help！！The problem about calculation of TMR in ATK

[huangshenjie]

HELLO!!

We all know that TMR=(Gp-Gap)/Gap=(Ip-Iap)/Iap. I calculated the TMR of our model refering to the tutorial:spin transport in Fe-MgO-Fe.

I also calculate the TMR through the currents, through the equation, TMR=(Ip-Iap)/Iap (the currents are also calculated by ATK), but the results of this two calculations are extremely different!!

For example, in 1.5V, the TMR through the method  TMR=(Gp-Gap)/Gap is positive, however, the TMR through (Ip-Iap)/Iap is negtive!

This result really confuses me!! In my opinion, these two results should be corresponding. 
Waiting for you answer.......

[huangshenjie]

I don't know why the current and conductance calculated by ATK don't adapt to the equation G=I/U.

[Anders Blom]

Already asked and answered: http://quantumwise.com/forum/index.php?topic=1402.msg7673#msg7673

[mads.engelund]

Well 'G=I/V' is just an approximative rule-of-thumb for simple systems. And I can tell you, if it always held, then our work would be a lot easier.

In reality conductance can be a complicated function of voltage, especially on the atomic scale. As an example, engineers are often searching for systems with negative differential conductance, meaning systems where the conductance goes down when the bias increases.
Reality is more complicated than 'G=I/V'.

[huangshenjie]

Oh，thanks for your reply! In your opinion, the equation TMR=(Ip-Iap)/Iap is not precise enough,  but I have saw many papers still cit this equation. 

[Anders Blom]

It's primarily a matter of clarity and definitions. As long as you say you use the optimistic definition (and explain why you choose to do so), nobody can fault you for it.

[huangshenjie]

Thanks for your reply. But I didn't get your point. In my view, the optimistic and pessimistic definition of TMR just means TMR=(Gp-Gap)/Gap or (Gp-Gap)/(Gap+Gp). I really don't the physical meaning of these two definitions.

If I choose the optimistic definition, say TMR=(Gp-Gap)/Gap, do you mean that now the ohm's law is effective?

waiting for your reply si r

[Anders Blom]

You typically want to use the pessimistic version, since Iap is often much smaller than Ip and so the optimistic TMR can tend to infinity which doesn't make that much sense. The use of either definition is not related to the linearity of the I-V curve, these are two separate issues.

[huangshenjie]

Thank you sir and I got your point!

But I still have another question, in our model, Gap is bigger than Gp in zero bias, as a result, the TMR in zero bias is negative. Is this possible?? Or there is something wrong with our calculation??
 

[Anders Blom]

Of course it's possible. I suggest you do a literature search on the topic.