[Pw_forum] Pseudopotentials for efg.x

[Josef W. Zwanziger]

Dear Colleagues,

Can someone post the input files that were used to make the pseudopotentials
used in example24 of the examples distribution? This is the example that
uses efg.x to compute the electric field gradients for SiO2. The example
uses the pseudopotential files SiPBE_nc.UPF and OPBE_nc.UPF, and the
reconstruction files Si.recon and O.recon. EFG.X requires the use of
norm-conserving pseudopotentials, and the UPF files seem to suggest that the
ones used are single-projector pseudopotentials. Using the supplied files, I
get about 6 MHz for O-17 in SiO2, which is reasonable, but I cannot
reproduce anything like this using pseudopotentials I make with the ld1
code. The ones I make from ld1 seem in other respects quite good (good
agreement between log derivatives for ae wf's and ps wf's, good agreement of
eigenvalues, better than 1 meV, and for example using my Si ncpp and PBE,
the lattice constant of Si optimizes to 0.5% longer than experiment, which
seems reasonable). But the efg's that result from these pseudopotentials are
terrible. It would help to see in detail how the ones used in the example
were made.

While I'm on the subject of efg.x, can anyone comment on the input units for
Q, the subject of my last post? The data is requested in 10^-30 m^2, but the
example seems to be using barns, which is 10^-28 m^2. What's going on here?

Thanks much,

Joe

Josef W. Zwanziger                                 
Prof. of Chemistry
Canada Research Chair in NMR Studies of Materials
Director, Atlantic Region Magnetic Resonance Centre
Dalhousie University
Halifax, NS B3H 4J3 
Canada

tel: 902.494.1960     net: jzwanzig at dal.ca
fax: 902.494.1867     web: http://jwz.chem.dal.ca