What is the correct way to calculate the enthalpy of a solid as a function of temperature using molecular dynamics?

31 January 2024 2 8K Report

I would like to get the enthalpy as a function of temperature for BCC lithium at zero pressure.

I have performed a series of NVT simulations with 500 atoms using a Nose-Hoover thermostat at the corresponding equilibrium volumes (found using the volume average of NPT simulations) and calculated the enthalpy as 𝐻=𝑈+𝑝𝑉 which at zero pressure is just the total energy in the simulation. When I compare the result with experimental values from NIST referenced to the enthalpy at 0K, the enthalpy I get is significantly higher.

Things I've thought about:

It is not an offset so it's not like a constant contribution like zero point energy is missing and besides the referencing should fix that.

It is not a constant factor difference either and I think my units are fine.

The pressure is indeed 0 and fluctuates by about 0.005GPa which is tiny i.e. pV term fluctuation is less than 1meV/atom

The simulation is stable, it remains BCC the entire team as seen from Common Neighbor Analysis and the eye test.

My questions are:

Am I thinking about this wrong? Is there some reason why this is not a valid simulation protocol for getting the enthalpy of a solid? Perhaps a classical simulation near 0K is not valid since quantum effects dominate?

Am I missing some term? It would have to be a decreasing function of temperature and any other contribution such as electronic enthalpy (from integrating electronic heat capacity) would make it worse by increasing the enthalpy

Is there a paper where someone has computed the enthalpy as a function of pressure of a solid using MD/DFT, ideally near 0K?

Wang Jinkai

Just from the nice pic, I think the data different between MD and exp. is fairly closed from my point.

A small tip is that in exp. materials often consist of defects, vacancies, dislocation, grain boundaries and so on but in MD the materials is a perfect crystal in your case.

So I think you can accept this result.

And if you're not satisfied, I would like to propose two aspects which you can do furhter tests.

1. increase the number of atoms in the system. A big system would result in better results( or not).

2. chose a "good" potential. A good potential is critical to the properties of systems. Before you do further production run, a systematical tets of exsists potentials is essential.

Şükrü Aktaş

Moleküler dinamiği kullanarak bir maddenin moleküler bazda herhangi bir fonksiyon belirtmesi için şartlar ne olursa olsun yine moleküler bazda hesaplama yapmak gerekir.

MD MOE RMSD

MD MOE RMSD

fatal error

fatal error

fatal error

fatal error

Physics: Why do we describe light by referring to its wavelength, but microwaves and radar pulses by referring to their frequency?

What kind of fluid could I use for a pitfall trap that also does not invalidate molecular testing?

Does the pressure(-600 mmHg) affects pH?

Why can't make aluminum nitride film successfully by use sputter?