UT theoretical chemistry code forum

we followed the instruction to calculate an adsorbed ortho diiodoebenzene, C6H4I2 on Cu...In DISPLACAR, we ask all molecular atoms to move 0.015 by x, y, z, and used the wonderful scripts to visualize these modes. I thought each mode would corresponds to the motion of a single atom in the molecules, but it turned to out each mode is a mutual motion of several atoms.

Our question is
How it is achieved?

Great! Yes, that is exactly right. If the atoms are coupled together, the normal modes will be collective motions of all of the atoms.

thank, Graeme,

I have a further question, how does dyn matrix generate the set of normal modes and generate the movies? Does it have something to do with the way to set up the DISPLACECAR?

In a naive thought of my case of C6H4I2, i pictured the modes as 3 trans +3 rotation +30 vibration; and in each vib mode, each atom displaced either along the C--H or C-I bonds, perpendicular to these bonds inside the molecular plane, or perpendicular to the molecular plane. But the modes (judged from the movie) generated by Dyn Matrix seems to be quite different from my thought, but a highly mixed one.

I understand, in my case, that any set of 30 orthogonal modes is sufficient to produce any vibration of C6H4I2; but is there a way to control how dyn matrix choose the modes?


[quote="graeme"]Great! Yes, that is exactly right. If the atoms are coupled together, the normal modes will be collective motions of all of the atoms.[/quote]

Our implementation uses Cartesian displacements of the atoms as defined in the DISPLACECAR. The Hessian is then diagonalized to find the normal modes, from which the movies are generated.

It is certainly possible to use another set of displacements to generate the Hessian. In fact, any combination of vectors which span the full space are acceptable. The normal modes, however, will not depend upon the displacements.

thanks Graeme.....

when I looked at the modes at almost 0 cm-1, in addition to the expected rotation/translation, there is some stretch of bonds...is that because of the displacement in the cartesian coordinates....anyway to do it in vasp to work in the internal coordinates?

There should be six near-zero modes for an isolated molecule. They may not be perfect translation and rotation, but they should be close. If there are any more near-zero modes, there is a problem. Again, the modes are independent of how the displacements are done (formally, and ignoring issues of noise in the forces). Vasp can not displace in internal coordinates.

i am with you now, Graeme...thanks