[CP2K-user] [CP2K:20820] Re: Gallium + CO2 lack of convergence

[Marcella Iannuzzi]

Dear Michela, 

You mentioned a liquid, but the xyz you posted shows a sort of lattice 
forming a cubic cage containing the CO2. 
For sure the system sie is very small and I doubt it represents a reliable 
model for liquid Ga with CO2.  Ga crystalline in a base centred 
orthorhombic structure, maybe you can start to verify whether you can 
reproduce that.
>From the output, one observes that the pressure is rather negative 
(compressive), indeed the volume very rapidly shrinks and the temperature 
of the barostat becomes crazy.  
To generate a reasonable liquid-model  rather long sampling is required and 
some structural parameters (e.g., correlation functions) and electronic 
parameters should be used to verify the reliability of the model. 
Once the model of the liquid is OK, one can think of inserting a solute and 
re-equilibrate. 

Regards
Marcella

On Thursday, October 24, 2024 at 3:47:21 PM UTC+2 bnzmi... at gmail.com wrote:

> Hi Marcella and everyone,
>
> I wanted to follow up on your advice. I used an equilibrated structure of 
> 64 Ga atoms, eliminated ~8 central Ga atoms and added the CO2 molecule. The 
> MD steps converged until step #14, then the SCF loop did not converge. I 
> wanted to update anyone who may be looking at this thread for guidance: I 
> will be sizing up my cell dimensions.
>
> Thank you,
>
> Michela
>
> On Thursday, October 3, 2024 at 1:22:48 PM UTC-4 Marcella Iannuzzi wrote:
>
>> Hi
>> If you have a good equilibrated Ga liquid box (right density and low 
>> stress tensor) I wouldn't run GEO_OPT
>> Obviously introducing CO2 is going to change the conditions, but I would 
>> anyway start with a NVT run for a first equilibration and then run NPT to 
>> re-equilibrate the volume. I would add the CO2 molecule and remove all the 
>> Ga atoms within a certain radius from the center of mass of the molecule, 
>> and then run the equilibrations as described above. 
>> The only way to lower the concentration is to increase the amount of Ga, 
>> i.e., increase the box. 
>> Regards
>> Marcella 
>>
>> On Thursday, October 3, 2024 at 5:02:46 PM UTC+2 bnzmi... at gmail.com 
>> wrote:
>>
>>> Hi Marcella,
>>>
>>> Thank you again!
>>>
>>> 1) should I run GEO_OPT on the pure Ga liquid, then add CO2?
>>> 2) How should I form a cavity artificially without disrupting the newly 
>>> equilibrated Ga structure?
>>> 3) I only have one molecule of CO2 in there - How should I go about 
>>> lowering the concentration? Should I just increase my #Ga atoms and add 1 
>>> CO2 molecule?
>>>
>>> Best,
>>>
>>> Michela
>>> On Thursday, October 3, 2024 at 10:20:25 AM UTC-4 Marcella Iannuzzi 
>>> wrote:
>>>
>>>>
>>>> Dear Michela, 
>>>>
>>>> The procedure you describe  does not sound very appropriate to me. 
>>>> You should first obtain a liquid system, without solute. 
>>>> I suppose you should check for density and other properties and have a 
>>>> sufficiently large box. 
>>>> Then you can create a cavity in the equilibrated liquid and insert the 
>>>> solute, still with the right C-O bond length.
>>>> If the SCF does not converge anymore after a few steps it is probably 
>>>> because of the coordinates. 
>>>> The concentration of CO2 seems rather high. 
>>>>
>>>> You can use GAPW. It is more commonly used for all electron 
>>>> calculations. With PP, GPW is as accurate.
>>>>
>>>> Regards
>>>> Marcella
>>>>
>>>>
>>>>
>>>> On Thursday, October 3, 2024 at 3:02:10 PM UTC+2 bnzmi... at gmail.com 
>>>> wrote:
>>>>
>>>>> Hi Marcella,
>>>>>
>>>>> thank you for your kind response and your time.
>>>>>
>>>>> 1) I just switched to double zeta quality a few hours ago, but my MD 
>>>>> just crashed because, weirdly, it converged for the first few SCF loops, 
>>>>> but then it stopped converging (attached the output file here to explain).
>>>>>
>>>>> 2) I am using GAPW because I found that augmented plane waves method 
>>>>> worked really well with my liquid Al systems before. that method is also 
>>>>> reported in DFT literature for liquid Ga. Rationalizing it, I think it 
>>>>> works because it samples regions of space with different charge densities 
>>>>> with more accuracy. Do you think I should consider something else?
>>>>>
>>>>> 3) I used a cell size to represent the density of liquid Ga with the # 
>>>>> of atoms I have. I prepared my coordinates with a Python script, then 
>>>>> relaxed the geometry in Avogadro2 software and inserted CO2 such that it 
>>>>> was a distance of min 2.5 A to minimize initial repulsion with Ga atoms. Do 
>>>>> you have any suggestions to prepare a structure? I am leaving 1-2 A on all 
>>>>> sides from the unit cell boundaries because I have been worried about Ga 
>>>>> atoms being too close to neighbors across periodic boundaries.
>>>>>
>>>>> Michela
>>>>> On Thursday, October 3, 2024 at 8:23:26 AM UTC-4 Marcella Iannuzzi 
>>>>> wrote:
>>>>>
>>>>>> Dear Michela, 
>>>>>>
>>>>>> The basis set you are using is of poor quality. 
>>>>>> The coordinates you sent show a rather strange C-O bond length
>>>>>> The cell is very small, but still there is vacuum space among the 
>>>>>> replicas in all directions, it is a rather weird choice of coordinates. 
>>>>>>
>>>>>> Is there a reason why you are using GAPW?
>>>>>>
>>>>>> Regards
>>>>>> Marcella
>>>>>> On Wednesday, October 2, 2024 at 7:27:59 PM UTC+2 bnzmi... at gmail.com 
>>>>>> wrote:
>>>>>>
>>>>>>> Good morning dear CP2K community,
>>>>>>>
>>>>>>> how are you? You may know me from previous posts on liquid Al (+CO2) 
>>>>>>> MD troubleshooting. All of your responses have been super helpful so far, 
>>>>>>> and I am coming here again for a different liquid metal.
>>>>>>>
>>>>>>> My simulations with pure liquid Gallium have been less troublesome 
>>>>>>> than all of my liquid Al simulations, but the MD with 1 CO2 molecule won't 
>>>>>>> converge. Can I please get some help troubleshooting?
>>>>>>>
>>>>>>> Thank you,
>>>>>>>
>>>>>>> Michela
>>>>>>>
>>>>>>

-- 
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/1f787ddc-2f7a-4294-bd34-92ae21762ae8n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20241026/39b1e5d5/attachment.htm>