Molcas Forum

tumitumi123

Member

Registered: 2017-06-13

Posts: 16

how to calculate molecule polarizability?

Dear all
When molecule interact with laser field the potential curve will change
V=V0-uF+1/2*alfa*F^2
Here u is the permanent dipole, F is electric field, alfa is the polarizability.
So I want to calculate the polarizability.
I see in FFPT, it claim that this module can calculate the polarizability. But it give me the modified energy after CASSCF.
However, it is better to know the polarizability instead of modified energy. And I found that the energy change does not behave linear with dipole perturbation.
any suggestion on polarizability is welcomed.
SHAN XUE

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: how to calculate molecule polarizability?

At the end of the FFPT description it says:

Response properties are obtained by numerical differentiation of the total energy with respect to the field parameter. For definitions of the response properties the interested reader is referred to the paper of A.D. Buckingham [60]. According to the definition of the dipole moment, it is obtained as the first derivative of the energy with respect to the field strength. Similarly, the dipole polarizability is given by the second derivative of the energy with respect to the field strength.

To do the numerical differentiation you could use a number of formulas of different accuracy. The simplest is probably (https://en.wikipedia.org/wiki/Finite_di … ifferences):

alpha ~ [ E(F=-h) - 2*E(F=0) + E(F=+h) ] / h^2

tumitumi123

Member

Registered: 2017-06-13

Posts: 16

Re: how to calculate molecule polarizability?

At the end of the FFPT description it says:

Response properties are obtained by numerical differentiation of the total energy with respect to the field parameter. For definitions of the response properties the interested reader is referred to the paper of A.D. Buckingham [60]. According to the definition of the dipole moment, it is obtained as the first derivative of the energy with respect to the field strength. Similarly, the dipole polarizability is given by the second derivative of the energy with respect to the field strength.

To do the numerical differentiation you could use a number of formulas of different accuracy. The simplest is probably (https://en.wikipedia.org/wiki/Finite_di … ifferences):

alpha ~ [ E(F=-h) - 2*E(F=0) + E(F=+h) ] / h^2

Thank you, lgnacio, you mean I should calculate this by hand. But I think if I use different field strength, it give me different answer.
I want to include a electric field, which should I choose, DIPO or EFLD?
for DIPO, if I write the input like Z  0.01, does this mean a operator zF, F=0.01a.u. is the field strength(I strongly doubt this, for Br2, energyshift(F=0.02)/energyshift(F=0.01) should be 4, but it's not).
for EFLD, I don’t understand the meaning of manual
"In addition, the origin of the perturbation operator also needs to be specified by entering a line starting with the string ORIG followed by the coordinates." 
why there is a center of operator since the electric field is a plane wave.

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: how to calculate molecule polarizability?

But I think if I use different field strength, it give me different answer.

Of course, with different strengths you will get different energies. If you mean that the result of the numerical differentiation will be different, yes, that means the response is not purely quadratic. The formula is only exact as the strength approaches 0, as per the definition of derivatives. But using very weak fields you have the problem that the energy difference will also be very small and you get large numerical errors.

for DIPO, if I write the input like Z  0.01, does this mean a operator zF, F=0.01a.u. is the field strength(I strongly doubt this, for Br2, energyshift(F=0.02)/energyshift(F=0.01) should be 4, but it's not).

That is my understanding. If the response is not approximately 4 times when you double the field, either the field is too strong, or the hyperpolarizabilities are large, or there is a bug or convergence problem somewhere.

for EFLD, I don’t understand the meaning of manual

I'm sorry, I can't help there. Maybe someone else can.