Fluctuating Charge Embedded-atom method

(FQ EAM)

January 11, 2017

The force field for Embedded-atom method (EAM) is given by [1]

X
X
1X
ij (rij )
U=
Fi
fj0 (rij ) +
2 i,j
i

(1)

j6=i

i6=j

where F is the embedding energy i.e the kinetic, exchange, and correlation
contribution to the energy; f 0 is the spherically averaged atomic density and
is the electrostatic, two body interaction.
In FQ EAM, we let charge to fluctuate from one atom to other atom so we
can study the charge dynamics in a crystal which helps us to understand the
electronic properties of crystals.
Let ith atom receives qi charge measured in the units of electronic charge.
As, ith atom has Ni valance electrons, so after fluctuation the electrons (valance)
in ith atom decreases by qi . The valance electron in ith atom is Ni qi . So the
resulting charge density for ith atom is fi0 (1 Nqii ).
qi
) must give the total charge in ith atom
Checking: The integral of fi0 (1 N
i
i.e Ni qi


Z
qi
3
0
d rfi 1
Ni
all space
Z
Z
qi
=
d3 rfi0
d3 rfi0
Ni
all space
all space
qi
= Ni
Ni
Ni
= Ni q i
The force field for FQ EAM is



X
X
X
X1
q
j
+ 1
U=
Fi
fj0 (rij ) 1
ij (rij ) +
kj qj2
N
2
2
j
i
i,j
j
j6=i

i6=j

(2)

where the modified spherically averaged atomic density is taken into account
and a harmonic potential (self energy) wrt fluctuating charge is added to the
previous EAM force field.
Previous implementation uses the definition of pair potential between two
atoms as the mean of the pair potential of same atom [1].
ij = mean[i , j ]

(3)
i

where the mean could be arithmetic or geometric and is the pair potential
defined for ith atom.
However, Johnson defined the pair potential between two different atom
as [2]
#
"
1 fj0 i fi0 j
ij
+ 0
(4)
=
2 fi0
fj
In FQ EAM, the charge density of each atom is changing so the atoms are behaving as the atoms with different valance electrons i.e atoms of different types.
So, we tried to modify the potential as defined by Johnson. But, the question
was if the scaling charge density in the Johnson Potential be the instantaneous
charge density or the equilibrium charge density. Mimicking the derivation of
Johnson we concluded that the charge distribution should be instantaneous not
the equilibrium.
So, the Johnson potential for FQ EAM is
#
" 0
q
qi
0
fj (1 Njj )
)
f
(1

1
i
N
i
i
ij =
j
(5)
q
qi + 0
)
2 fi0 (1 N
fj (1 Njj )
i
Confusion: According to Daw [1], equation(4) the pair potential is defined
Z
Z
ni (r1 )nj (r2 )
ij = d3 r1 d3 r2
(6)
r12
where ni = fi0 (r Ri ) Zi (r Ri ), which shows that charge distribution
contributes to the pair potential. In our model, the charge fluctuation changes
the charge distribution so changes the pair potential. So, I think we are missing
the change in the pair potential due to charge fluctuation.
If we look at Daws paper [1], equation(26), where he modifies the pair potential by including the electron-core interaction to the functional. The modified
pair potential also includes the contribution from the charge distribution, which
is excluded in our new model.

as

References
[1] M. S. Daw. Model of metallic cohesion: The embedded-atom method. Physical Review B, 39(11):74417452, 1989.
[2] R. A. Johnson. Alloy models with the embedded-atom method. 39(17):554
559, 1989.