I have been running Materials Studio (MS) to investigate methane transport in kerogen structure.
-----------------------------------------------------------------------------------------------------------------------------------------------------
The running procedures:
- Import the kerogen molecule .xyz file
- Calculate bonds (function in MS)
- Geometry optimization (Forcite)
- Amorphous cell construction: 6/8/9/12/20/30 kerogen units with a target density of 0.1/0.6/0.7/0.8 g/cm3
- Geometry optimization (Forcite)
- NPT: 5MPa, 338K, 1ns, then analyze kerogen density, porosity and pore size distribution, and compare with the published experimental/simulation data.
- Methane adsorption simulation (at a given Pressure and Temperature) (Sorption Fixed pressure)
- Based on the snapshots from Step 7, run MD simulations (first NVT for 500ps and then NVE for 2ns)
- Forcite Analysis of methane molecules to calculate the self-diffusion coefficient
-----------------------------------------------------------------------------------------------------------------------------------------------------
Issues/observations I am having now:
- After Step 6, my pores are smaller than the published data: 0.5-4 Angstrom vs. 1-12 Angstrom.
- From Step 8, I can see that a lot of my methane molecules are “stuck”, which is my biggest concern now.
- From Step 9, MSD vs. Time curves are not what I have expected, please see the attached for several examples.
-----------------------------------------------------------------------------------------------------------------------------------------------------
Some potential reasons in my mind and Something I want to try:
- I have the kerogen .xyz file downloaded from online. My question is how much could I trust on the “Bonds Calculation” function in MS?
- For Step 6, I have tried different procedures to construct the kerogen molecular structure: (1) Multiple NPTs (a. NVT, 800K; b. NPT, 20MPa, 800,700,500,300K; c. NPT, 300 K and1 bar (=0.1MPa); (2) Simulated Annealing; (3) One NPT. Which one you would recommend?
- Now I am using COMPASS force field, in some published papers, they are using CVFF force field. So I am wondering how much effect the force field type might have in my simulations?
- For Step 9, since we found that some of the methane molecules are stuck there, we plan to delete these “stuck methane molecules” and only consider the actually moving methane molecules in our MSD calculation. So I am wondering whether this is a proper direction to go?
- For the self-diffusion calculation, some papers said that NVE run should be used instead of NVT. But some papers are actually using NVT run to calculate self-diffusion coefficient. What are your comments?
- Some published papers are using LAMMPS, so how much influence using different software might have on the simulation results?
- What are the potential reasons for the differences between my simulation results and the published data?
-----------------------------------------------------------------------------------------------------------------------------------------------------
Any comments would be greatly appreciated!!
