Dear LAMMPS documenter,

I think I found a problem in the new documentation of the TIP3P water model.

In the page “8.4.3. TIP3P water model — LAMMPS documentation” the sigma value for OH and HH was incorrect and should be 0.0.

It was correct in older documentation on “8.5.2. TIP3P water model — LAMMPS documentation”

Please let me know if I made a mistake.

Best regards

Sahar Rabet

The sigma value is irrelevant here because epsilon is zero, so there is no interaction. In general it is better to have a non-zero sigma to avoid a division by zero.

Hey!

Continuing the conversation, I found that in the table given on page 8.4.4. TIP4P water model — LAMMPS documentation , the epsilon for OO interaction of the TIP4P/Ice model is given as 0.1577 kcal/mole, whereas the epsilon in the original paper is reported as 106.1 (eps/Kb) which when converted to kcal/mole comes out to be 0.210831.

Is that just a typo or am I making some mistake?

Thanks and regards,

Chaitanya

Most likely your conversion was not done correctly. A value of over 0.2 is very large. Typical is around 0.15 to 0.18.

The values in the LAMMPS manual have been like this for a very long time. If they were so largely different and “wrong” people would have complained long ago.

Hey,

I still think my conversion is correct:

eps/Kb = 106.1 (units: K);

if Kb = 1.380649e-23 (J/K)

Thus, eps = (106.1)*(Kb)*(6.023e23) (units: J/mol)

= 882.29 (J/mol)

= 882.29/4.184 (cal/mol)

= 210.87 (cal/mol)

= 0.21087 (kcal/mol)

Also, the reason people might not be complaining is they might be using the forcefield files to generate the lammps input file using fftool or other packages, or it might only be a typo, and the value used in implicit tip4p is actually the above one.

I am trying out the documented value for my calculations anyways.

@akohlmey I must say that to my surprise, I think that there is a typo in the manual. Looking at several papers, the value of TIP4Pice seems to be 36% larger than the original TIP4P Jorgensen model (106.1 \mathrm{K} vs 78.0\ \mathrm{K}).

Using k_B=1.987*10^{-3}\ \mathrm{kcal\cdot{}mol^{-1}\cdot{}K^{-1}}, 78\ \mathrm{K} gives the usual value of 0.155\ \mathrm{kcal\cdot{}mol^{-1}}. For comparison, the value of 93.2\ \mathrm{K} of TIP4P2005 converts to a value of 0.185\ \mathrm{kcal\cdot{}mol^{-1}} and 106.1\ \mathrm{K} indeed converts to 0.211\ \mathrm{kcal\cdot{}mol^{-1}}.

Some references:

- Original TIP4P/Ice model from Abascal
- Abascal 2005 on several other models (note the very high melting temperature of TIP4P compared to all other TIP4P models)
- Conde 2017 using the 106.1 value for \epsilon
- Chialvo et al. also comparing several TIP4P models

Important to note that the previous documentation of TIP4P appears to be correct: 8.5.3. TIP4P water model — LAMMPS documentation which would explain why people didn’t complain. I suspect a typo from a confusion with the d_{OM} parameter which happens to be 0.1577 \mathrm{A} during the copy and I think that @chaitu is correct in their conversion.

I stand corrected. If the epsilon was different in the previous version, then this is definitely a mistake of the conversion of the lists into tables. All entries should be the same except for the sigma = 0.0 which have been changed to sigma = 1.0.

I will try to correct the online version as quickly as possible.

Update: online version should be corrected now.