Molecular dynamics simulations of D2O ice photodesorption.

Molecular dynamics (MD) calculations have been performed to study the ultraviolet (UV) photodissociation of D(2)O in an amorphous D(2)O ice surface at 10, 20, 60, and 90 K, in order to investigate the influence of isotope effects on the photodesorption processes. As for H(2)O, the main processes after UV photodissociation are trapping and desorption of either fragments or D(2)O molecules. Trapping mainly takes place in the deeper monolayers of the ice, whereas desorption occurs in the uppermost layers. There are three desorption processes: D atom, OD radical, and D(2)O molecule photodesorption. D(2)O desorption takes places either by direct desorption of a recombined D(2)O molecule, or when an energetic D atom produced by photodissociation kicks a surrounding D(2)O molecule out of the surface by transferring part of its momentum. Desorption probabilities are calculated for photoexcitation of D(2)O in the top four monolayers and are compared quantitatively with those for H(2)O obtained from previous MD simulations of UV photodissociation of amorphous water ice at different ice temperatures [Arasa et al., J. Chem. Phys. 132, 184510 (2010)]. The main conclusions are the same, but the average D atom photodesorption probability is smaller than that of the H atom (by about a factor of 0.9) because D has lower kinetic energy than H, whereas the average OD radical photodesorption probability is larger than that of OH (by about a factor of 2.5-2.9 depending on ice temperature) because OD has higher translational energy than OH for every ice temperature studied. The average D(2)O photodesorption probability is larger than that of H(2)O (by about a factor of 1.4-2.3 depending on ice temperature), and this is entirely due to a larger contribution of the D(2)O kick-out mechanism. This is an isotope effect: the kick-out mechanism is more efficient for D(2)O ice, because the D atom formed after D(2)O photodissociation has a larger momentum than photogenerated H atoms from H(2)O, and D transfers momentum more easily to D(2)O than H to H(2)O. The total (OD + D(2)O) yield has been compared with experiments and the total (OH + H(2)O) yield from previous simulations. We find better agreement when we compare experimental yields with calculated yields for D(2)O ice than when we compare with calculated yields for H(2)O ice.