1. THE SODIUM EXOSPHERE OF MERCURY: COMPARISON BETWEEN OBSERVATIONS AND MODEL A.Mura, P. Wurz, H. Lichtenegger, H. Lammer, A. Milillo, S. Orsini, S. Massetti, H. Schleicher, M. Kodachenko Abstract. In this study we compare the sodium observations made by Schleicher et al. (2004) with the result of a numerical simulation. The observations, made during the transit of Mercury across the solar disk on May 7, 2003, shown a maximum of sodium emission near the polar regions, with north prevalence, and the presence of a dawn- dusk asymmetry. We interpret this distribution as the resulting effect of two combined process: the s/w proton precipitation causing chemical alteration of the surface, freeing the Na atoms from bounds in the crystalline structure on the surface, and the subsequent photon stimulated desorption of the Na particles. The observed and simulated distributions agree very well, indicating that the proposed process is able to explain the observed features.

2. Observations 2 Mangano et al., 2007 Potter et al., 2002 2

3. More observations… 3 3

4. Transit observations (1) 4 4 From Schleicher, 2004

5. Transit observations (2) 5 Observed Na column density. Data from Schleicher et al., (2004), obtained during the Mercury transit of May. 7, 2003. Y and Z axis are orientated according to the MSE frame i.e. Z is positive towards north; Y is positive towards dusk. Max density2500 cm -3 Max Col. den.7  10 10 cm -2 Total amount3  10 27 Scale height200  500 km parallel doppler width 1.6 km/s 5

6. Transit observations (2) 6 6 Anomaly150° Distance0.45 AU S/W velocity700 km/s IMF-20, 10, -10 nT Radiation pressure-60 cm/s 2 Measured by ACE, during May 7-9, 2003, distance from spacecraft to Earth, components of magnetic field, solar wind proton speed and density (http://www.srl.caltech.e du/ACE/ASC/level2/ind ex.html).www.srl.caltech.e du

7. Influence of IMF 7 Negative Bx component of IMF causes reconnection in the North Emisphere (Sarantos et al., 2003, Kallio et al, 2003, Massetti et al., 2005) 7 Reconnection in the North Emisphere causes higher S/W proton precipitation fluxes North South

8. Simulated H + flux 8 Day Night North-South Asymmetry Simulated precipitation flux using Montecarlo single-particle model, 10 6 test- particles/run 8

9. Using only Ion sputtering… 9 9 1st Run: S/W precipitation causes Na ion sputtering Results: scale height too high, density too low (factor 100), no dawn dusk asymmetry Y(RM) col dens. (cm -2 )

10. Other model 10 Surface element Na TD PSD S/W 1) Thermal desorption and PSD depleted Na contents 2) S/W causes chemical alteration of the surface, freeing the Na atoms from bounds in the crystalline structure on the surface 2H + Na2SiO3 → 2Na + SiO2 + H2O Production of sodium and water by proton sputtering of sodium- bearing silicates was considered by the following mechanism (Potter, 1995) Proton precipitation Montecarlo model Surface evolution model Sodium exospheric Montecarlo model

11. Sodium variability 11 Dawn Dusk Rotation 11 Ions PSD Ions TD PSD Ions TD

12. Some equations… 12 12 24 x 48 elements surface grid, time step = 10 m total simulation time: 2 Mercury years S/W flux: from numerical model PSD flux: TD flux:

13. …more equations. 13 13 The surface elements is “moved” and the following equation is solved numerically:  loss = k  spu If we assume that particles released by TD always fall back onto the surface, this process does not contribute to the net flux from the surface; however, the Na atoms fall back within an area of radius <300 km. Thus, thermal desorption will also cause a smearing of the places of Na release on the day side. This effect has been simulated using 10000 test-particle for each time-step and k takes into account the overall process yield and the probability that the proton found a Na atoms in the surface, considering the fraction of Na bearing minerals in the regolith (Wurz et al., 2008). For the elements in the dayside:

14. Time evolution of Na: first hour 14 14 TSC = N /  loss

15. Time evolution of Na: a year 15 15

16. H + and Na fluxes: comparison 16 16 Na flux from the surface. The blurring is due to Thermal desorption H + flux onto the surface

17. Finding optimal parameters 17 17 The simulation parameters can be tuned to better match the observed quantities. Since the PSD flux is directly proportional to the H + one, the main parameter is the energy distribution (controlled by the parameter U) The energy distribution of the source controls also the scale height and the density of Na Here we have chosen to use a value for U in order to match the wavelength dependence of the excess adsorption, which is related to the Doppler velocity shift along the line of sight

18. Finding optimal parameters 18 18 Parallel (x) velocity distribution of the simulated particles (blue line). The simulated data can be fitted by using a gaussian function:  exp (-v 2 /v th 2 ), with v th = 1.4 km/s. The observed velocity distribution can be reproduced by a gaussian function with with v th = 1.4 km/s (in red). From Schleicher et al. 2004

20. Results 20 20 ParameterDataSimulation Density (max)2500 cm -3 1000 cm -3 Column dens. (max) 7  10 10 cm -2 9  10 10 cm -2 Total amount 3  10 27 9  10 27 (*) Scale height 200  500 km ~1000 km parallel doppler width1.6 km/s1.4 km/s OBSERVATIONS SIMULATION 2 10 6 test particle run

21. Model limits 21 21 A critical hypothesis is to assume that the proton precipitation flux (on the night side) is constant over a very long time-scale (~ weeks): not realistic; the time-scale for the equilibrium between H + and PSD fluxes, on the dayside, is very short (~ one hour); the north-south asymmetry is due to H + precipitation on the dayside, which rapidly results in an enhancement of Na density in the high latitude regions; the dawn-dusk asymmetry is caused by the planetary rotation and by the H+ precipitation on the night side; such a precipitation is predicted for most of the IMF conditions (for example, Kallio et al., 2003 )

22. Results (2) 22 22 From Kallio and Janhunen, GRL, 2003 From Delcourt et al, AG, 2003

23. Results (2) 23 23 Including a uniform precipitation of 10 7 cm -2 s -1 on the night side (corresponding to a Na flux of 10 6 cm -2 s -1 ), we obtain a better fit of the dawn-dusk asymmetry. OBSERVATIONS SIMULATION 10 5 test particle run

24. Conclusions 24 24 1)Neither PSD, nor IS alone are able to explain the observed features 2)There is a very good agreement between: Column densities Scale heights Doppler widths 3) Effect of other ion precipitation, and more refilling mechanism are going to be added 4) A paper has be submitted to Icarus; poster to EGU: EGU2008-A-09131