GEOPHYSICAL RESEARCH LETTERS, VOL. 17, NO.4, PAGES 449-452, MARCH SUPPLEMENT 1990 CONDENSATION OF HNO3 ON FALLING ICE PARTICLES: MECHANISM FOR DENITRIFICATION OF THE POLAR STRATOSPHERE S.C.Wofsy, R.J.Salawitch, J.H. Yatteau, and M. B. McElroy Harvard University, Cambridge, MA B. W. Gandrud, J.E. Dye, and D. Baumgardner National Center for Atmospheric Research, Boulder, CO Abstract. Ice particles created in polar stratospheric cool- ing events arepredicted to descend •ntoType I PSCs and accrete a coating of nitric acidtrihydrate (NAT) that inhibits evaporation. Coated particles efficiently strip HNO3 from theatmosphere, providinga mechanism for denitrification M'thout significant dehydration. Coatings that disintegrate may release large particles of NAT thatinfluence subsequent .pa•cle growth. Introduction Removal of HNO3 (denitrification) is a key feature of the polar ozone "hole", observed in both the arctic and the antarctic [Fahey et al., 1989; Kawa et al., 1990], but the mechanism is uncertain. Crystalline nitric acid trihydrate (NAT) may form at temperatures 3 to 5 K above the frost point [Type I polarstratospheric clouds (PSCs)] [Toonet al., !98.6; Poole andMcCormick,1988a]. Stratospheric cooling rates exceed 5 K/day on synoptic scales [Tuck, 1989; Jones eta!., 1989], and small scalemotionsinduce fluctuations of 1 to 2 K in a few minutes [Gary, 1989]. Most of the back- ground aerosols are expected to nucleate during condensation in these rapidcooling cycles, forming particles of NAT too small to precipitate [Poole and McCormick, 1988b; Jones et al., 1989; Toonet al., 1989, 1990b; Wilson et al., 1989; Wofsy et al., 1990]. If temperatures fall belowthe frostpoint,ice particles (Type II PSCs) maynucleate onNAT aerosols. Onlya small fraction of ice particles grow large enough to fall [Toon et al.,1989].If NAT is uniformly distributed in small aerosols Mgnificant removal of HNO3by precipitation requires com- p.arable dehydration. Observations showthat denitrification is associated with little or no dehydration in the arctic [Kawa eta!., 1990] and perhaps also in theantarctic [Salawitch et al., 1989; Hamill et al., 1990]. Lidar[Toon et al., 1990a] and in situ [Holmann et al., 1989] data show that some NAT clouds contain small numbers of particles larger than 1gmradius. Salawitch et al. [1989] suggested that preferential nucleation of ice on these particles could set the stage for denitrification, with little associated dehydration. The atmosphere mustcool very slowly to produce such large particles by direct growth on background condensation nuclei[Poole and McCormick, 1988b; Wofsy eta!.,1990; Toon etal.,1990b], but observed Copyright1990 by the AmericanGeophysical Union. •per num•oer 90GL00332. •94-8276/90/9•00332503.00 449 temperature fields may be too perturbed to allow slow growth [Gary,1989; unpublished data,1989]. Thispaper proposes a mechanism for removal of HNO3 by ice particles that descend from Type II PSCs into Type I PSCs. Whenthe atmosphere is cooling, partial pressures of HNO3 exceed equilibrium and precipitating ice particles become coated with NAT, inhibiting evaporation. Coated ice particles efficiently scavenge HNO3, denitrifying the atmo- sphere without significantdehydration. Large particles observed in some Type I PSCs may be ice coated with NAT. Model description A column of air forced upward cools adiabatically, form- ing Type II PSCs at temperatures belowthefrost point. Type I PSCsform at altitudes above and belowtheType II PSCs. Condensation of H20 and HNO3 on existing particles is drivenby vaporpressures in excess of equilibrium values. Excess vapor pressures are maintained as long as tempera- tures continue to decline, and the degree of supersaturation increases as the cooling rate increases. If thereare 10 parti- cles of NAT per cm 3, a molecule of HNO3 hits a particle every I0 to 20 minutes,which is approximately the lag between vapor pressure and particle temperature. For cooling rates between 5 and 10 K/day and unit sticking probability, the saturation ratio for HNO3 (relative to NAT) should be between 1.05and 1.2 [see Wofsy et al., 1990]. Saturation ratios of at least 2, andperhaps greater than5, are requiredto nucleate NAT on background aerosols, equivalent to supercooling by ! to 2 K [Wofsy et al., 1990]. During a cooling event thereshould be a zoneat the base of the cloud where NAT particles are absent because nucleation hasnot yet occurred. Maximum saturation ratios between 2 and5 areexpected in this region [Wofsyet al., 1990]. The size of a particle changes by evaporation of water and condensation of NAT. For a spherical particle sur- rounded by a spherical shell of NAT, the steady-state solu- tionof thediffusion equation for H20 and HNO3 gives [ew 1 r^ 1 - r^(1 - ----- R Pice r}• D'w drB rA 2 drA D'n MNAT dt r•2 dt r BR PNAT (2) where rA is the radius of theice core, t is time, DA is the dif- fusion constant for waterthrough solidNAT (see below), rt! is the particle radius including the NAT coating, D• is the