somewhere in time

There have two email about I-V curve of molecular device in experiments and theoretical works.

Dear Siesta users,

I’ve followed tutorials for TranSiesta (which was done TS school 2009)
For one of the exercise, Gold_111, I have some general questions.

I calculated I-V characteristic for a given strcture and the electrical resistivity is estimated to be ~ 58 *10^(-7)ohm m, which is much larger than experimental value of 22*10^(-9)ohm m(at T=20C).

Is the difference from only considering ballistic transport and not including the diffusive transport part? (if yes, then calculated number should be smaller then experimental value. doesn’t it?)

Or temperature factor?

Can you explain clearly about my questions.?

I appreciate any help.

Thanks.
Kyoung

Dear Kyoung:

When it comes to I-V characteristics of molecular size devices, comparisons to experimental results is not straight forward.

Even if there is only ballistic transport in the experimental device (which is so for metallic point

contacts and low biases to a 99% accuracy), there are many unknowns in the experimental devices, which imply that a direct comparison might not be completely meaningful.

Some of these difficult (almost impossible) to control characteristics of the experimental objects are the following:

1- The specific (down to the atomic detailed) geometry of the nano-thing connected to the electrodes.

2- The specific atomic detail of the contact-regions (regions between the central object and the electrodes).

3- The influence of (uncontrolled) tensile (or compressive) stress in the contact region.

4- The influence of (thermal or mechanical) vibrations of the central region.

5- The nano-scale structure of the electrodes near the contact, local crystaline orientation.

6- The presence of uncontrolled, unwanted, chemical species at the contact (typically originated from the residual gas present event under high vaccuum conditions)…

7,8,9… – Influence of spurious electrical fields , … magnetic fields …, local heating , …

…

…

All the former may affect notably your computed I-V results.

So this amounts to say that:

–> A simple comparison of a set of calculations with a set of experiments sometimes is usually not enough to conclude that the computation is correct or not.

Rather, when facing a big discrepancy, one should wonder about the following:

–> What is the difference between the computed (model) system and the real system?

- In some cases the answer to this question may be: “The real system does not present the molecule attached in such or such way (or such site) to the electrodes.”

- In other cases, the answer may be: “There are likely some residual gas atoms glued to the nano-object that we did not take into account.”

…

…

…

- Yet, in some other cases the answer may be: “Ballistic transport only accounts for a fraction of the current here, most of it is in the diffusive regime.”

But the latter is usually not the answer found, given the atomic scale sizes of the system usually studied. (The diffusive regime does not usually appear for very small devices)

Sorry about the long explanation, the summary is that comparison to experiments in the case of I-V characteristics of atomic scale things is usually much trickier than other typical comparisons in DFT (e.g., stable structures of molecules)

Yours

Jose A.

事实上，安装Inelastica并不难

本来习惯于使用ubuntu，安装软件只需要sudo apt-get install 就差不多了

但是最近新实验室中操作系统换成了CentOS，虽然也有yum install，但是其库显然没有ubuntu那么丰富，比如python，其版本还是2.4。安装Inelastica的建议是安装python2.6，只好通过源代码编译。但是蹊跷的是numpy始终不能安装成功，本人基于lapack, atlas等库函数一直没有成功。迫不得已，最后安装intel_MKL，最终终于编译成功了。

看到可爱的Inelastica顺利运行，是为记。

IETS is a powerfull tools in design molecular device both in experiments and theoretical works. In the latest PRL, there is one paper about IETS probing efficient spin transport in magnetic materials. As mentioned above, the IETS method we used in our lab is one of the first calculations all over the wide. I don’t know the advance of this Lett. till now and I will read it as soon as possiable.