Units are a nightmare and makes you waste a lot of time trying to transalte a forcefield in a paper to the proper one, for example in GROMACS... for example why GROMACS is usin nm, or why the constant in the repulsive part of Buckingham is defined as the inverse of anyone else is a mistery to me... So let's try to help with some relations:

  • 1 kJ/mol=0.010366 eV/particle
  • Coulomb constant:  138.9118  kJ/mol·nm/e-2 (if charges are in fundamental charge units, and distance in nm, as Gromacs wants)
  • Coulomb constant:  1389.118 kJ/mol·A/e-2 (if charges are in fundamental charge units, and distance in A)
  • Coulomb constant: 14.3996 ev·A/e2
For polarizable models α=Kc q2/k,  where Kc is the coulomb constant and k is the elastic constant of the spring. The coulomb constant is:
  • Kc=14.399629·10-3 if the elastic constant is in eV/A2 and charges are in fundamental charge units.


It is perfectly and nicely explained in the manual... but you must search it in two different chapters. I gather the info about forcefields... I keep forgeting this basics always:


Lennard Jones:

Normally I use combination rule 2 in Gromacs(σAB=(σAB)/2 ; εAB=(εA·εB)1/2  In this case:

ULJ=4ε[(σ/r)12-(σ/r)6]=ε[(r0/r)12 -(r0/r)6 ]

And the units [σ]=nm and [ε]=kj/mol being σ the distance for which the potential is zero.

An alternative way of writing the potetial is using the distance related to the minimum of the potential r0. The relation between r and σ is easy: 


Please, look in this page above if you need to transform units. Click here if you want to know the relationship with amber.


Buckingham is defined as A exp(-r/ρ)-C/r6 in most paper. WARNING in gromacs is defined as Aexp(-Br)-C/r6. That means that you must inverse ρ and (usually) change units!!!!

There is a way to obtain the parameters from buckingham to LJ and vice versa. This should not be used unless a case of extreme necessity: of course you will loose information and this should only be done in the case the repulsive side of your potential is not important (i.e. pressure can be affected!)


Here you can find an explanation to translate a force field from amber to GAFF (amber) to gromacs format, summarizing.

For non-bonded:

  • εgroGAFF·4.184
  • σgro=RGAFF·25/6  * 1/10 = 0.17818·RGAFF

For bonded:

  • Bonds: b0gro = b0GAFF / 10   ;    kb,gro = kb,GAFF /10 *2 *4.184
  • Angles: Egro = Emin/2 cth = EGAFF *2 * 4.184
  • Dihedrals: kdgro = PKGAFF * 4.184 / IDIVFGAFF


(please, if you find an error I would be more than happy to write me a warning mail to