K2-18b. We urgently need to get our shit together about life detection

We have not discovered life on K2-18b.

We have not even discovered water on K2-18b.

All we can say is that the planet does not… not… have an ocean.

Which is probably the same line you used when trying to impress your ninth grade crush with talk of your family’s summer getaway. In Arkansas.

But this isn’t a mundane ball of nothingness. The observations of K2-18b are awesome, and the possibilities are WILD. But first, we have to debunk a week of ridiculous headlines.

So snap on your rubbery marigolds, it’s time to shovel some exo-bullshit.

Artist concept of K2-18b. It’s pretty, but we don’t know anything about the surface, just the upper atmosphere. (Credits: Illustration: NASA, CSA, ESA, J. Olmsted (STScI), Science: N. Madhusudhan (Cambridge University))

Haven’t we been here before with K2-18b?

Oh my good friends, yes we have. In 2019, observations with the Hubble Space Telescope appeared to spy water in the atmosphere in K2-18b; an extrasolar planet that happens to orbit in the so-called habitable zone. It was an amazing observation made by a telescope that was launched before we had discovered a single planet around another star. But it caused a torrent of liquid-rich bovine faeces to splatter across our news screens.

And this was because a basic point was ignored: K2-18b is not Earth-sized.

K2-18b has a radius that is 2.6 x Earth and a mass that is 8.6 x Earth, which seems chonky until you calculate a density of only 2.7 g/cm3. This is super low. For contrast, the Earth has an average density of 5.5 g/cm3 and if you scaled up the Earth so that our planet was the same physical size at K2-18b, the density would sit around 10 g/cm3. So to have such a low density, the volume of K2-18b must be filled with something fluffier than Earth. The most like explanation is an atmosphere of hydrogen and helium (the lightest and fluffiest of all gases) and/or a global colossal ocean.

Both these options make habitability… tricky.

Adding hydrogen to the planet atmosphere is like giving the planet an extra warm coat, much thicker than the Earth’s atmosphere, and risks over heating the surface in the habitable zone. Alternatively, too much water can result in deep sea ices, that cut off any link between the planet’s rock and the surrounding environment. This is a problem for adjusting the temperature and nutrients for life.

Like carbon dioxide, hydrogen is a greenhouse gas that traps heat. Calculations by Eric Lopez and Jonathan Fortney estimate that even 0.5% mass of hydrogen and helium in the atmosphere can wrap a planet in a blanket so thick that the surface pressure reaches twenty times higher than the bottom of the Marianas Trench (ain’t nothing living there but James Cameron’s sub) and temperatures soaring to thousands of centigrade.

Meanwhile, a huge global ocean puts so much weight on the planet that it throttles the geology. Forming under high pressure, deep sea ices seal off the rocky parts of the planet and (at the very least) shuts down the carbon-silicate cycle. On Earth, it is this cycle that keeps our surface temperate as it moves carbon dioxide between the atmosphere and rocks in response to cooler or hotter temperatures. The movement also cycles nutrients needed for life from the planet interior to the surface, so is an all round GOOD THING. Without it… well, we’ll get to that… but it ain’t Earth.

Anyway, I got mad about the whole thing at the time and wrote a rant about it [here].

So what’s happened now?

Now, my good people, JWST has taken a squint at this planet: our newest and most incredible space telescope. And it saw…

No water.

So was the Hubble data false? Actually, it seems it had just confused methane for water.

The Hubble and JWST observations of the atmosphere of K2-18 b. The black points on the left show the data collected by Hubble. The JWST data covers a much much larger range!

Both Hubble and JWST examine the atmosphere of planets through a technique called transmission spectroscopy. This involves looking at the starlight that passes through the planet’s atmosphere, and seeing which wavelengths of that light are missing. Each type of molecule in the atmosphere will absorb a particular set of wavelengths to create a ‘fingerprint’ of missing light. By noting which wavelengths did not make it through the gas, we can therefore tell what molecules are hanging around.

However… Hubble’s Wide Field Camera 3 can only see wavelengths between 1.1 - 1.7 microns. And in that range, the fingerprint for water looks very similar to that of methane. It is like looking at the lines of a human fingerprint, but only being able to see the top quarter of a finger and thereby mixing up two suspects. It was only when JWST observed the much wider range of wavelengths between about 2.73 - 5.17 microns with the telescope’s NIRSpec and NIRISS instruments that a more complete picture of the fingerprint could be seen. It then became obvious that the molecule doing the absorbing was methane, not water.

A dry world?

So why did we get headlines such as,

Tantalising sign of possible life on faraway Earth” (BBC), “NASA’s new telescope may have made biggest discovery of the century” (NY Post), “In the hunt for alien life, this planet just became a top suspect” (National Geographic), and “Huge breakthrough in the search for aliens”.

…OK, that last one was from the Daily Mail but the rest presumably read something that wasn’t a synopsis for “Men in Black”.

This is where things get a bit wild, so pour that double espresso and look alive.

The JWST results show strong evidence for both methane and carbon dioxide in a hydrogen-rich atmosphere. And… the paper notes… a sort-of wiggle that might indicate a molecule called dimethyl sulphide.

Dimethyl sulphide is a terrestrial biosignature. That is, it is a product of living organisms on Earth. Most of it is puffed out by phytoplankton in the sea. Seeing it in the atmosphere of another planet might therefore suggest life.

Moreover, transmission spectroscopy (the technique we’re using to spot these molecules) only detects stuff in the upper atmosphere. So, it’s possible that an ocean exists but water just never reaches the stratosphere.

And it’s those two last ideas that reversed the non-detection of water and sent the news into LaLa land.

Hidden oceans

Let’s start by taking a look at the water. Suggesting there’s an undetected ocean sounds like a giant cop-out, but the possibility is given traction by theoretical models. In particular, the journal paper describing these results points out that a planet that harbours an ocean under a relatively thin hydrogen atmosphere is predicted to produce carbon dioxide and methane in the upper atmosphere at abundances consistent with the JWST observation.

The lead author of this study, Nikku Madhusudhan, is clearly pretty keen on this interpretation. They are a proponent for the existence of so-called “Hycean worlds”, where a global ocean exists under a hydrogen-rich atmosphere that is thin enough to prevent cooking the planet. Two earlier papers by Madhusudhan prior to the JWST observations already proposed this for K2-18b.

JWST (via transmission spectroscopy) can measure gas in the upper atmosphere of the planet. Deeper in the atmosphere and closer to the surface, we need to rely on theoretical models. JWST spotted methane and carbon dioxide. Models suggested that this is not incompatible with a global ocean, but no water was detected.

And… they may be right. The paper compares predictions for the upper atmosphere from a Hycean world and planets with thick and thin hydrogen atmospheres but no ocean. In addition to detections of carbon dioxide and methane, the Hycean model also predicts that water would condense at lower altitudes and appear only with a low abundance in the upper atmosphere, along with low abundances of ammonia and carbon monoxide which were also not detected by JWST. The other models without an ocean appear to match the data less well.

It is therefore seems not implausible that K2-18b could still harbour an ocean. But… there are a few points to take home.

Firstly, water has not been detected. It’s a deduction based on a theoretical model of a planet very unlike our own. Therefore, we might come up with other ideas in the future that match the data better, or conduct observations that can probe to lower altitudes.

Secondly, that hydrogen atmosphere is a potential problem for a comfortable water-y getaway. The paper notes that the greenhouse effect of hydrogen could turn your nice splash pad in the habitable zone into a supercritical phase of water that exists somewhere between a solid and liquid. But if you make the hydrogen atmosphere too thin, then it will escape the planet. This issue could be alleviated by clouds. If the planet atmosphere has a significant pile of clouds or haze, then these could scatter the sunlight and keep the planet cool. The observations… marginally support clouds… but it’s not strongly compelling, so we don’t know if this could be happening or not.

Thirdly, if K2-18b is a Hycean world, then it’s got a global ocean. And there’s the issue that a global ocean can squeeze the geological life out of a planet. If the carbon silicate cycle is cut-off, then alternative mechanisms are needed to adjust the greenhouse gases to keep the planet temperate, and provide the nutrients that would be available from the silicate rocks. This… is not impossible. Carbon silicate cycle substitutes have been proposed for worlds with global oceans, and one of Madhusudhan’s earlier papers has suggested sufficient to support nutrients for life could arrive externally through asteroids or dust. But these ideas are theoretical without (yet) any known planet where we have seen this played out.

The probability of being probable

Now possibly this rather challenging world would not have been staked out as grounds for a new Starbucks if it wasn’t for the claim of dimethyl sulphide. And the problem here is that it’s a whiff of a possibility, but it is not a detection.

To put it in statistical terms, confidence that a result is not down to chance is usually expressed as a sigma level. Real confidence where we’re prepared to say it’s a “detection” usually needs to be at least 3-sigma. The detection of methane and carbon dioxide in an hydrogen-rich atmosphere for K2-18b is at 5-sigma and 3-sigma, respectively. However… for dimethyl sulphide the detection is only around 1-sigma for the best fit to the observational data.

Flip a coin 100 times. If you get 55 heads, would you suspect it was biased? That’s a 1-sigma detection. 65 heads is 3-sigma (which would be way more suspicious).

The journal paper does discuss the possibility of dimethyl sulphide at length, but it is clear that there is not a lot of evidence for its presence. Indeed, the authors note that a very high biological production rate would be needed if their most optimistic abundance estimates proves to be true, which would require a re-think on how dimethyl sulphide chemistry works.

This brings us to a risk with biosignatures. No scenario for K2-18b gives it an Earth-like environment. This means that we need a lot more information about conditions on that planet to be sure that a biosignature on Earth could not be abiotically created in a different environment. Such studies have not yet been done for K2-18b, or Hycean worlds in general.

Despite these major caveats, the presence of dimethyl sulphide is mentioned in both press released by NASA and Cambridge, giving an impression of a far more certain discovery, and also promoting the idea that K2-18b must have an ocean.

A further important point was noted by Ryan MacDonald, a NASA Sagan Fellow at the University of Michigan. Posting on Twitter and Bluesky, MacDonald pointed out that analysing atmospheric data with JWST is still a very tricky and very new task. This makes all the results super sensitive to the technique used to analyse the data. The way around this is for different teams to try their hand at it, and (if possible) to repeat the observations. The observational data will become public in one year, allowing a second opinion on these intriguing results.

Not like Earth

And the results are intriguing. I did love this paper, and I especially love the idea of Hycean worlds because they are so COMPLETELY DIFFERENT to anything in our own Solar System. So I am truly excited by the prospect the K2-18b might be a giant alien water world, and we might get the opportunity to probe its environment and geology in the future.

However…

The current results cannot tell us that K2-18b has water.

There is no biosignature detection.

Let’s talk detection

And here is the problem. We’re now moving from the era of planet detection, to that of planet characterisation. Changing from planet discoveries where we know little about the new world other than its size and orbit, to beginning to probe actual conditions on that planet.

It’s incredibly exciting.

But unless we can clearly describe what we’re finding, every press release from now on risks being misleading and confusing.

What would really help is a way to clearly describe possible biosignature detections. A method that could be used in press releases to say “this discovery is at [this point] on the scale of certainty”.

Without this clarity, we risk the loss of support for the big expensive post-JWST missions that will take us to the next step of examining planets like K2-18b, and (even more sadly) we may undermine everything we are discovering now by pretending it’s something else.

The development of such a scale has been discussed in the planetary science and astrophysical communities. Jim Green, former Chief Scientist at NASA, proposed the CoLD (Confidence of Life Detection) scale in a paper published in Nature in 2021. Running from levels 1 to 7, detection of a signal that can result from biological activity is at level 1, below steps such as necessary checks for contamination (level 2), ruling out abiotic options for that environment (level 4) and stronger evidence for life such as the detection of a second independent signal (level 5).

A workshop on biosignatures standards was also led by Victoria Meadows and Heather Graham in 2022, which produced a community report on the same topic, and a new workshop in 2024 to bring together professional scientists and science communicators is also planned.

The issue is that detecting life is never going to be one-step process, which makes it very difficult to estimate certainty of any one result. But we need a way of communicating this better, because the discovery of worlds beyond our own is incredible and should be shared with everyone on this world.


Almost all of this information (baring a bit of background) is from the journal paper, which was accepted to ApJ Letters. It’s freely available here:
Carbon-baring Molecules in a Possible Hycean Atmosphere

One extra I wanted to add in, but thought it was distracting, is a discussion of what the habitable zone really means. Do check out this comparison with coats!

I also really recommend this fantastic article by Knicole Colon on the capabilities of JWST regarding planet characterisation. It’s an incredible telescope, but this job is blisteringly hard.