In a few sentences,
the presentation of both
RMC and RDM methods
(personal views) :
The Reverse Monte Carlo method (RMC) is now widely used for structural modelling.
RMC produces glass structure models tending to the very best fit of diffraction data, almost in a systematical way.
The composition and density
minimal/maximal interatomic distances should be known, and
coordination numbers are welcome when dealing with network glasses.
Model size involves at
least 1000 atoms.
Modelling may start from random atomic positions.
In case of network glasses, the building of a model satisfying the coordination constraint requirements (for instance, a 4-connected three-dimensional network for SiO2 or ZnCl2) may become tedious and require a final by-hand intervention.
The models tend to present
non-requested features such as tetrahedra edge sharing, pending oxygen
atoms, exception to the expected coordination.
On the other hand, the Rietveld for disordered material method (RDM) needs a crystal structure as a starting mean model, so that the coordination constraints are respected ab initio.
Glasses or nanocrystalline material diffraction data are fitted using microstrain effects on line broadening.
Model size involves less than 25 independent atoms of which the coordinates are refined.
The approach is controversial...
Describing glasses by space groups different from P1...
Reconciling RMC and RDM
How to reconcile both methods, that lead to quite different models, is the aim of this conference.
Is examined the behaviour of RDM models, enlarged to nearly one thousand atoms or more (by multiplying the cell edges), and submitted to limited random moves by the RMC method.
Three examples are presented :
glassy SiO2, ZnCl2 and NaPbM2F9 (M = Fe, V), all previously submitted to RMC and RDM modelling.