Damping constant

[Suman Mishra]

Hi all,

I am doing Curie Temperature calculation for 2D materials using Vampire software. I have one trivial doubt about damping constant. Does anyone know how to choose a damping constant for specific material? Any suggestion is highly appreciated.

Thanks

Suman

[milton...@gmail.com]

If I had to find the damping constant of a material then I would look for FMR measurements of that material, since the FMR linewidth depends linearly on the damping constant. Or you could try to find out what other people have used when simulating the same material.

If you do Monte Carlo simulations then you don't have to worry about the damping constant. For dynamical simulations of Tc you should set critical damping, alpha=1.0, to try to compensate for the speed of the temperature sweep (0 to 1000 K in 1 ns or less). Lower damping will make it seem like the Tc is higher than it actually is, just because the spins didn't have time to relax.

[Suman Mishra]

Does anyone have information regarding this?

---------- Forwarded message ---------
From: Suman Mishra <mishras...@gmail.com>
Date: Tue, Jan 25, 2022 at 11:18 AM
Subject: Re: Damping constant
To: milton...@gmail.com <milton...@gmail.com>

Also in the case of Monte Carlo simulations, are you using it in the serial version? Because I checked in tutorials of Vampire it is not parallel so it takes lots of time for systems with large number of atoms in a unit cell. Apart from that in the case of LLG, does the Curie temperature depend on the size of the system? For example I have a unit cell of 7x7 Angstrom and I used a 5nmx5nm sheet for simulation. Is it possible that if I use a 10nm sheet curie temperature will differ from what I got in 5nm?

Thanks

On Tue, Jan 25, 2022 at 10:59 AM Suman Mishra <mishras...@gmail.com> wrote:

Thanks for this nice note, so to get spin relaxed I need to set alpha=1.0 and I don't need to worry about material damping constant?

--
You received this message because you are subscribed to the Google Groups "Vampire Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to vampire-user...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/vampire-users/2d98080d-affb-477a-a9e5-25c2d5d8139dn%40googlegroups.com.

[milton...@gmail.com]

There is only serial Monte Carlo but you can in a way parallelize it by running many small simulations at the same time, just don't forget to change the seeds for the random number generators if you're doing many repetitions of the same simulation. When it comes to scaling effects you can find the true Curie temperature with "cumulant crossing methods". Basically there is a quantity called the 4th order Binder cumulant which, when plotted versus temperature for different size systems, will intersect precisely at the Curie temperature. Vampire doesn't calculate the Binder cumulant currently but you can calculate it yourself from time averages of the magnetization if you simulate each temperature separately with "time-series" simulations. I think this is only for Monte Carlo though. For LLG the size also affects the Curie temperature but I don't know how you would extract the true one. Also, as mentioned in the slides for the first Vampire workshop, you should use a 0.1 fs time step when you try to find the Curie temperature with LLG simulations, instead of the usual 1.0 fs.