GEOPHYSICAL RESEARCHLETTERS, VOL. 23, NO. 23, PAGES 3353-3356, NOVEMBER 15, 1996 Loss of H and O from Mars: Implications for the planetary water inventory Helmut Lammer Dept. for Extraterrestrial Physics, Space Research Institute, Austrian Academy of Sciences, Graz, Austria Willibald Stumptner and Siegfried J. Bauer Institute for Meteorology and Geophysics, University of Graz, and Dept. for Extraterrestrial Physics, Space Research Institute, Austrian Academy of Sciences, Graz, Austria Abstract. The evolution of the Mars atmosphere, with regard to water, is influencedby non-thermal es- cape of oxygen atoms. In our evaluation of the various escape processes we recognized that the widely quoted escaperates for oxygen atoms originating from disso- ciativerecombination (exospheric O) are too high by an order of magnitude. Thus the originally postulated couplingof H and O escape in the ratio of 2:1 can no longer be maintained. This has important implications for the stability of the Martian CO2 atmosphere. This study takes into account temperature dependent pho- todissociationcoefficients, optical depths, ozone chem- istry, hydrogen and oxygen escape fluxesfor conditions of an ancient dense Mars atmosphere. Since the only source of hydrogen is water, the calculated hydrogen escapeflux representsan upper limit for an equivalent water loss to space. Under this constraint, we find much lower oxygen escaperates originating from water than current estimates,implying a lossof H20 to spacecor- respondingto an equivalent depth <_ 5 meter over the last 3.5 billion years. Our results cast serious doubt on previousestimates that as much as 50-80 meters of H20 were lost to spaceduring this period. Introduction The present thin Mars atmosphere with a surface pressure of about 7 mbar has been one of the great puzzles of planetary science. Ancient fluvial networks on the surface of Mars suggestthat it was warmer and wetter three billion years ago. Surface features resemblingmassiveoutflow channels provide evidence that the Martian crust contained the equivalent of a planet wide reservoir of H20 several hundred meters deep[Cart, 1987]. There are two possibilities for the fate of early H20 and CO2: they are either sequestered somewhere on the planet or have been lost to space. Since Mars does not have an appreciable intrinsic magneticfield, the atmosphere can be erodedby solar wind interaction. Nonthermal escapeprocesses such as dissociative recombination of molecular ions are known to leadto loss of atomic constituents (O, N, C) on Mars where the gravitational potential is easily overcome by Copyright 1996 by the AmericanGeophysical Union. Paper number 96GL03153. 0094-8534/96/96GL-03153505.00 the energy imparted in the production of hot atoms [Fox and Dal#arno, 1983;Lainmet and Bauer, 1991; Fox, 1993]. Impact erosion [Walker, 1986; Melosh and Vickery, 1989] and late impact accretion [Chyba, 1990; Kargl and Bauer, 1995]are other processes whichcould have played an important role in affecting the early atmo- sphere. Because both the sun [Zahnle and Walker, 1982] and the Martian atmosphere have changed over time, the importance of these evolutionary processes cannot be estimated by simply multiplying the contem- porary loss rates by the ageof the solar system. Models of these loss mechanisms must include the evolution of solar EUV intensity and solar wind effects. Luhmann et al. [1992] calculated sputtering rates of CO2 molecules for 1 EUV (present), 3 EUV (2.5 GYr before present [B.P.]) and 6 EUV (3.5 GYr B.P.) epochs of the Mar- tian atmosphere by reentered O+ pick up ions. They found an integrated CO2 lossequivalentto about 0.14 bar. From the calculated oxygen loss they estimated that Mars has lost an equivalent depth of 50 meters of H20 over the last 3.5 billion years. Kass and Yung [1995] argued that early solarwind conditionsled to an enhancedsputtering lossof atmo- spheric constituents. Their results imply that about 3 bar of CO2 could have been sputtered over the last 3.5 billion years, if Mars had "no" intrinsic magneticfield in the past. On the other hand, if Mars onceshouldhave had an intrinsic magnetic field, sputtering would be a negligible escape mechanism and only loss of O atoms originating from dissociative recombination and upward flowing ionospheric CO} andO• + ions over the polar capsof its magnetosphere could be important [Kar, 1990; Larnrner and Bauer, 1992]. Kass and Yung [1995] obtained an equivalent loss of H20 to space cor- responding to 80 meters of water. In their analysis of exospheric oxygen escape rates, Luhmannet al. [1992] andKass andYung [1995] used the rates of Zhang et al. [1993a] for their calculations. The calculated oxygen escape rateof about 8x 10 • s -• at present [Zhang et al., 1993a] was in apparent agree- mentwith the earlyestimates by McElroy [1972].How- ever, this is more than one order of magnitude larger than those of Lamrner andBauer[1991] (5x1024 s -•) and the even more thorough analysis of hot oxygenes- cape processes of Fox [1993] (2.5x1024 s -•) for simil- iar assumptions. We found that the quoted values of Zhang et al. [1993a] are not consistent with their in- tegrated valuesof the upward moving hot exospheric atomic oxygen fluxes. We thus obtain lowerexospheric 3353