Detection of Water in Interstellar Regions by its Microwave Radiation

by

A. C. CHEUNG
D. M. RANK
C. H. TOWNES

Department of Physics, University of California, Berkeley

D. D. THORNTON
W. J. WELCH

Radio Astronomy Laboratory and Department of Electrical Engineering, University of California, Berkeley

A report of the detection of microwave radiation from water molecules in space, by the group which recently detected interstellar ammonia emmission.

Microwave emission from the C₁₆5₂₃ rotational transition of H₂O has been observed from the directions of Sgr B2, the Orion Nebula and the W49 source. This radiation, at 1.35 cm wavelength, was detected with the twenty foot radio telescope at the Hat Creek Observatory using techniques described earlier for the detection of the NH₃ spectrum¹. In the case of Sgr B2, the H₂O emission is from the same direction in which considerable NH₃ is observed (unpublished work of A. C. C. et al.), although there is reason to believe the two molecular species may not be closely associated. Strong H₂O radiation producing an antenna temperature of 14° K is observed from the Orion Nebula (where no NH₃ was detected), and an antenna temperature at least as high as 55° was found for H₂O radiation from W49.

Fig. 1 shows the spectral intensity near the H₂O resonance in the approximate direction of Sgr B2, the antenna being pointed at the position a₁₉₅₀ = 17 h 44 m 23 s. d₁₉₅₀ = –28° 25′. The observed antenna temperature is plotted as a function of Doppler velocity. The mean Doppler velocity of about 70 km s^–1 for the spectral line observed is not very different from the 68 km s^–1 mean velocity found for one of the OH emission and broad OH absorption features observed in this region², the 62 km s^–1 Doppler velocity of a small nearby HII region³, and the velocity of about 58 km s^–1 found for NH₃, (unpublished work of A. C. C. et al.) observed in this direction. The results shown in Fig. 1 were obtained with filters producing a spectral resolution of about 1.3 MHz.

Fig. 1 Observed spectral intensity of 616→523 H2O rotational transition in the direction of Sgr B2 at a1950·0 = 17 h 44 m 23 s ± 6 s, d1950·0 = 28° 24·9′ ± 1′. | high-resolution version |

Fig. 2 shows the antenna temperature as a function of Doppler velocities observed in the Orion Nebula at a₁₉₅₀ = 5 h 32 m 57 s ± 4 s and d₁₉₅₀ = –5° 25.5'± 1.0'. In Orion, the radiation intensity was sufficiently high to make practical the use of filters producing a spectral resolution of about 350 kHz. In Fig. 2 the solid line represents the continuum temperature as it was measured with filters of width 2 MHz; the plotted points represent observations made with the filters having bandwidths of 350 kHz. These points are, in some places, closer together than the filter widths of 350 kHz because measurements were made during different runs in which the central frequencies of the filters were displaced in order to examine the structure of the spectral line, which is composite, representing at least two distinct Doppler velocities. The mean Doppler velocity of about + 7 km s^–1 found for the water line is somewhat different from the HII Doppler shifts in the range of –2 to –6 km s^–1 found in the same direction^3,4. OH radiation from this direction shows Doppler velocities near 20 km s^–1 and 7 km s^–1 (ref. 2); the latter coincides more closely with the H₂O velocity.

Fig. 2 Observed spectral intensity of 616→523 H2O rotational transition in the direction of the Orion Nebula. The position of peak emission is a1950·0 =5 h 32 m 57 s ± 4 s, d1950·0 = –5° 25.5′ ± 1.0′. | high-resolution version |

Fig. 3 shows similar data for the still more intense source W49, at a₁₉₅₀ = 19 h 7 m 55 s ± 4 s and d₁₉₅₀ = 9° 0.4′ ± 1′. Over most of the spectral line, the narrow filters of width 350 kHz were used and several narrow and intense features were revealed. The measured antenna temperature was as high as 55° K, but allowance for atmospheric attenuation of this source, which was rather low in the sky during the observations, raises its effective antenna temperature to somewhat more than 70°. Presumably, narrower filters would give still higher antenna temperatures at some frequencies. The solid lines on either side of the diagram show the radiation intensity observed with filters about 2 MHz wide. Radiation present on the extreme right of Fig. 3 seems to be above the continuum level, showing that H₂O radiation also occurs within the band of this last wide filter. The Doppler shift of OH in W49 ranges from –2 km s^–1 to + 23 km s^–1 (ref. 2) and that of the HII region is + 6 km s^–1 (ref. 3). These are in the same general range, but by no means similar to the dominant values of Doppler shifts shown in Fig. 3.

Fig. 3 Observed spectral intensity of H2O transition in the direction of W49. The position of peak emission is a1950·0 = 19 h 07 m 55 s ± 5 s, d1950·0 = + 09° 0.4'± 1'. | high-resolution version |

In all three figures, the continuum temperatures should be scaled down by a factor of two, because the heterodyne detection system received continuum radiation in both side bands.

The Orion and W49 sources were not large enough in angular size noticeably to broaden the beam width of 8.8'in drift scans across the sources. Thus we conclude that these radiating sources are not larger in angular diameter than 3'.

The radiation found is attributed to H₂O because its frequency coincides very closely to that found for H₂O in the laboratory, and no other known atomic or molecular species can explain the observations. The Doppler shifts plotted in the figures represent small departures from the laboratory frequency of 22,235.22 MHz for H₂O, for which 10 km s^–1 corresponds to a frequency shift of 0.75 MHz. No other simple molecular or atomic species which can be expected in interstellar space is known to produce a line within several tens of MHz of this frequency except NH₃, the (3,1) inversion transition of which lies at 22,234.53 MHz. This is a rather weak NH₃ transition, but it is still important to examine carefully the possibility that the observed radiation might be caused by NH₃. A search was therefore made for other stronger transitions of NH₃. The (3,3) transition is somewhat more than an order of magnitude stronger than this (3,1) transition at excitation temperatures above 50° K, while the (1,1) transition is similarly more intense at temperatures below 50° K. Even at excitation temperatures considerably above 50° K, the (1,1) transition is always more than twice as intense as the (3,1) transition. A search for these two NH₃ transitions in the Orion Nebula was made. They were not found, with the upper detection limit of antenna temperature being set at less than 0.07° K for both the (1,1) and the (3,3) inversion transitions. This limit is almost two orders of magnitude less than the intensity of radiation found at about 22,235 MHz, and hence we conclude that the radiation we have observed cannot be caused by the (3,1) transition of NH₃, and can be explained only by H₂O.

In the case of Sgr B2, the situation is more complex. Rather strong NH₃ radiation was, in fact, observed from the same (or nearly the same) direction as the H₂O radiation. The (1,1), (2,2), (3,3) and (4,4) inversion transitions of NH₃ were all found with antenna temperatures between 0.5° K and 1.5° K (unpublished work of A. C. C. et al.). The possibility that the line observed at 22,235 MHz is caused by NH₃ may be ruled out by the following argument, however. The NH₃ rotational states (1,1), (2,2), (3,3) and (4,4) can be excited by collision, but are metastable, with exceedingly long radiation lifetimes. The (3,1) state can radiate in the infrared, however, by making a transition to the (2,1) state. The mean radiation lifetime of the upper state is about 50 s; thus unless the state is excited in a comparable time, it should be very much less populated than any of the states which were observed. The (4,3) state also radiates with a lifetime comparable with that of the (3,1) state. Thus a search was made for (4,3) level inversion radiation to find out whether or not some excitation mechanism was producing a significant population in the (4,3) level and, by inference, also producing an appreciable number of NH₃ molecules in the (3,1) state. The (4,3) inversion radiation was not observed, the limit of detection being an antenna temperature of 0.07° K. The (4,3) transition searched for is about an order of magnitude more intense than the (3,1) transition. If the observed antenna temperature of 1° K in Sgr B2 at 22,235 MHz were caused by the (3,1) transition, the (4,3) transition would have an antenna temperature of 10° K. No such temperature is observed, so we conclude that the observed radiation at 22,235 MHz must be caused by H₂O rather than this weak NH₃ transition.

In the case of W49, a search was made for the NH₃ (1,1) inversion transition. It was not found, and an upper limit of 0.5° could be established for the antenna temperature at its frequency.

There is, of course, H₂O in the Earth's atmosphere. One might wonder if it could produce the observed radiation. It can be eliminated as a possible source for several reasons. One is that the antenna beam is switched from one part of the sky to a neighbouring part, so that only radiation which varies very rapidly with angle is detected. A second reason is that the microwave resonant line was detected in the three particular fixed directions in space, and not nearby. Still a third is that an atmospheric water line would have occurred at the laboratory frequency, which corresponds to a Doppler shift of + 15 km s^–1 in Fig. 1 and –30 km s^–1 in Fig. 2.

It has previously been suggested that the 6₁₆5₂₃ transition of H₂O could be of interest to radio astronomy⁵, and recently a substantial proposal for its detection has been made by Snyder and Buhl at the AAS meeting, Dallas, December 1968. But it is surprising that the transition is as strong as observed, because it involves levels of rotational energy of 456 cm^–1 which can radiate to lower states in about 10 s. They require moderately high temperatures and frequent excitations, either by collisions or by radiation. Rotational states of NH₃ which can similarly radiate have not been found, indicating that the NH₃ and H₂O have been detected in rather different regions. Presumably the H₂O is present in rather special regions of higher than normal excitation The matrix element for this H₂O line is appreciably less than that for the NH₃ inversion levels. Thus if there is thermal equilibrium between the 6₁₆ and 5₂₃ states, the population in these two levels, rather high above the ground state, must have a column density greater than that found for NH₃ (ref. 1), or about 10¹⁷ cm² in the Sgr B2 cloud.

In Orion, the actual microwave brightness temperature of the source would have to be at least as high as a few hundred degrees, and in W49 at least as great as about one thousand degrees. The actual temperatures may be much higher if these sources are smaller than the upper limit of 3'of are given, or if they are optically thin. The high intensity, very narrow, lines suggest that perhaps thermal equilibrium does not occur and that there may even be maser action. Further study of the distribution and condition of H₂O in interstellar space is clearly needed. If thermal equilibrium does in fact apply, this intense microwave radiation from H₂O and the existence of strong HDO transitions in the radio region should allow an interesting measurement of the hydrogen–deuterium abundance ratio.

We thank Professor Harold Weaver for giving us his positions of OH clouds in the Sagittarius region and for discussions, and Paul Rhodes for his help with the observations. This work is supported in part by NASA, the US Office of Naval Research and the US National Science Foundation.

Received January 13, 1969.

1. Cheung, A. C., Rank, D. M., Townes, C. H., Thornton, D. D., and Welch, W. J., Phys. Rev. Lett., 21, 1701 (1968).
2. Weaver, H. F., Dieter, N. H., and Williams, D. R. W., Ap. J. Suppl., 16, 219 (1968).
3. Mezger, P. G., and Höglund, B., Ap. J., 147, 490 (1967).
4. Gordon, M. A., and Meeks, M. L., Ap. J., 152, 417 (1968).
5. Townes, C. H., in Fourth IAU Symp., Manchester, 1955 (edit. by van de Hulst, H. C.) (Cambridge Univ. Press, 1957).

| return to looking back index page |