Re: [AMBER] EPM2 Force field for CO2

From: Liyang Zhu <>
Date: Mon, 6 Feb 2017 13:28:43 -0800

Dear David,

Thank you for your correction. It still has a problem when using the
corrected atomic radii as follows.

          angstrom kcal/mol
   c1 1.547 0.0558
   o 1.702 0.1600

The system contains 2000 CO2 was minimized around 1000 steps, then heat
quickly 0->300K->700K->333K, then run density equilibrium under NTP

The main problem comes from the low wall depth value of 0.0558 kcal/mol (*
epsilon/kB=28.129K, as reported*). The density increased at first 1ps, then
continuously deceased from 0.78 to 0.46, until there was a message came

SANDER BOMB in subroutine nonbond_list
  volume of ucell too big, too many subcells
  list grid memory needs to be reallocated, restart sander

When switch to pmemd, the density can decrease to 0.2 g/ml without error
message, until I kill the simulation.
I noticed that the ETOT, EPTOT and VDWAALS increased all the time, the
pressure kept at 100 atm (pres0=1.0 in input file). It was confusing why
ETOT increased, but there was no force to prevent the expansion of volume.

When change the wall depth of C atom to about 0.15 kcal/mol for test, the
density was equilibrated around 0.8~0.9.

So I do not understand, why so many published paper used EPM2 parameter for
CO2, but the wall depth of carbon is too low to work on NTP ensemble. Are
there some points that I not noticed?

The attached are the files of input and output of equilibrium stage.

Many thanks,

2017-02-06 8:03 GMT-08:00 David Case <>:

> On Sun, Feb 05, 2017, Liyang Zhu wrote:
> > Typically the VDW parameters in Amber is as follows:
> >
> > * angstrom kcal/mol *
> > c1 1.9525 0.1596
> > *o 1.7101 0.1463*
> >
> > And get to know that there is a EPM2 model for CO2 used in many papers,
> > which is high accurate for density properties. However, the parameters
> can
> > not be used in amber correctly, because there is always a error of sander
> > bomb during NTP equilibrium. The parameter of EPM2 is as follows:
> >
> > * angstrom kcal/mol *
> > c1 2.757 0.0558
> > *o 3.033 0.1600*
> The Amber Rmin parameters in force field files are actually Rmin/2: that
> is,
> they are atomic radii, not diameters. You don't say what you actually did,
> but that might be a part of the problem. If the EPM2 number above is
> a sigma value(?), then the Rmin value Amber needs is 3.033*(2**1/6)/2
> = 1.702, which is close to the GAFF value you cite above.
> If this is not the problem, then you would need to give more information:
> just
> saying "sander bomb" doesn't allow those on the list to offer much real
> help.
> ....dac
> _______________________________________________
> AMBER mailing list

AMBER mailing list

Received on Mon Feb 06 2017 - 13:30:03 PST
Custom Search