*Bump*

Bump

finding convergence for surface adsorbed species can be extremely difficult. where exactly are you having trouble? what happens to the H atom after several structural operations? does it release completely from the surface? or what? do you get electronic SCF convergence?

several tips that might help are the following: try several initial location guesses, these can be educated guesses, like top/bridge/triple-point locations, or just brute force grid division of initial points. same goes for guess of bond length. also, you can try to fix for instance the Z coordinate of the H atom, and let it relax only in the surface plane (using selective dynamics), and only later release this constraint, same with inner slab layers just to reduce the workload. usually the numerical parameters are not that important for initial convergence, but its worth a shot just to check (encut, ismear, Lreal, kgrid, etc.), but definitely i would write out the wave function in case you need to re-run several geometrical optimizations. and also, start small - set NSW to say 10, and check after 10 iterations to see if things are going well, if not there is no point waiting 300 iterations. 

if on the other hand the problem is converging the wave function, this could be more complicated, and usually is strongly related to the initial input geometry. in this case spin constraints might help, as well as releasing some constraints on electronic convergence just to get an initial wave function to work with.

finally, if all else fails, and the H atom keeps dissociating from the surface, you can also try a smaller slab such that if you find a good geometry in the smaller slab you might have a good guess for the larger slab.

good luck,

Ofer.