Can the ATK2009.12 handle with the charged system for DeviceConfigurations?

[qingling]

For the recently released ATK2009.12,one difference is it can allow to calculate charged system. But after reading through the manual, I find the charged system is limited to MoleculeConfiguration and BulkCondiguration. Is it really that the ATK2009.12 cannot handle with the charged system for DeviceConfigurations yet?

[Anders Blom]

It doesn't really make sense to talk about charged device configurations in this context. For molecular and bulk systems, charge is conserved, and the standard is to have a neutral system. If you need a charged system, it must be specified explicitly.

For a device or two-probe system, charge is not conserved, so even if we specified an initial additional charge, chances are it would just "fly away". The system will under any circumstance self-consistently converge towards the steady-state charge distribution, which in most cases involves a certain amount of charge either being added or removed from the central region. Also, since the system is connected to infinite reservoirs of electrons, an additional charge in the total system is negligible.

That is not to say that perhaps one could try to set up some dipole structures in the middle as initial state, that is something we might consider later on...

[qingling]

Yes，your reply is right,indeed.
However, I find many charged system may have very interesting conductive properties. Compared them with the nuetral systems, it can have high on-off retio in  current. May be this can not realize in fact ? But there are many related reports.
Thank you very much!

[zh]

Could you explain the physics meaning of the charged system in your so-called reports? or give an example for this kind of system?

[Anders Blom]

Yes，your reply is right,indeed.
However, I find many charged system may have very interesting conductive properties. Compared them with the nuetral systems, it can have high on-off retio in  current. May be this can not realize in fact ? But there are many related reports.
Thank you very much!

If you can provide some references (articles), we can have a look. It might be that the systems need to be set up as charged in simpler models that do not take the charge flow explicitly into account.

[Nordland]

As Dr. Blom says, you can not add charge to a device configuration for the simple reason, that any additional charge will be simply vanish into the electron reservoirs of the electrode.
If you want to charge a device, you will have to give the device a physical reason for it having an additional electron in the device region. This can be done by introducing a metallic gate, this will allow you to modify the energy levels in the device, and hence you will be able to populate an additional electron in the device.

Otherwise around, if you want to remove a electron(s) from the device, you will need to adjust that metallic gate such that it is a energetic favorable for the device to emit a electron into the electron reservoirs of the electrode.

[Anders Blom]

... and ... fortunately, with the advent of version 2009.12, you can add such metallic gates to your DFT or semi-empirical two-probe calculations with ATK!

[qingling]

Thank you for your kindly replies!
A redox-avtive molecule is one important candidate for molecular switch. Its charge can be controled by electrochemistry gate,for example:J. Am. Chem. Soc., 2008, 130 (37), pp 12204–12205;. Am. Chem. Soc., 2008, 130 (28), pp 9013–9018.

 Now, I am confused with the difference between the ATK's gate and the  electrochemistry gate. I feel their effects maybe different.
I don't know whether the charge of  this kind molecules flows or not in real two-probe system. In my calculations, I meet many molecules,which geomtry and electronic properties can change with the charge state. I alwayes want to compare their conductive properties with the neutral ones. But I don't know for two-probe system in ATK how to distinguish them.

[zh]

The gate voltage implemented in ATK is based on a crude approximation that the electrostatic potential of molecule (or conductor) in two-probe system is rigidly shifted. In principles, the purpose of applying gate voltage  is to tune the potential difference between conductor and source (or drain). The electrochemistry gate takes the same effect on the molecule, i.e., it will tune the relative energy position of HOMO or LUMO energy levels of molecule with respect to the Fermi levels of metal electrodes.  In addition, the effect of the solvent should be taken into account in the application of electrochemistry gate. However, to treat the solvent in DFT is a challenge. The change of energy position for the HOMO and LUMO energy levels of molecule applied with electrochemistry gate comes from:  i) carrier (i.e., electrons or holes) doping due to gate voltage, ii) change of geometry structure of molecule.  When the carrier doping happens to a molecule due to the gate voltage, the molecule will be charged. In other words, the charge state of molecule is tuned by the gate voltage.

[zh]

I don't know whether the charge of  this kind molecules flows or not in real two-probe system. In my calculations, I meet many molecules,which geomtry and electronic properties can change with the charge state. I alwayes want to compare their conductive properties with the neutral ones. But I don't know for two-probe system in ATK how to distinguish them.

The right way is to compare the conductivities of molecule with and without gate voltage. In addition, the optimization of geometry structure of molecule under gate voltage seems very important, and the effect of applying gate voltage on the geometry of molecule should be taken into account. 

[Anders Blom]

The gate voltage implemented in ATK is based on a crude approximation that the electrostatic potential of molecule (or conductor) in two-probe system is rigidly shifted.

This refers to ATK 2008.10 and earlier. The gates used in ATK 2009.xx and 2010.01 are more physical, since they take into account the fully self-consistent response of the electrostatic potential distribution due to the gates, which also have a physical presence in the geometric setup.

Also, solvents effect are - at least on a basic level - included in 2009.12 and later via the specification of an ambient dielectric constant.