Artist's View of Extrasolar Planet HD 189733b

Missing Molecules in Exoplanet Atmospheres

Article written: 14 Oct , 2010
Updated: 24 Dec , 2015


Every day, I wake up and flip through the titles and abstracts of recent articles posted to arXiv. With increasing regularity, papers pop up announcing the discovery of a new extra-solar planet. At this point, I keep scrolling. How many more hot Jupiters do you really want to hear about? If it’s a record setter in some way, I’ll read it. Another way I’ll pay attention is if there’s reports of detections of spectroscopic detection of components of the atmosphere. While a fistful of transiting planets have had spectral lines discovered, they’re still pretty rare and new discoveries will help constrain our understanding of how planets form.

The holy grail in this field would be to discover elemental signatures of molecules that don’t form naturally and are characteristic of life (as we know it). In 2008, a paper announced the first detection of CO2 in an exoplanet atmosphere (that of HD 189733b), which, although not exclusively, is one of the tracer molecules for life. While HD 189733b isn’t a candidate for searches for ET, it was still a notable first.

Then again, perhaps not. A new study casts doubt on the discovery as well as the report of various molecules in the atmospheres of another exoplanet.

Thus far there have been two methods by which astronomers have attempted to identify molecular species in the atmosphere of exoplanets. The first is by using starlight, filtered by the planet’s atmosphere to search for spectral lines that are only present during transit. The difficulty with this method is that, spreading the light out to detect the spectra weakens the signal, sometimes down to the very point that it’s lost in systematic noise from the telescope itself. The alternative is to use photometric observations, which look at the change in light in different color ranges, to characterize the molecules. Since the ranges are all lumped together, this can improve the signal, but this is a relatively new technique and statistical methodology for this technique is still shaky. Additionally, since only one filter can be used at a time, the observations must generally be taken on different transits, which allow the characteristics of the star to change due to star spots.

The 2008 study by Swain et al. that announced the presence of CO2 used the first of these methods. Their trouble started the following year when a followup study by Sing et al. failed to reproduce the results. In their paper, Sing’s team stated,”Either the planet’s transmission spectrum is variable, or residual systematic errors still plague the edges of the Swain et al. spectrum.”

The new study, by Gibson, Pont, and Aigrain (working from the Universities of Oxford and Exeter) suggests that the claims of Swain’s team were a result of the latter. They suggest that the signal is swamped with more noise than Swain et al. accounted for. This noise comes from the telescope itself (in this case Hubble since these observations would need to be made out of Earth’s atmosphere which would add its own spectral signature). Specifically, they report that since there’s changes in the state of the detector itself that are often hard to identify and correct for, Swain’s team underestimated the error, leading to a false positive. Gibson’s team was able to reproduce the results using Swain’s method, but when they applied a more complete method which didn’t assume that the detector could be calibrated so easily by using observations of the star outside the transit and on different Hubble orbits, the estimation of the errors increased significantly, swamping the signal Swain claimed to have observed.

Gibson’s team also reviewed the case of detections of molecules in the atmosphere of an extra solar planet around XO-1 (on which Tinetti et al. reported to have found methane, water, and CO2). In both cases, they again find that detections of were overstated and the ability to tease signal from the data was dependent on questionable methods.

This week seems to be a bad week for those hoping to find life on extra-solar planets. With this article casting doubt on our ability to detect molecules in distant atmospheres and the recent caution on the detection of Gliese 581g, one might worry about our ability to explore these new frontiers, but what this really underscores is the need to refine our techniques and keep taking deeper looks. This has been a frank reassessment of the current state of knowledge, but does not in any way claim to limit our future discoveries. Additionally, this is how science works; scientists review each others data and conclusions. So, looking on the bright side, science works, even if it’s not exactly telling us what we’d like to hear.

8 Responses

  1. Uncle Fred says

    Dedicated NEO telescope in order?


    At the first paragraph, in the fourth line, and also at the sixth paragraph, in the sixth line, the contraction “there’s” (there is) is incorrect because the indicative, in both cases, is plural; therefore, it should be: there are.

    Um… I’ll get my coat and see myself out…

  3. Manu says

    “detection of CO2 … one of the tracer molecules for life”
    Surely this can’t apply to Mars & Venus, so what is the key?

  4. William928 says


    If you had read the article more closely, you would have noticed that Jon preceded the statement “one of the tracer molecules for life” with “although not exclusively”. As you mention, CO2 is certainly not an indicator for life on Mars and Venus. I believe the point of the article is to point out the two current methods utilized to attempt to determine atmospheric components of exoplanets. The conclusion I reach in reading this is that there’s alot more technological advancement that needs to take place in the study of exoplanets in general, and their atmospheres in particular. Still, science has come a long way in the 15 years since the discovery of the first extrasolar planet. I can’t wait to see where we are in another 15 years.

  5. Jon Hanford says

    Knowing some of the difficulty in obtaining spectra of exoplanets and their atmospheres detailed here, I was heartened by another announcement today of recent success with the Apodizing Phase Plate coronagraph developed for use on the VLT:

    Besides getting a closeup look at exoplanets in the sweet spots of nearby stars, study of young stars, dusty disks and AGN could also benefit from this instrument.

  6. William928 says

    Thanks for the link Jon. This is encouraging news!

  7. Member
    Jon Voisey says

    @Manu: William was on the right lines. CO2 is A tracer. In and of itself, it doesn’t mean much as it can be produced in a variety of ways. But add a bunch of other molecules (like O2, and the other components we see in our atmosphere), you can make the case for respiration and feedback cycles.

  8. Lawrence B. Crowell says

    A plausible signature of life is the presence of two chemicals which react with each other in some way. Methane and oxygen are good candidates. A spectral signature of water helps as well. Nothing is certain of course. You can’t really know for sure whether a planet has life until you actually go there and look on the ground. This is not at all easy, and probably it is at best possible to send a probe within a 25 to 50 light year radius.


Comments are closed.