Re: [AMBER] EPM2 Force field for CO2

From: David Case <>
Date: Wed, 8 Feb 2017 19:17:14 -0500

On Wed, Feb 08, 2017, Liyang Zhu wrote:

> Dear David,

Please send amber-related questions to the mail reflector,,
and not to me personally. That way, many people can see your question and try
to help, and the answers can help others with similar questions. See for information on how to

> The system is at 333 K and 20 atm (20 MPa), which is supercritical CO2 with
> a density of 0.8 g/ml.
> Initially I tried gaff parameter, which gave a density of 1.2, that is to
> high, so if some solutes were in the supercritical CO2, the diffusion would
> be very slow. I also tried gaff2 parameter, which gave a better density of
> around 0.95.

It's not surprising that gaff/gaff2 do not give great performance for
supercritical CO2, which is very far from the sort of molecules that were used
in its parameterization.

> I tried your suggestion, set ntb=2, ntp=1, pres0=20.0, taup=999999.0,
> irest=1, ntx=5 (if ntx=7, the initial temperature is 0 K, not previously
> heated 333 K). It shows that the pressure was at first as high as 900 atm
> at 333K, and then decreased to 140 atm. During the pressure decreasing, the
> temperature also decreased from 333 K to around 300 K, and then back to
> 333K. So eventually it equilibrated at 140 atm, 333 K, with a initial
> density of 0.5.

Why not try a initial simulation using the experimental density? If you
use a density of 0.5 under thermodynamic conditions where the denisty
"should" be 0.8, you may be seing bubbles in your system, and it could
be very hard to equilibrate. Using the experimental density will also
allow you to gain confidence (or not) that you have transferred the EMP2
potential function correctly to Amber. (I'm not at all familiar with this
potential, but linear molecules can be tough to handle in cartesian codes.
But work first to get a well equilibrated system.)

[To be clear here: I'm suggested that you the volume (and hence the
density) fixed, regulate the temperature to the value you want, make sure
that you have reached equilibrium, and see what the average pressure is.]

> I noticed that most of the articles using EMP2 are for Monte Carlo
> simulation, is it because the parameters of 12-6 LJ potential (EMP2) is not
> transferable from Monte Carlo simulation to Molecular Dynamic simulation?


....good luck...dac

AMBER mailing list
Received on Wed Feb 08 2017 - 16:30:02 PST
Custom Search