Li Zeng

Home Planet Models Media Publication Courses C.V. Honors Taichi Research Experience Teaching & Public Outreach


Media Report

Water Worlds


Water Worlds
Image Courtesy: NASA

External links:

BBC Radio (14:00 onward)

Discovery Magazine

list of press coverage:

Goldschmidt: the world’s main geochemistry conference
Water-worlds are common: exoplanets may contain vast amounts of water

  • What has been found? That the known masses and sizes of many exoplanets of two to four times the size of Earth can be explained by large amounts of water.
  • Why is it important? Water has been implied previously on individual exoplanets, but this work concludes that water-rich planets are common. This bodes well for planet formation of Earth-like planets with water and the search for life beyond our Solar System.

Scientists have shown that water is likely to be a major component of those exoplanets (planets orbiting other stars) which are between two to four times the size of Earth. It will have implications for the search of life in our Galaxy. The work is presented at the Goldschmidt conference in Boston.

The 1992 discovery of exoplanets orbiting other stars has sparked interest in understanding the composition of these planets to determine, among other goals, whether they are suitable for the development of life. Now a new evaluation of data from the exoplanet-hunting Kepler Space Telescope and the Gaia mission indicates that many of the known planets may contain as much as 50% water. This is much more than the Earth’s 0.02% (by weight) water content.

“It was a huge surprise to realize that there must be so many water-worlds”, said lead researcher Dr Li Zeng (Harvard University),

Scientists have found that many of the 4000 confirmed or candidate exoplanets discovered so far fall into two size categories: those with the planetary radius averaging around 1.5 that of the Earth, and those averaging around 2.5 times the radius of the Earth.

Now a group of International scientists, after analyzing the exoplanets with mass measurements and recent radius measurements from the Gaia satellite, have developed a model of their internal structure.

We have looked at how mass relates to radius, and developed a model which might explain the relationship”, said Li Zeng. The model indicates that those exoplanets which have a radius of around x1.5 Earth radius tend to be rocky planets (of typically x5 the mass of the Earth), while those with a radius of x2.5 Earth radius (with a mass around x10 that of the Earth) are probably water worlds”.

“This is water, but not as commonly found here on Earth”, said Li Zeng. “Their surface temperature is expected to be in the 200 to 500 degree Celsius range. Their surface may be shrouded in a water-vapor-dominated atmosphere, with a liquid water layer underneath. Moving deeper, one would expect to find this water transforms into high-pressure ices before we reaching the solid rocky core. The beauty of the model is that it explains just how composition relates to the known facts about these planets”.

Li Zeng continued, “Our data indicate that about 35% of all known exoplanets which are bigger than Earth should be water-rich. These water worlds likely formed in similar ways to the giant planet cores (Jupiter, Saturn, Uranus, Neptune) which we find in our own solar system. The newly-launched TESS mission will find many more of them, with the help of ground-based spectroscopic follow-up. The next generation space telescope, the James Webb Space Telescope*, will hopefully characterize the atmosphere of some of them. This is an exciting time for those interested in these remote worlds”.

Professor Sara Seager, Professor of Planetary Science at Massachusetts Institute of Technology, and deputy science director of the recently-launched TESS (Transiting Exoplanet Survey Satellite) mission, which will search for exoplanets, said:

It’s amazing to think that the enigmatic intermediate-size exoplanets could be water worlds with vast amounts of water.  Hopefully atmosphere observations in the future—of thick steam atmospheres---can support or refute the new findings”.

Image:  Exoplanets similar to Earth, artist concept. Image from NASA

*JWST is the James Webb Space Telescope, which will be the successor to the Hubble space telescope. It is due to be launched in 2021, see for background.

This press release is based on work presented at the Goldschmidt conference, Boston, Abstract Growth Model Interpretation of Planet Size Distribution, see notes for full abstract.

Notes for editors
Please mention the Goldschmidt Conference in any press story arising from this press release.
Contact details:
Li Zeng        
Goldschmidt Press Officer, Tom Parkhill :  tel +39 349 238 8191
Conference Abstract: Growth Model Interpretation of Planet Size Distribution
1. Earth and Planetary Sciences, Harvard University , MA  (*correspondence: ); 2. Harvard - Smithsonian Center for Astrophysics , MA; 3. Department of Astronomy, UT Austin , TX; 4. Sandia National Laboratories , Albuquerque , NM; 5. School of Physics, Georgia Institute of Technology, Atlanta, GA 30313; 6. INAF-Osservatorio Astrofisico di Torino, via Osservatorio 20, 10025 Pino Torinese, Italy; 7. Institute for Astronomy, University of Hawaii, 2680 Woodlawn Drive, Honolulu, Hawaii, 96822, USA

The radii of over 4000 exoplanet candidates have been precisely measured by the NASA Kepler Mission, along with their orbital periods and other parameters [1]. Their radii show a bi - modal distribution, with the main and secondary peaks likely corresponding to Earth - like rocky planets and larger intermediate - sized planets, respectively [2 – 4]. The masses of planets can be determined by ground - based spectroscopic observations, but only for planets orbiting the brightest stars. These observations, allow calculations of average densities and, thus, constraining their bulk compositions and internal structures. Hence, an important question about the compositions of the planets ranging from 2 to 4 Earth radii still remain [5,6] . They may either have a rocky core enveloped in a massive Hydrogen - Helium gas (gas dwarfs) [3,7 – 9] or contain a significant amount of multi - component,  H2O - dominated  ices/fluids  (water  worlds).  The growth model tracks how mass and radius change when a planet population grow from rocky core and subsequently accrete either  O - H - C - N - ices or  Hydrogen - Helium gas . The observational radius and mass - radius distribution can be reproduced by the growth model with a Monte Carlo simulation. Because their composition cannot be uniquely constrained, we use growth model and Monte Carlo simulation for these planets to argue that many intermediate - sized planets are “water worlds”. [1]  Akeson et al . (2013)  Publ. Astron. Soc. Pacific 125 , 989 .  [2] Zeng et al . (2017)  LPSC abstract  1576 . [3] Fulton  et al .  (2017)  Astron. J. 154 , 109 .  [4]  Zeng et al . (2017)  RNAAS 1 ,  32.  [5]  Rogers & Seager (2010) Astrophys. J. 712 , 974 – 991 .  [6]  Adams et al . (2008)  Astrophys. J. 673 , 1160 – 1164 . [7]  Buchhave  et al . (2014)  Nature 509 , 593 – 595 .  [8] Lehmer &  Catling (2017) Astrophys. J. 845 , 130 .  [9]  Owen &  Wu (2017)  ApJ 847 , 29. 

Water-worlds FAQ

1. You say that according to your study, 35% of exoplanets are waterworlds….how many exoplanets is that (is it 35% of 4000, ie, about 1400)?

Answer: Yes, that’s correct, 35% out of 4000. 

2. What data did you have on each of the exoplanets you analysed (eg radius/mass…anything else?), and which telescopes collected each type of data?

Answer: The radius data were mainly collected by the Kepler mission, with important modifications of host stellar parameters by Travis Berger (Univ. of Hawaii, from Gaia mission Data Release 2. 

The mass data were mainly obtained by many different ground-based spectroscopic follow-ups, such as HARPS-North (  ), HARPS (,... and the international collaborations among them. The mass measurements are the most difficult part to do. Among our collaborators, Dr. Andrew Vanderburg from UT Texas, Dr. Mercedes Lopez-Morales from Harvard-Smithsonian Center for Astrophysics (, and Dr. Mario Damasso and Dr. Aldo Bonomo from INAF-Osservatorio Astrofisico di Torino, Italy ( are observers who make these precise mass measurements. 

3. It sounds from the press release as if you ran this data through a special “growth model” which would work out the likely composition of exoplanets of varying sizes. Can you explain how this growth model works, and how it works out the likely composition of a planet just from the radius and mass?

Answer: The mass and radius tell you the average bulk density of a planet. But there could still be ambiguity, since different mixtures of lighter and denser components can produce similar density. That’s where the growth model comes into play, that we systematically compare mass-radius distributions of different planet populations, which reveals a general trend of ~5 Earth-mass rocky core + ~5 Earth-mass ices + additional gaseous envelope, which can connect the small exoplanets all the way up to the gas giant exoplanets. This trend can be interpreted as a general pathway of planet formation. 

Observationally speaking, there is a lack (relative depletion in number) of planets in between 1.8~2 Earth radii. Our model can then naturally explain this lack of planets by the mass and density differences between the rocky planet population and the water world population. 

4. Can you explain in simple terms why the smaller planets would be rocky and the larger ones water-worlds….wouldn’t factors other than just radius and mass affect this?

Answer: To correct this statement a little bit, smaller planets are rocky, intermediate-sized planets are water-worlds, larger planets contain more gas. This reflect the cosmic element hierarchy of three major planet-building materials: Rock (+metal), Ice, and Gas. 

In more detail: in all chemical and physical aspects, the three major planet forming materials: Rocky (including metals), Ice, and Hydrogen/Helium gas, form a hierarchy in their properties: (1) volatility (characterized by equilibrium condensation temperature in the nebula), (2) density, and (3) cosmic abundance, rock (silicates)+metals (Si, Fe, etc.), Ice forming elements (O, N, C), and H2/He always form a ladder or hierarchy, with the elements O, N, C always falling in the middle of this hierarchy:

 Table 1. Hierachy of Planet-Building Elements

  Cosmic Abundance (by mass) Condensation
Temperature (K)
H2/He 1000 1~10 0.2
O, N, C 6+1+3 100~300 2
Mg-Silicates 2 1300~1400 4
Fe, Ni metal 1 1300~1400 8

Therefore, this natural ladder is reflected in planets: as you go from small to large planets, you expect the correlation between the mass or size of planets with their composition, and thus also with the cosmo-chemical sequence.

5. On what assumptions—based on real data about planet formation from our own solar system--do you base your assessments of the composition of all these exoplanets?

Answer: The first direct constraint comes from our understanding of the chemistry of the solar system, which comes from our understanding of meteorites in our own solar system, that my postdoc advisor Professor Stein B. Jacobsen ( is an expert on and our geochemical lab at Harvard analyzes a lot of these meteorite samples from all over the world. The meteorites tell us that the compositions of planets are not arbitrary, but intimately connect to the composition of the central host star, with various degrees of depletion of more volatile elements. The stellar composition itself is not arbitrary but controlled by common nuclear synthesis pathways of elements heavier than Helium in the history of our galaxy, which render similar major planet-building element proportions for most solar-like main-sequence stars. 

Secondly, all planets in the solar system orbit near the same plane. This well-defined plane is the evidence that all planets formed within a disk surrounding the central host star. The most important feature of this disk is the snowline, beyond which, ices can exist as solids and greatly enhance planet formation and growth. In more detail: The disk solid material density jump at the snowlike of water-ice by a factor of 2~3. For planet formation and growth, whether by collisional growth, or gravitational interaction, or other physics, the interaction typically goes as density-squared. So this jump of density is a huge effect (4~9 at least) in facilitating planet growth. Therefore, snowline should act as a factory for these water worlds, or icy cores. No wonder why there are so many of them. :)

Our solar system is in line with the growth model. The four inner planets are rocky and small. The four outer giant planets are thought to have formed near or beyond the snowline, and each is estimated to contain a ~10 Earth mass (rock+ice) core enveloped in various amounts of Hydrogen/Helium gas. Therefore, we speculate these abundant water worlds formed in similar manner as giant planet cores near or beyond the snowline. The only difference is that in their evolutionary history, these water worlds did not have the chance to accrete additional gaseous envelope, but migrate inward to their current location of 0.1~0.2 AU distance to their host stars.

6. Finally, to what extent do your results add to the likelihood that life may have evolved elsewhere in the universe?

Answer: In one sentence, life could be a universal phenomenon, whenever and wherever conditions are right, life will emerge and evolve, it is a natural consequence of energy/information flow of our universe.

One has to realize that, although water (H 2 O) appears to be precious and rarer on Earth and other inner solar system terrestrial planets, it is in fact one of the most abundant substance in the universe, since Oxygen is the third most abundant element after Hydrogen and Helium. Whenever conditions are appropriate, such as on dust grain surface in cold nebula, the Oxygen atom will readily combine with the Hydrogen atoms to form the water (H 2 O) molecule. Here, we have found a population of exoplanets corresponding to water worlds, which are planets consist of significant amount, as much as half of their masses as (H 2 O). These water worlds are mostly in between twice up to four times the size of Earth, but in terms of volume, ranging from 8 up to 64 times that of Earth volume. They are likely formed in similar manner as the cores of gas giants (J, S, U, N) in our own solar system, but they took different pathways in certain stage of formation in the proto-planetary disk, instead of acquiring significant gaseous envelopes, they migrate inward much too fast to become close to their parent stars and stay there ever since. Although they failed in their fate to become the gas giants, they were successful to transit their parent stars, at much higher chance, to catch our attention. They might be one of the most abundant planet types in our galaxy.

Life could develop in certain near-surface layer on these water worlds, when the pressure, temperature, and chemical conditions are appropriate. 

7. Expertise of co-authors:

(1) For Harvard Univerisity Origins of Life Initiative (, Prof. Dimitar D. Sasselov is the director. 
(2) For meteorite analysis, Dr. Michail Petaev is the expert, in Prof. Stein Jacobsen's geochemistry lab ( at Harvard University. 
(3) For statistical analysis and interpretation, Dr. Juan Perez-Mercader and Prof. Gongjie Li are the experts. 
(4) For planet orbital dynamics, Prof. Gongjie Li is the expert.
(5) For planet formation and migration simulations, Matthew Z. Heising is the expert, who is also a PhD candidate in Astronomy & Astrophysics at Harvard University. 
(6) For understanding the physics of water under high pressure in the deep interior of water worlds, Dr. Thomas Mattsson is the expert, who is also the Manager of the High Energy Density Physics Theory Division at the Sandia National Laboratories (
(7) For understanding the surface layer chemistry and physics of water worlds, Dr. Amit Levi is the expert. 
(8) For solar system giant planet interior, Dr. Hao Cao is the expert
(9) For planetary climate and atmospheric evolution (, Prof. Robin Wordsworth is the expert. 

This research is supported mainly by the Simons Foundation ( and the Harvard Origins of Life Initiative (



Trappist-1 Planetary System

Image Courtesy: JPL & NASA

External links:

The nature article is citing our planet interior structure model developed at Harvard EPS (Fig.2 in the article is made based on the Zeng, Sasselov & Jacobsen 2016 model):

Mass-Radius Relation for Rocky Planets based on PREM”. Li Zeng, Dimitar D. Sasselov, and Stein B. Jacobsen. ApJ, 819, 127, 2016. (ADS link) (PDF)


Click on the Images to access the links of following contents:

Harvard Gazette

New York Times

New York Times Report of Li Zeng



Harvard Smithsonian Center for Astrophysics Press Release



Washington Post

Back to Top