Magnetic property of Molecules

[firemarsh]

Say I have a molecule with two metal centers and some bridging ligands binding the two metal centers, how do I know whether the molecule is ferromagnetic or antiferromagnetic (or neither)? What exactly dictate whether this molecule would be ferromagnetic or antiferromagnetic?

 

[shawl]

In a molecule, it is not easy to understand the magnetism.

The magnetism is related not only on the molecule structure, but also on the details of the chemical bond.

You'd better give a detail materials (molecule) if you really have a problem in your research,

 

[shawl]

Of course, if you have apparatus, you may measure it.

The measurement method is including: M-H, ac susceptibility, etc.

 

[DrDu]

A single molecule can never be said to be ferro- or anti-ferromagnetic as this is a property of macroscopic bodies, not of molecules.
You can only say that the spins in the molecule are either coupled to a non-zero total spin or to zero total spin.

 

[shawl]

It may be hard to have a discussion with the macroscopic magnetism in a molecule, but it is allowed to have a discussion with the spin alignment of each magnetic ion in a molecule.

There are many possible states, such as ferro-alignment, antiferro-alignment, ferri-alignment.

If the spin alignment is structure dependence. The topic of spin frustration arises, which will bring much more novel phenomenon in study of a molecule.

 

[cgk]

The computational method of answering this question would be to simply calculate the energies of the various spin states (from closed-shell/low spin to high spin) on the relaxed geometries of the relevant spin states, using some DFT or wave function methods. The energetically most favorable one would then be your guess for the preferred spin coupling (at zero temperature, in the gas phase).
Of course, in practice these calculations may be anywhere from trivial to presently infeasible.

 

[firemarsh]

Yes I think what I want to ask is ferromagnetic interaction between the metal atoms.
The only thing I have is the bond angles and lengths, and the alignment of d orbitals of the metal centers (i.e. staggered/eclipse), is it possible to tell the ferromagnetic/antiferromagnetic interaction form these information without in-depth calculation?

 

[cgk]

Yes I think what I want to ask is ferromagnetic interaction between the metal atoms.
The only thing I have is the bond angles and lengths, and the alignment of d orbitals of the metal centers (i.e. staggered/eclipse), is it possible to tell the ferromagnetic/antiferromagnetic interaction form these information without in-depth calculation?

No, it is not. There is no actual ferromagnetic or anti-ferromagnetic interaction; these are just effective pictures for what happens when the whole system figures out its covalent bonding situation. This picture is encoded in the exchange coupling constants of the effective Hamiltonian; but in order to obtain these constants (which decide about ferromagnetic or anti-ferromagnetic coupling) one needs to perform calculations on the total electronic structure. There was a recent nifty review on DFT methods by Frank Neese, which also goes into these issues:
http://dx.doi.org/10.1016/j.ccr.2008.05.014
Maybe that would clarify things for you.

 

[alxm]

Maybe I should weigh in on this, since I've done a number of these calcs (yay, I'm even cited in Neese's review).

As cgk says, the only way to get the value theoretically is by explicit calculation, I'm afraid, and DFT would be the most likely candidate. While DFT calcs on, say, simple organic molecules and reactions can be done today without a lot of in-depth knowledge, calculating open-shell metalloorganic systems take a bit of knowing what you're doing.

To begin with there's the simple question of what functional to use. As cgk mentions, the relative energies of your spin states depend on the exchange energy, and so they depend almost entirely on which functional you choose. Especially with hybrid functionals where you might choose different values for the HF-exchange component, and the improvement on one aspect doesn't necessarily translate to other ones. There's no consensus on what's best. (According to Per Siegbahn, B3LYP* with 15% HF-exchange gives better energies for transition-metal atoms. According to Martin Kaupp, BHLYP, with 50% HF-exchange gives better spin-densities. But it also increases the risk of spin contamination, which is always a problem.) On top of all that, there's a purely practical issue in that most programs have convergence problems with that kind of systems, and usually require explicit starting guesses, vshifting etc to 'find' the correct wave function. It can be determined experimentally though, using EPR spectroscopy.

 

[DrDu]

DFT is fine maybe for explicit calculations, but nevertheless for many substances it is possible on a qualitative basis to derive the sign of the effective coupling without using a computer. There are many models for direct or indirect exchange or itinerant magnetism.

In many molecules with more or less well defined bonds maybe the best starting point from a qualitative point of view is valence bond theory as it directly yields an effective hamiltonian for the spin system. If one wants to, ab initio Valence bond programs give solutions which are, at least for small molecules where they are feasible, potentially better results than semi-empirical methods like DFT.

 

[alxm]

In many molecules with more or less well defined bonds maybe the best starting point from a qualitative point of view is valence bond theory as it directly yields an effective hamiltonian for the spin system. If one wants to, ab initio Valence bond programs give solutions which are, at least for small molecules where they are feasible, potentially better results than semi-empirical methods like DFT.

I've never heard that claim made before. Can you back it up?

VBT methods are not particularily accurate in general, the spin couplings are weak, and in a wavefunction method you'd typically need a relatively large number of determinants to accurately calculate it. I don't see how VBT would be useful here. A cursory search shows people doing the http://jcp.aip.org/resource/1/jcpsa6/v116/i10/p3985_s1" ; using DFT to calculate valence effective hamiltonians.

I wouldn't consider DFT methods aren't 'semi-empricial' in comparison to VBT methods.