Enthalpy of vaporization calculation in CP2K

garold g.murd... at gmail.com
Tue Dec 9 14:07:09 UTC 2014

Hi Bharat,

We had some extensive email correspondence a few months ago when I gave you 
additional assistance, both input and output files and software changes. I 
hope that was helpful. Please let me know how it went.

As for your current problem, I think you need to review some basics which 
can be found in any molecular dynamics book. You need to ask yourself, What 
is internal energy and how do I obtain it consistently from an MD code? 
Basically, internal energy U can be approximated as E_int/n where E_int is 
the interaction energy obtained from differences of potential energies. For 
a liquid with n molecules, with a DFT (or semiempirical) simulation you may 
calculate this as:

U = [ (Average potential energy of n waters from the liquid simulation under 
some thermodynamic conditions) - n*(Potential energy of 1 water in the same 
box size, with same cutoffs, etc.) ] / n

The assumption here is of course that the molecules in the gas phase are 
quite diffuse and barely interact with one another (this is why you use 1 
molecule). You can see my earlier paper (or other similar papers) on water 
clusters for additional discussion: J. Chem. Phys. 132, 164102  (2010) , in 
particular page 164102-10 and also Figure 2 to see how E_int/n of water 
clusters start to approach the experimental liquid internal energy as the 
cluster size gets large, provided the correct Hamiltonian is used. Another 
assumption is that the vibrations and thus kinetic energies will be similar 
in the liquid and in the gas (this allows you to use the potential energies 
rather than the total energies). Of course, as also mentioned by the other 
respondents, with the PM3 Hamiltonian you will not obtain the experimental 
result but the exercise is still useful for you to see that you understand 
what you are doing.

Are you you doing this or are you doing something else?


On Monday, December 8, 2014 7:58:54 PM UTC+2, Jano... at googlemail.com wrote:
> Dear Bharat,
> I am not convinced about the usage of semiempericals like PM3 for such a 
> purpose, but it is your business... 
> Concerning the approach:
> Calculating internal energy is simply calculating the average of the total 
> energy.  But, as the average of the kinetic energy is defined on a given 
> temperature (I assume, you run NVT simulations), I usually calculate the 
> average of the potential energy. The contribution from kinetic energy 
> should cancel when you subtract the liquid phase value from the gas phase 
> value.
> Both potential energy and total energy are reported in the *.ener file in 
> cp2k.
> Otherwise do you get the right value by simply adding RT to dU? Just to 
> test your approximation:
> why don't you add d(pV), that you can calculate from experimental 
> densities.
> Janos
> On Monday, December 8, 2014 3:43:56 PM UTC+1, bharat wrote:
>> Hi Samuel,
>> It's a semiempirical calculation. I does not have any functional forms 
>> like GGA or hybrid. Calculation is correct because I was able to reproduce 
>> other properties. Isn't internal energy "Total energy" in CP2K? if not how 
>> do I calculate internal energy from cp2k results?
>> Thanks.
>> Bharat
>> On Monday, December 8, 2014 3:52:56 AM UTC-5, Samuel Andermatt wrote:
>>> You will need to post your input and output files. Do you do GGA or 
>>> hybrid calculations, how do you account for the vdW forces?
>>> On Friday, December 5, 2014 4:17:45 PM UTC+1, bharat wrote:
>>>> Hello,
>>>> This is friendly reminder. Any suggestions?
>>>> Bharat
>>>> On Wednesday, December 3, 2014 12:56:10 PM UTC-5, bharat wrote:
>>>>> Hello Experts,
>>>>> I am trying to reproduce Enthalpy of vaporization using PM3.
>>>>> Here are my calculation: 
>>>>> (I took Total energy value from CP2K output file as an Internal 
>>>>> energy, average is calculated over the MD. Experimental density is used for 
>>>>> the constant volume for both liquid and vapor calculation). Am I taking 
>>>>> correct energy value for internal energy? 
>>>>> U_vapor(Avg) = -20797.3 eV 
>>>>> U_liquid(Avg) = -20802.9 eV
>>>>> Delta_U = 5.6 eV 
>>>>> I divided with 64 (# of water molecules) and converted to kcal/mol
>>>>> = 2.02 kcal/mol
>>>>> I got the half value (reported value is 4.00 kcal/mol in G. Murdachaew 
>>>>> et al. J. Phys. Chem. A 115, 6046 (2011)). Because of this value, I got 
>>>>> only ~2.60 kcal/mol (RT= ~0.60 kcal/mol) of enthalpy of vaporization 
>>>>> (Delta_H = delta_U + RT). 
>>>>> Can anyone please tell me where I am doing wrong? Where is the factor 
>>>>> 2 missing?
>>>>> Thank you.
>>>>> Bharat 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20141209/166c8767/attachment.htm>

More information about the CP2K-user mailing list