Hello,
I have a general question about neb calculations (not related to vtst scripts). If the energy barrier for a particular migration is very high, does it mean that experimentally it is not possible? Because I got a high barrier of 4.0 eV for migration of vacancy from one site to another. But then experimentally i was able to synthesize the structure with vacancies. How then can this be explained given the high migration barrier?
Thank you.
Migration barrier
Moderator: moderators
Re: Migration barrier
It is certainly true that a process with a 4 eV barrier will not occur at room temperature. If you want to upload the calculation, I can take a look at it. There are lots of ways to mess up a calculation and get an unphysical result.
Re: Migration barrier
Dear Prof. Henkelman,
Thank you for your reply. Attached is the calculation I have tried - in short, it is the study of migration of Na atom from one cage to another. And the barrier is 4 eV. But experimentally I was able to synthesize the (almost) Na free structure in the temp range 300-350 C.
I had also tried another case using your recommendations from a previous answer - I used looser settings first to converge the structures and then used the newly obtained structures and re submitted using higher accuracy settings. Even in that case I got a high barrier (3.85 eV).
Thank you for your time.
Thank you for your reply. Attached is the calculation I have tried - in short, it is the study of migration of Na atom from one cage to another. And the barrier is 4 eV. But experimentally I was able to synthesize the (almost) Na free structure in the temp range 300-350 C.
I had also tried another case using your recommendations from a previous answer - I used looser settings first to converge the structures and then used the newly obtained structures and re submitted using higher accuracy settings. Even in that case I got a high barrier (3.85 eV).
Thank you for your time.
- Attachments
-
- lar to sm.zip
- (176.04 MiB) Downloaded 9670 times
Re: Migration barrier
Hello Prof. Henkelman,
Hope you had the time to go over this.
Thank you.
Hope you had the time to go over this.
Thank you.
Re: Migration barrier
Goodness, normally I can find simple problems with simple solutions. In this case, your initial and starting points looks reasonable, the band converged nicely, but the energy barrier is just very high.
Unfortunately, I don't have a simple solution for you. One thing I would check is just to converge your band to 0.005 eV/Ang. It is very unlikely that this will change anything, but it is an easy test. The one thing that I'm thinking is that the path could relax to an intermediate minimum. Again, this is not likely.
The second thing that I would do is to run some high temperature MD until you see a transition (something you can check by periodic minimization) and see if there is a more appropriate final state for diffusion. It's a little hard for me to see in the structure, but if there is a more facile diffusion mechanism, you might see it by running for some picoseconds at temperatures around or above 1000K.
We also have some fancier methods to find reaction pathways (dimer and akmc) and if this turns out to be important, we could help with those kinds of simulations.
One final thought: is it possible that diffusion is mediated by a defect such as a vacancy in the material?
Best of luck, and sorry for the slow response.
Unfortunately, I don't have a simple solution for you. One thing I would check is just to converge your band to 0.005 eV/Ang. It is very unlikely that this will change anything, but it is an easy test. The one thing that I'm thinking is that the path could relax to an intermediate minimum. Again, this is not likely.
The second thing that I would do is to run some high temperature MD until you see a transition (something you can check by periodic minimization) and see if there is a more appropriate final state for diffusion. It's a little hard for me to see in the structure, but if there is a more facile diffusion mechanism, you might see it by running for some picoseconds at temperatures around or above 1000K.
We also have some fancier methods to find reaction pathways (dimer and akmc) and if this turns out to be important, we could help with those kinds of simulations.
One final thought: is it possible that diffusion is mediated by a defect such as a vacancy in the material?
Best of luck, and sorry for the slow response.
Re: Migration barrier
Dear Prof. Henkelman,
Thank you for your reply and for your suggestions. I shall try them out.
As a follow up to the high temperature MD - if the facile diffusion mechanism is visible only > 1000K would it still be valid for my case since I see the material at ~ 350 degree C? (~600 to 700 K)?
I do think that the diffusion could be mediated by vacancies in the material.
Thank you.
Thank you for your reply and for your suggestions. I shall try them out.
As a follow up to the high temperature MD - if the facile diffusion mechanism is visible only > 1000K would it still be valid for my case since I see the material at ~ 350 degree C? (~600 to 700 K)?
I do think that the diffusion could be mediated by vacancies in the material.
Thank you.
Re: Migration barrier
The use of high temperature is just a trick to find possible diffusion mechanisms.
If your material is active at 650 K on a time scale of seconds, that would imply a diffusion barrier of around 1.5 eV. But of course, you can only simulate maybe 10 ps, so if you want to see that same 1.5 eV barrier, you will have to simulate at a temperature of above 2000K. But I would start with a somewhat lower temperature if you don't see any diffusion mechanism in the time scale that you can afford to run MD, then heat it up until you do see something. Again, the point is not the MD simulation to model dynamics, it is a way to find sensible mechanisms so that you can go back and calculate barriers to get diffusion rates at the temperature that you care about - closer to 650 K.
If your material is active at 650 K on a time scale of seconds, that would imply a diffusion barrier of around 1.5 eV. But of course, you can only simulate maybe 10 ps, so if you want to see that same 1.5 eV barrier, you will have to simulate at a temperature of above 2000K. But I would start with a somewhat lower temperature if you don't see any diffusion mechanism in the time scale that you can afford to run MD, then heat it up until you do see something. Again, the point is not the MD simulation to model dynamics, it is a way to find sensible mechanisms so that you can go back and calculate barriers to get diffusion rates at the temperature that you care about - closer to 650 K.