A Second Look at 10 Planets Says the Odd Signals Are Molecules, Not Noise

Four astronomers have reanalyzed JWST measurements of starlight filtered through the atmospheres of 10 planets orbiting other stars, using one consistent method for all of them. They report that the contested features in three of those planets look more like real molecules than like noise or an instrument artifact. The paper by Martin Binet, Nikku Madhusudhan, Måns Holmberg and Subhajit Sarkar was posted to the preprint server arXiv on Sept. 4 and has not been peer reviewed.

The authors say the survey was built to tell those two possibilities apart. The argument it addresses started with the first mid-infrared reading of K2-18 b, which reported possible traces of a molecule called dimethyl sulfide. Later papers argued the signal could as easily be random noise, an instrument effect or a different molecule, and similar objections followed for other planets studied the same way.

Madhusudhan is an author of this reanalysis and led the original K2-18 b work whose statistical significance is disputed, so the group re-examining the contested result is the group that produced it.

Rather than making new observations, the team worked from spectra already collected with JWST's mid-infrared instrument, MIRI. They first fitted each of the 10 spectra with models carrying no features at all and compared the spectra against each other, and then searched for more than 150 trace gases in every planet's atmosphere. Hints of complex molecules turned up in three planets (K2-18 b, TOI-732 c and TOI-270 d) and in none of the other seven, which run from hot rocky worlds to gas giants. More observations are needed to identify the specific molecules and to establish what chemistry produces them, the authors write, leaving open whether biology is involved.

The three are all temperate sub-Neptunes, sized between Earth and Neptune, and candidate hycean worlds, a proposed class with deep oceans under hydrogen-rich air. The authors say their results mark that group out as distinct.