There is a tutorial at the Opium pseudopotential generation project, see
link below.  Although you might not want to take the time to fiddle with the
software, the tutorial does a good job explaining the different
modifications that can be made to norm-conserving pseudopotentials.

Scott Beckman 

Assistant Professor
Department of Material Science and Engineering
Iowa State University

On 7/13/09 2:57 PM, "Hui Tang" <> wrote:

> When we build a pseudopotential (psp), we froze the inner core states and
> treat only the valence electrons due to the fact that the properties of
> materials are largely determined by the outer-layer valence electrons. In
> order to obtain a psp with good transferability, there should be very small
> overlap between core and valence electron distributions which is true for most
> light elements. For most the elements, the overlap between core and valence
> electron can be so large that the psp built in the usual way cannot lead to
> the correct properties of materials composed of that element (e.g., Na, K).
> There are two methods to solve the above problem. 1) Nonlinear core
> correction, we use an artificial core charge distribution to replace the real
> core charge distribution. When building psp, we substract the contribution to
> Vxc and VHartree, which comes from the artificial core, and when using the
> psp, add the corresponding ones (see PRB 26, 1378). When doing nonlinear core
> correction, we don't increase the number of valence electrons but only
> increase the cutoff energy for plane-wave calculations due to the added core.
> 2) Semicore states, which means including some states that you usually treat
> as core states when you build a pseudopotential. For example, when building a
> psp for Ca, we not only include 4s, 4p, 3d but also include 3s, 3p in the
> valence states. For this method, we include a lot more electrons (e.g. 8 more
> e/atom for Ca), and also need a much large cutoff energy because we have
> included much more localized 3s, 3p electrons. So including semicore states
> will result in much harder calculations.
> In fact, for most cases, nonlinear core correction is already enough for a
> good psp and including semicore states is kind of overkill. So my suggestion
> is: use the nonlinear core correction when you need it, use semicore states
> unless you have to.
> Hui
> ----- Original Message -----
>> From:  daijianhong001 <>
>> To:
>> Sent: Monday, July 13, 2009 2:42 AM
>> Subject: [SIESTA-L] semicore state  meaning
>> Dear all,
>> Some pseudopotential includes semicore state.I wonder what is the meaning  of
>> semicore state and which element includes semicore state?
>> Best wishes!
>> Dai
>>  200万种商品,最低价格,疯狂诱惑你
>> <>

Reply via email to