Radiation Damage

LAMMPS LAMMPS Mailing List Mirror
1

Depending on what you mean by "radiation effects," there are ways. I
typically identify Frenkel defects in BCC materials through
post-processing, for example. I have written LAMMPS compute and dump
classes that are capable of doing this right within LAMMPS, but I'm still
testing them and they are very much alpha releases at this point. For all I know, the classes I wrote may only work on BCC materials; possibly the
techniques it uses only make sense for BCC (or at least cubic) materials,
though though they're intended to work on any lattice LAMMPS can generate,
rotated or otherwise.

If there is strong interest in this, I can think about moving the
development process forward more quickly, but for now this has very much
been a side project for me.

One reason I'm hesitant to throw this type of thing out there is that once
it's implemented, there are always going to be people who use it without
any understanding whatsoever of how it works (and more importantly what
can go wrong).

Karl D. Hammond
University of Tennessee, Knoxville

2

Karl,

3

Oh, I agree---that's not the reason I haven't released it. The real
reason is that it didn't work at all until very recently. :slight_smile: And
because it's not directly used by me, meaning I'm not testing it every day
or something similar.

My (perhaps overly cautious) statement is more a statement that something
like this involves extra work documenting the possible problems, even
after the implementation itself is finished. And THAT is much harder to
do---if you're too verbose, no one reads it, and if you say nothing, you
(at some level) have opened the door to bad interpretations. Not your
fault, of course, but still not something for which to strive.

Karl D. Hammond
University of Tennessee, Knoxville
[email protected]...

"You can never know everything, and part of what you know is always
   wrong. A portion of wisdom lies in knowing that. A portion of courage
   lies in going on anyway."
"Nothing ever goes as you expect. Expect nothing, and you will not be
   surprised."

4

In “Molecular dynamics simulation or radiation damage in CaCd6 quasicrystal cubic approximant up to 10keV” by Chen et. al,

They talk about how interstitials and vacancies can be found using the Voronoi method which was mentioned earlier in this thread. (I quoted the paper below). They also used pe/atom which would work as well. Karl, that software would work really well for me if it is indeed about ready to go.

“For the current work, we found that two approaches were
practically useful for damage analysis. The methods chosen
were potential energy and Voronoi cell analysis. In the potential energy method, those atoms with a potential energy
higher than the equilibrium value were labelled as defects.
This is a convenient way to identify the damage zones, but it
does not give detailed information on the defect structures on
an atomic level. The Voronoi cell method was used to rec-
ognize interstitials and vacancies by analysing the atom posi-
tions with respect to the lattice defined by the original undis-
turbed simulation cell. A lattice site with an empty Voronoi
cell was labelled a vacancy and a cell with multiple atoms an
interstitial. Since irradiation damage in quasicrystals has not
been analysed earlier, we report results for both methods to
allow comparison.”

Ben

5

One reason I’m hesitant to throw this type of thing out there is that once
it’s implemented, there are always going to be people who use it without
any understanding whatsoever of how it works (and more importantly what
can go wrong).

What’s the worst that can happen? Someone uses it, gets upset,
hunts you down, and breaks your knee-caps. That’s what alias
gmail accounts are for.

Seriously, if you write a doc page, we’re happy to release it
in USER-MISC, and as Carlos said, it’s “caveat emptor”.

Steve

6

One reason I'm hesitant to throw this type of thing out there is that once
it's implemented, there are always going to be people who use it without
any understanding whatsoever of how it works (and more importantly what
can go wrong).

What's the worst that can happen? Someone uses it, gets upset,
hunts you down, and breaks your knee-caps. That's what alias
gmail accounts are for.

right, by that token, *i* should already wearing some nice concrete
overshoes and explore the mutations floating through the schuykill
river. not that i want to give anybody ideas here... :wink:

axel.

7

You guys know how to make my day. :slight_smile:

-Karl

8

right, by that token, i should already wearing some nice concrete
overshoes and explore the mutations floating through the schuykill
river. not that i want to give anybody ideas here… :wink:

This from a guy who now hangs out in Italy (for protection?)

Steve

9

right, by that token, *i* should already wearing some nice concrete
overshoes and explore the mutations floating through the schuykill
river. not that i want to give anybody ideas here... :wink:

This from a guy who now hangs out in Italy (for protection?)

indeed. i was made an offer that i could not refuse.

axel.

10

I have been studying the voronoi/atom feature in LAMMPS. Is it true that a lattice site with an empty Voronoi cell is a vacancy and a lattice site with multiple cells is an interstitial? If so, I am confused since most research papers I have read state that LAMMPS does not provide the ability to analyze Frenkel defects.

Ben

11

Ben,
the voronoi/atom compute does not work in that way. It only computes
the voronoi tessellation of the current atom arrangement.
What the rad damage application of voronoi cells needs is an
occupation analysis of the atoms in the current (post irradiation)
frame with respect to the tessellation of the pristine lattice.
Daniel

12

So in other words, LAMMPS provides absolutely no methods of analyzing Frenkel pairs - meaning this needs and has to be done in self-written software?

Ben

13

Correct. Doing it for one particular lattice is usually not all that
difficult---it's doing it across processes and/or for generalized lattice
vectors that may take some doing. But you don't have to do it that way.
Post-processing is typically more straightforward, as you only have one
process to deal with.

I do plan to submit my classes that do this on-the-fly to USER-MISC, but
as I said, they're not quite ready yet.

-Karl

14

A few things to think about:

a) LAMMPS has no concept of a “vacancy”. It
only knows about particles. You keep referring
to a “lattice” as if it stores locations. In LAMMPS
it doesn’t. It is just a geometric definition of
a unit cell. Other commands, like create_atoms,
use it to generate sites and create atoms, but
that is all.

b) even if you had lattice sites, LAMMPS models
many kinds of systems that have nothing to do
with a single lattice. E.g. amorphous systems,
systems with extended “defects”, e.g. grain
boundaries where mutiple lattices would define
different regions of the systems, interfaces where
2 materials come together. How would you imagine
defining Frenkel pairs for those kinds of systems?

c) Lattices are not static in dynamical systems. E.g.
if you heat a system or have a free surface or run
NPT so the box volume changes, the lattice sites
are not static. How are you imagining LAMMPS would
track these fictitious sites in those scenarios?

Having said that there are various LAMMPS commands
that analyze atoms and their neighborhoods and
can tell you something about point defects.

compute cna/atom, compute voronoi/atom, compute coord/atom
compute centro/atom, compute displace/atom, fix store/state,
atom-style variables

If you’re creative you can get info out of those that would help
you identify Frenkel pairs or other kinds of defects.

E.g. you could use compute displace/atom to track
how far atoms move from their original positions. And
dump those that move further than some threshhold distance. If
you also dump the original position (e.g. from fix store/state),
then you have both the interstitial coords and the original (vacancy)
coords.

Failing that, if you describe precisely what you want to
calculate to identify a “Frenkel pair”, then developers can
advise you as to whether it is possible. But keep in mind
it needs to make sense for (a), (b), © above to be of
general use.

Steve