Detection of the Water Reservoir in a Forming Planetary System

See allHide authors and affiliations

Science  21 Oct 2011:
Vol. 334, Issue 6054, pp. 338-340
DOI: 10.1126/science.1208931

This article has a correction. Please see:


Icy bodies may have delivered the oceans to the early Earth, yet little is known about water in the ice-dominated regions of extrasolar planet-forming disks. The Heterodyne Instrument for the Far-Infrared on board the Herschel Space Observatory has detected emission lines from both spin isomers of cold water vapor from the disk around the young star TW Hydrae. This water vapor likely originates from ice-coated solids near the disk surface, hinting at a water ice reservoir equivalent to several thousand Earth oceans in mass. The water’s ortho-to-para ratio falls well below that of solar system comets, suggesting that comets contain heterogeneous ice mixtures collected across the entire solar nebula during the early stages of planetary birth.

Water in the solar nebula is thought to have been frozen out onto dust grains outside ∼3 astronomical units (AU) (1, 2). Stored in icy bodies, this water provided a reservoir for impact delivery of oceans to the Earth (3). In planet-forming disks, water vapor is thought to be abundant only in the hot (>250 K) inner regions, where ice sublimates and gas-phase chemistry locks up all oxygen in H2O. Emission from hot (>250 K) water has been detected from several disks around young stars (4, 5). In the cold (∼20 K) outer disk, water vapor freezes out, evidenced by spectral features of water ice in a few disks (6, 7). However, (inter)stellar ultraviolet radiation penetrating the upper disk layers desorbs a small fraction of water ice molecules back into the gas phase (8), suggesting that cold (<100 K) water vapor exists throughout the radial extent of the disk. The detection of this water vapor would signal the presence of a hidden ice reservoir.

We report detection of ground-state rotational emission lines of both spin isomers of water (JKAKC110-101 from ortho-H2O and 111-000 from para-H2O) from the disk around the pre–main-sequence star TW Hydrae (TW Hya) using the Heterodyne Instrument for the Far-Infrared (HIFI) spectrometer (9) on board the Herschel Space Observatory (10) (Fig. 1) (11, 12). TW Hya is a 0.6 M (solar mass), 10-million-year-old T Tauri star (13) 53.7 ± 6.2 pc away from Earth. Its 196-AU-radius disk is the closest protoplanetary disk to Earth with strong gas emission lines. The disk’s mass is estimated at 2 × 10−4 to 6 × 10−4 M in dust and, using different tracers and assumptions, between 4 × 10−5 and 0.06 M in gas (1416). The velocity widths of the H2O lines (0.96 to 1.17 km s−1) (table S1) exceed by ∼40% those of cold CO (14). These correspond to CO emission from the full 196-AU-radius rotating disk inclined at ∼7° with only little (<65 m s−1) turbulence (17). The wider H2O lines suggest that the water emission extends to ∼115 AU, where the gas orbits the star at higher velocities compared with 196 AU.

Fig. 1

Spectra of para-H2O 111-000 (A) and ortho-H2O 110-101 (B) obtained with HIFI on the Herschel Space Observatory toward the protoplanetary disk around TW Hya after subtraction of the continuum emission. The vertical dotted lines show the system’s velocity of +2.8 km s−1 relative to the Sun’s local environment (local standard of rest).

To quantify the amount of water vapor traced by the detected lines, we performed detailed simulations of the water chemistry and line formation using a realistic disk model matching previous observations (12, 18). We adopted a conservatively low dust mass of 1.9 × 10−4 M and, using a standard gas-to-dust mass ratio of 100, a gas mass of 1.9 × 10−2 M. We explored the effects of much lower gas-to-dust ratios. We followed the penetration of the stellar ultraviolet and x-ray radiation into the disk; calculated the resulting photodesorption of water and ensuing gas-phase chemistry, including photodissociation; and solved the statistical-equilibrium excitation and line formation. The balance of photodesorption of water ice and photodissociation of water vapor results in an equilibrium column of water H2O vapor throughout the disk (Fig. 2). Consistent with other studies (19), we find a layer of maximum water vapor abundance of 0.5 × 10−7 to 2 × 10−7 relative to H2 at an intermediate height in the disk. Above this layer, water is photodissociated; below it, little photodesorption occurs and water is frozen out, with an ice abundance, set by available oxygen, of 10−4 relative to H2.

Fig. 2

Adopted model for the TW Hya protoplanetary disk. (A) H2 number density, (B) dust temperature, (C) the number density of water vapor molecules and contours of volume-averaged water ice abundance decreasing from white to black as 2 × 10−4, 2 × 10−5, 2 × 10−6, 2 × 10−7, and 2 × 10−8, relative to H2, and (D) one quadrant of the resulting water emission line intensity from the near face-on disk, in arbitrary units. In (A) and (B), the blue contour delineates the layer of maximum water vapor abundance.

In our model, the 100- to 196-AU region dominates the line emission, which exceeds observations in strength by factors of 5.3 ± 0.2 for H2O 110-101 and 3.3 ± 0.2 for H2O 111-000. A lower gas mass does not decrease the line intensities, if we assume that the water ice, from which the water vapor derives, formed early in the disk’s evolution, before substantial gas loss occurred, and remains frozen on grains. The most plausible explanation involves a difference in the relative location of small, bare grains regulating the ultraviolet radiative transport and larger, ice-carrying grains. Differential settling of large grains relative to small grains moves much of the ice reservoir below the reach of the ultraviolet radiation, resulting in less water vapor and weaker lines. Our model matches the observations if only 12% of the original ice content remains above this line (20). A radially increasing degree of settling of icy grains explains the observed H2O line widths.

The detected water vapor, resulting from photodesorption, implies an ice reservoir in the giant planet formation zone and beyond. In our simulations, the 7.3 × 1021 g of detected water vapor (equivalent to 0.005 times the mass of Earth’s oceans) originate from a total ice reservoir of 9 × 1027 g (or several thousands of Earth’s oceans) throughout the disk. The size of this reservoir is tied to the dust mass contained in the disk, for which we adopt a conservatively low value. Although the ice reservoir is only observed indirectly, no known mechanism can remove it from the regions probed by Herschel. Any smaller ice reservoir implies the corresponding absence of elemental oxygen that efficiently reforms water ice on the grains.

The detection of both spin isomers of water vapor allows its ortho-to-para ratio (OPR) to be derived, because our simulations indicate that the lines are optically thin. An OPR of 0.77 ± 0.07 matches our observations (12). This value is much lower than the OPR range of 2 to 3 observed for solar system comets (21). It is common practice to associate the OPR with a spin temperature Tspin at which a Boltzmann distribution reproduces the ratio of spin isomers. Our derived OPR corresponds to Tspin = 13.5 ± 0.5 K, whereas the range for solar system comets yields a Tspin of >20 K.

Radiative conversion between spin isomers is not allowed in the gas phase, preserving the OPR for long time scales. Gas-phase formation of water occurs through exothermic reactions leading to an OPR of 3. On grains, water forms and survives at low temperatures, and it is tempting to equate Tspin with the grain temperature. However, the energetics of water formation and ortho-to-para exchange on grains are poorly understood (22), and the water OPR may be changed by photodesorption. This process starts by dissociating water to H and OH in the ice and continues with the energetic H kicking out a neighboring H2O molecule from the ice matrix or with the H and OH recombining in the ice to form H2O with sufficient internal energy to sublimate (23). The latter route drives the OPR to at least unity, implying an even lower original ice OPR, to yield a resulting OPR of 0.77. Cometary volatiles are released through thermal sublimation, and their measured OPRs are interpreted to reflect the OPR of their ice constituents. Equating Tspin with the physical temperature of the grain on which the ice formed is supported by the similarity of measured Tspin of NH3 and H2O in several individual solar system comets (24).

Solar system comets consist of a heterogeneous mixture of ices and solids, likely assembled in the giant planet formation zone by mixing local material with material that drifted in from larger radii (25). Our water vapor observations probe cold, ice-coated precometary grains residing beyond >50 AU, representing the bulk of the latter material. The presence in comets of crystalline silicates, requiring formation temperatures >800 K (26), together with CO and H2O ices that condense at 20 to 100 K, argues for transport of hot material from near the star to the icy outer regions of the solar nebula (27). Provided that spin temperatures reflect formation histories, the different Tspin inferred for the water ice in TW Hya (<13 K) and solar system comets (>20 K) indicates a similar mixing of volatiles throughout the entire solar nebula, blending water formed at >50 K and an OPR of 3 with water formed at 10 to 20 K and an OPR < 1 probed by our observations. In this case, the range of Tspin values of the cometary inventory reflects the stochastic nature of transport and mixing.

Our Herschel detection of cold water vapor in the outer disk of TW Hya demonstrates the presence of a considerable reservoir of water ice in this protoplanetary disk, sufficient to form several thousand Earth oceans’ worth of icy bodies. Our observations only directly trace the tip of the iceberg of 0.005 Earth oceans in the form of water vapor.

Supporting Online Material

Materials and Methods

Table S1

References (28–39)

References and Notes

  1. One astronomical unit (AU) is the mean distance between Earth and the Sun of 1.49598 × 1011 m.
  2. Water in Earth’s atmosphere obstructs ground-based detection of cold water vapor in planet-forming disks. Although Herschel cannot spatially resolve even the closest disk in water ground-state emission lines, HIFI spectrally resolves the H2O line profiles. Comparison with previous spectrally and spatially resolved observations of CO confirms the disk origin of the H2O lines.
  3. Materials and methods are available as supporting material on Science Online.
  4. Acknowledgments: Herschel is a European Space Agency space observatory with science instruments provided by European-led principal investigator consortia and with important participation from NASA. This work was partially supported by Nederlandse Organisatie voor Wetenschappelijk Onderzoek grant 639.042.404, NSF grant 0707777, and, as part of the NASA Herschel HIFI guaranteed time program, NASA. The data presented here are archived at the Herschel Science Archive,, under OBSID 1342198337 and 1342201585.
View Abstract

Stay Connected to Science

Navigate This Article