|From:||"Khourshaei Shar, Ali" <akhoursh@...6666...>|
|Date:||Wed, 9 May 2018 21:42:57 +0000|
Thank you for the response. Just for the sake of clarity, if you check BNC.tersoff for ( C B B ) which I attached here, lambda2 of it should be the mean of ( B B B ) and ( C C C ) according to my understanding. For (
B B B ) it is 2.0774982, and for ( C C C ) it is 2.2119 and Surprisingly for ( C B B ) it has been entered as 2.2054 which doesn't follow that mean rule. Am I wrong, or there is an explanation ? I really appreciate your help.
From: Ray Shan <rshan@...33....1795...>
Sent: Wednesday, May 9, 2018 5:14:32 PM
To: Khourshaei Shar, Ali
Subject: Re: [lammps-users] Question on tersoff parameters
This doesn’t matter. In Tersoff potential files a trio (A B C) means the bonding of B—A—C, which in many cases (e.g., Si Si C) have BA or AC bond lengths beyond the cutoff for the compound it parameterized for (SiC in this case). Therefore it doesn’t matter if parameters like A, B, lambda, etc are zeros or not, since Si-Si-C bonds do not exist. For your own potential file you need to determine if such bonds exist or not, but just make them non-zeros to be on the safe side.
I have a very serious problem on building a correct data file for Tersoff potential. Suppose that I want to build it for a system of two type atom, i.e. A and B. It is obvious that data for ( A A A ) and ( B B B ) rows are available in paper. But, how about other 6 rows of data ? I mean ( A A B ) , ... (B B A ). I tried to understand the logic behind it. But, for example in SiC so many parameters are equal with 0. However, when I go through BNC m and check for example ( B B C ), zeros terms are not in agreement with ( Si Si C ) for example. Any comment please ?