GROMACS version:
GROMACS modification: Yes/No
Here post your question
I have a system containing one type of sugar, one polar compound, and water. I am using the OPLS-AA force field for the solutes and SPC/E for water. Hydrogen bonds form between the sugar, the polar compound, water, and the other solute molecules.
Simulation protocol:
-
NVT(v-rescale): 1 ns
-
NPT(v-rescale & c-rescale): 3 ns
-
Production (NPT: Nose-Hoover & Parrinello-Rahman): 20 ns
Calculated diffusion coefficients (D ×10^-5 cm²/s):
| T (K) | D |
|---|---|
| 298 | 0.7661 |
| 303 | 0.9963 |
| 308 | 0.9118 |
At 298 K and 303 K, the diffusion coefficients match experimental values quite well. However, when I increase the temperature from 303 K to 308 K, D decreases, even though I would expect it to increase with temperature. In all three simulations, temperature and pressure are fully equilibrated.
I also tried running production in NVT. In that case, the diffusion coefficients behave as expected (increase with temperature), but the number of hydrogen bonds also increases with temperature, which seems counterintuitive.
What would be the best approach to resolve this issue? How can I obtain physically reasonable diffusion coefficients without generating unrealistic hydrogen bonding trends?