Hi,
The analysis script of bond connectivity data file (bonds.reax) uses a cutoff for each type of bond. These cutoffs are located in the Cutoff.dic file.
For example, C-N atom pair has a 0.3 bond order cutoff, whereas O-O has a value of 0.65.
In particular, some literature sources use 0.3 for all atom types, whereas more modern papers use different value for each pair of atoms.
I think the shorter (stronger) bonds should use smaller cutoff, so the analysis doesn’t “miss” a proper bonded system, whereas bonds which are typically very lengthy
can use higher cutoffs because they will still be regarded bonded. However, energy terms depend on bond orders, so changing cutoffs could be very tricky issue that can
lead to unphysical results.
So my question is how were these cutoffs determined ? and why same value for all pairs isn’t good ?
David Furman, Graduate Student | Fritz Haber Research Center for
Molecular Dynamics, The Hebrew University of Jerusalem | +972-8656-8909
Hi David,
Replies and comments below, thanks.
Ray
I think the shorter (stronger) bonds should use smaller cutoff, so the
analysis doesn't "miss" a proper bonded system, whereas bonds which are
typically very lengthy
can use higher cutoffs because they will still be regarded bonded.
It is actually the opposite. Stronger bonds have larger bond order
values (a triple bond has a value of 3.0), and weaker bonds have
smaller bond orders. Bond order values less than the cutoffs are not
considered for plotting the bond graphes, so that weaker (longer)
bonds need smaller bond order cutoffs.
However, energy terms depend on bond orders, so changing cutoffs could be very tricky
issue that can lead to unphysical results.
These bond order cutoffs are merely used for analyzing bond
connectivity to identify chemical species in a post-process fashion.
It does not affect atomic trajectories, energies and forces at all.
So my question is how were these cutoffs determined ? and why same value for
all pairs isn't good ?
You can choose whatever reasonable values for the bond order cutoffs
to identify the chemical species from your simulations. A general
value of 0.3 is good for most cases, but can be fine-tuned, as in the
Cutoff.dic case that is tuned for nitramines.