Why the distributions of centro-symmetry parameters are different for FCC and BCC monocrystals?

Dear steve and axel,

  Would anyone of you please take a look at the current email, which has been sent to the community for several days but not yet got any response.
  So sorry for having to send it again. Any comments or suggestions from you will be deeply appreciated.

I constructed a block-shaped monocrystal in BCC Fe and FCC Ni, respectively.Both monocrystals were constructed with PBCs and based on the crystallographic orientations of [100], [010] and [001].
When the two monocrystals were heated to various temperatures (e.g., 100, 400, 800 and 1200 K), i found significantly different features shown in the distributions of centro-symmetry parameters (CSPs) in them.

For the Ni monocrystal: irrespective of the temperature, the CSPs for atoms in the monocrystal overall showed an even distribution in the range of [0, max_CSP], where max_CSP varied with the temperature.

For the Fe monocrystal :
(1) when heated to relatively low temperatures (e.g., 100 and 200 K), CSPs overall showed an even distribution.
(2) when heated to intermediate temperatures (e.g., 400 and 500 K), CSPs overall fluctuated around two CSP values (roughly around 0 and 5, respectively).
(3) when heated to relatively high temperatures (e.g., >= 800 K), CSPs overall fluctuated around three CSP values ( roughly around 0, 5 and 10, respectively).

Please see the attached file, which illustrates CSPs distributions at various temperatures for both Ni and Fe monocrystals. Note that the above features shown in the CSPs distribution for Fe were also observed for BCC tungsten monocrystal. All simulations were run on lammps-7Aug19.

To my understanding, the features shown in the variation of CSPs distribution with temperature should be similar for FCC and BCC metals, as the CSP characterizes the local lattice disorder around an atom. Nevertheless, the actual CSP distributions for Fe and Ni showed significantly different features.

I wonder these differences are attributed to possible errors in my simulations (see the codes posted below) or the intrinsic structural properties of FCC and BCC metals (in other words, these differences are reasonable for FCC and BCC metals)?

#------------------------ Start of Codes ------------------------------#
#---------- Initialize Simulation --------------#
units metal
dimension 3
boundary p p p
atom_style atomic
atom_modify map array

#---------- Create Atoms ------------#
lattice bcc 2.8553 # for BCC Fe
region box block 0 1 0 1 0 1 units lattice
create_box 1 box
lattice bcc 2.8553 orient x 1 0 0 orient y 0 1 0 orient z 0 0 1
create_atoms 1 box
replicate 30 30 30

#---------- Define Interatomic Potential ------------#
pair_style eam/fs
pair_coeff * * Fe_2.eam.fs Fe
neighbor 1.0 bin
neigh_modify delay 5 check yes

#---------- Define Settings ------------#
compute eng all pe/atom
compute alleng all reduce sum c_eng
compute csym all centro/atom bcc

#---------- Run Minimization -----------#
reset_timestep 0
fix 1 all box/relax iso 0.0 vmax 0.001
thermo 10
thermo_style custom step pe lx ly lz press pxx pyy pzz c_alleng
min_style cg
minimize 1e-25 1e-25 5000 10000

#------------- Heating process ---------------#
reset_timestep 0
timestep 0.005

variable M equal 800
velocity all create 100 3882871

thermo 100
thermo_style custom step temp pe lx ly press pxx pyy pzz c_alleng

fix 15 all npt temp 100.0 100.0 0.5 iso 0.0 0.0 0.15
run 2000
fix 15 all npt temp 100.0 $M 0.5 iso 0.0 0.0 0.15
run 5000
fix 15 all npt temp $M $M 0.5 iso 0.0 0.0 0.15
run 2000

dump xyz_heat all custom 10 dump.monocrystal.*.800k.heat_Fe id type x y z c_csym c_eng
run 0
#-------------------------------------- END ----------------------------------------#


illustration of CSPs distribution in FCC Ni and BCC Fe.pdf (168 KB)

Sorry, I don’t know the answer to your Qs about CSP distributions.
It’s more of a conceptual Q than a LAMMPS Q
so you may have better luck in the literature than with
the mail list.

Your script is simple enough that I don’t
see any problems with it. I suggest your try an NVE
model first (not NPT, or at least do NPT to equilibrate,
but then NVE to gather CSP statistics) to see if eliminating box volume
fluctuations have any effect.


Dear Steve,

Thanks very much for your quick response and suggestion.

Before sending the email, i have tried several different ways (including the NVE model) to heat the Fe monocrystal,
but i still got the same CSP distributions.

For further validation, i also analyzed the CSP distributions for different Fe bicrystal systems at various temperatures.
Nevertheless, the variation of CSP distributions with the temperature was similar to that observed for Fe monocrystal.

So, i speculate that the difference between CSP distributions for FCC and BCC metals should be attributed to the intrinsic
difference in the structural properties of these two types of metal. I will turn to the literature for possible evidence
supporting this speculation.

Thanks again for your comments.

Best regards,
L. Yang