Question about the bond length of metel single atom line.

[fangyongxinxi]

Dr Sir,
 
        I use ATK to find the relationship between Energy and bond length of single atom line, the script as below said. As for the Au single atom line, we get the bond lenght is about 2.589 A, but the tutorial said the bond length is 2.886A. So, I confused. Could you give me some suggestion ? Thank you.

[zh]

You should take care of the the chosen calculation setup such as the basis set, cutoff, exchange-correlation functional, the size of k-point mesh, and etc.

[fangyongxinxi]

Thank you.
I did the test, found that the Energy is not very sensitive to the parameters you said, so I still confused.
If it is convenient, could you do one or two test ? Thank you so much.

[fangyongxinxi]

By the way, I did the test about Al. and found that the lattice constant is 2.39A, the same as the example of the ATK hand book said, so, I don't think the calculate parameters is wrong.

What's more, the I_V characteristics is sensitive to lattice constant, so I doubt that I can trust the lattce constant got by ATK or not.

[Nordland]

I have a couple of questions:

• You write Au ( Gold ) in your post, but the script is using Copper. Which one is correct?
• Since this is a wire, one should never have any kpoints in the X,Y plane, but the scripts uses 3,3,100
• Gold is very sensitive to mesh cutoff, so I would guess that this would be the starting point.
  

[fangyongxinxi]

thanks for Nordland's reply,
I know what is wrong with my question,

but first, as for Nordland said, the script I posted is cu-line, when you test different element, you should chang the script a little. For Au lattice constant 2.59A, it's no problem, because I use Au-line, when I doing this. As for KPoint, 3 3 100 is ok, just take a little more time, and the mesh cutoff 150Ry and 100Ry, a little differences for the result.

Au, 2.886A is the nearst-neighbor atom distance in the the [110] direction of bulk Gold and it appears in some papers, as for PRB 69,125409(2004), PRB 67,193104 (2003) did. The authors choose 2.886 as the distance of the Au atom line in TwoProbe configuration. Au, 2.589, as you get by the script, it seems like that people don't like to use optmized bond length of Au atom line to build structure.

Question is which one I should take, 2.88 or 2.59 ?
Looking forward to your reply.

[Nordland]

Hey again.

On the matter of k-points you are correct, that it does not lead to incorrect results, but it just takes longer - so therefore I have changed it to 1,1,100 for my test runs.
I have increased the mesh-cutoff to 200 Ry, and I also get a lattice constant of 2.59 Ang for the wire.

Regarding which one to take, it becomes a matter of taste. I would personally always use the "correct" lattice constant of 2.59 Ang for the Au-wire, unless I wanted to
reproduce the exact numbers of another paper where they have used an "incorrect" lattice constant.

I think that the best thing to do, would be to make a transmission spectrum for the 2.59 Ang and 2.886 Ang lattice constants and compare them. If they agree somewhat, I would continue using 2.59 Ang, however if they disagree I would continue using 2.59 Ang, but i would make a point or remark out of this difference.

[fangyongxinxi]

Thank Nordland again.