a few microlensing planet detections in terms of direct detection of the lens star. E.g. Identification of the OGLE-2003-BLG-235/MOA-2003-BLG-53 Planetary Host Star

    (I guess one of the most serious problems with microlensing as an exoplanet detection method is the truly horrible designations…)

  • Ron S April 9, 2013, 18:41

    andy: “How much of an issue is non-repeatability anyway? For example, you get a rich set of science coming out of things like supernovae and gamma-ray bursts.”

    You are correct that the data set is rich. However supernovae are not the data, the light is. We assign that data to a supernova event. This may seem a trivial observation since the event assignment isn’t difficult. For a data set from a GRB we assign it to…well, they’re still working on that though there certainly are candidates. Microlensing is perhaps similar in that exoplanets are strong candidates for particular data sets.

    Repeatability helps with the event assignment, making it “useful”. Parallax (thanks to Marshall for that reference) also seems to be useful in event assignment. It’s good to have this since we do not repeatability.

  • coolstar April 9, 2013, 18:54

    @Andy, yes, there have been a few followups on the lensing the star but the problem is that most of them are too faint to use any of the OTHER techniques on. That will probably change somewhat with the next generation of giant ground based telescopes though.

  • Matthew April 11, 2013, 11:36

    @coolstar @andy It is indeed very difficult to follow up microlensing detections with other techniques as the host stars are usually very distant and therefore very faint. However, there are rare cases that are amenable to follow up (see e.g. http://arxiv.org/abs/1302.0841). In exceptional microlensing events it is possible to measure some or all of the orbital parameters of the planet/star, and these can be used to make predictions about the radial velocity signatures that would be observed. Testing these golden follow-up systems will add confidence to the conclusions you draw from the larger sample of all microlensing events.

  • notetaker April 12, 2013, 8:34

    Am I correct in understanding that microlensing is focusing on a specific object on the far side of the star (system) that passes in front of it? If this is the case it would seem like it should only be a matter of time before we are able to focus on a specific spot of empty space behind any star system and achieve a similar microlensing affect – but my understanding of microlensing may be a bit limited.