temperature control & structure building

Hi,

I have two problems when using LAMMPS.

1 I am modeling molten NaCl under 1078K.
I control the temperature by commands of "fix npt" and then "fix nvt" with both one million steps. But the temperature cannot research the desired temperature, 1078K, rather than fluctuates dramatically, as shown in the attachment.
I want to know whether the data collected in that case are reliable or not.
And how to control the temperature to be the desired temperature within ignorable error.

2 In addition to the temperature control issue, I face another problem about the model structure building. I'd like to simulate charged molten salt systems with simple ions at high temperature, such as alkali metal halides or alkaline earth metal halides. When melting, the atoms in the systems are in random arrangement.
Firstly, I build the structure with the following commands:
"regionregbox block 0 18.96 0 18.96 0 18.96 units box
create_box 2 regbox
create_atoms 1 random 108 16 regbox units box
create_atoms 2 random 108 16 regbox units box
mass 1 23.0
mass 2 35.5
set type 1 charge 1.0
set type 2 charge -1.0 "
and the calculation came up with "nan", maybe due to substantial overlap of atoms resulting in too large energy .
Then I used "moltemplate" software to build my system, the involved files are shown in the attachment. The total charge of the system is determined by the random seed in the system*.lt file. If the total charge is not 0, also the energy of the system will be too large to be calculated. But, so far, the software has not considered about the charged systems, and its difficult to make the total charge to be right zero.
So I don't know now how to build my systems in which the atoms arranged randomly and not overlap, meanwhile, the total charge is OK.

I would highly appreciate if any advise!
Thanks & Best Regards!
Jia Wang
Postgraduate
East China University of Science and Technology
Shanghai, China
Tel: 18616024679

Temp-Steps.png

in.NaCl0502 (1.71 KB)

systemNaCl.data (5.81 KB)

NaCl.lt (1.18 KB)

systemNaCl.lt (971 Bytes)

Hi,

1). The temperature profile looks okay, as you will always have fluctuations. If you wish, you may increase the strength of tdamp (by decreasing the value). A bigger system will reduce the fluctuation as well.

2). I suggest you try with different seed values for the random keyword of create_atoms command.

3). You may charge-neutralize the structure created by any pre-processing tools by using set commands.

Cheers,
Ray

1). The temperature profile looks okay, as you will always have fluctuations. If you wish, you may increase the strength of >tdamp (by decreasing the value). A bigger system will reduce the fluctuation as well.

You’ve asked the first Q about 3-4 times now. As I said before,
you have 216 atoms in your system which is tiny. The temperature
will fluctuate (more for tiny systems). As Ray says, the temp
looks fine.

Steve

Hi,

I have two problems when using LAMMPS.

1 I am modeling molten NaCl under 1078K.
I control the temperature by commands of "fix npt" and then "fix nvt" with both one million steps. But the temperature cannot research the desired temperature, 1078K, rather than fluctuates dramatically, as shown in the attachment.
I want to know whether the data collected in that case are reliable or not.
And how to control the temperature to be the desired temperature within ignorable error.

this is not a problem of the simulation or LAMMPS, but a problem of
you incomplete understanding of statistical mechanics. the
instantaneous temperature from a simulation is precise only in the
limit of an infinitely large system.

2 In addition to the temperature control issue, I face another problem about the model structure building. I'd like to simulate charged molten salt systems with simple ions at high temperature, such as alkali metal halides or alkaline earth metal halides. When melting, the atoms in the systems are in random arrangement.
Firstly, I build the structure with the following commands:
"regionregbox block 0 18.96 0 18.96 0 18.96 units box
create_box 2 regbox
create_atoms 1 random 108 16 regbox units box
create_atoms 2 random 108 16 regbox units box
mass 1 23.0
mass 2 35.5
set type 1 charge 1.0
set type 2 charge -1.0 "
and the calculation came up with "nan", maybe due to substantial overlap of atoms resulting in too large energy .

have a look at the micelle example that describes how to use the
"soft" potential in combination with fix adapt how to resolve such
overlaps with a simulation. keep in mind that for your system the
change of potential will be very significant, so you will have to run
a minimization before you can run an MD with the final potential.

Then I used "moltemplate" software to build my system, the involved files are shown in the attachment. The total charge of the system is determined by the random seed in the system*.lt file. If the total charge is not 0, also the energy of the system will be too large to be calculated. But, so far, the software has not considered about the charged systems, and its difficult to make the total charge to be right zero.
So I don't know now how to build my systems in which the atoms arranged randomly and not overlap, meanwhile, the total charge is OK.

an alternative would be to melt a crystal at very high temperature and
use fix langevin with strong randomization.

axel.