[Wien] Chemical Potential Boltztrap

[Gavin Abo]

As I mentioned previously, ecut seems to be the big window, but the 
smaller efcut window was put in front of it.

0.34517+efcut = 0.34517+(0.15) = 0.49517 => That is about the 0.50167 
that you mention in case.trace

In fermiintegrals.F90, there is:

! similar to eferm+ecut+egap/TWO in bandana
!     IF(efermi>eferm+efcut+egap/TWO) then
!       EXIT
!     end if


      IF(efermi>cbm+efcut) then
        EXIT
      end if

The slight missing difference might be the cmb (and vmb) value. You 
could probably add a Fortran write statement and print out the cbm value 
to see if that is were the difference is coming from.

If you want the full ecut window, you likely need to make efcut greater 
than or equal to ecut.  Did you try to make efcut 0.3 and see if you get 
about the 0.34517-0.3 and 0.34517+0.3 range that you calculated.

As you have pointed out, ecut (0.3) and efcut (0.15) are less then the 
Fermilevel of 0.34517.  If you want to got negative, both the ecut and 
efcut values likely need to be greater than 0.34517.  Did you try that?

On 2/28/2017 4:42 AM, elsasab at alumni.uv.es wrote:
> Dear Prof. Gavin Abo;
>
> I'm so grateful to you for answering me.
>
> But related to the first question, I almost have checked the chemical potential coloumn in the trace file for all the examples in test directory of the BoltzTrap and I didn't find zero value nor negative values like what is shown in the chemical potential x-axis in the paper of Madsen.
>
> Also as I understand from the answer of the question in the mailing list
> (http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg08117.html) If we follow their way for the case of Bi2Te3 ,for example, we see that the third line in the intrans file is:
>
> 0.34517 0.0005 0.3 78.0   # Fermilevel (Ry), energygrid, energy span around Fermilevel, #of electrons
>
> So it means that we can expect that the first and last value in the chemical potential column in the trace file should be 0.34517-0.3 and 0.34517+0.3 respectively but it is not what I found in the trace file; since the first and last value of the chemical potential are .19367 and .50167 respectively
>
> In The second part, the range that is mentioned in the definition of ecut(ecut gives the range around efermi in which the bands are included) means the range of the energy that is occupied by the valence and conduction band around the fermi level or I'm confused?
>
> Best regards
> Ana
>    
>
>
>
>> I'm NOT an expert on BoltzTraP, but I try to answer your questions below.
>>
>>> I just started to use the Boltztrap program. I have read a lot of papers using this program but I still have some questions related with the input and output files.
>>>
>>> Firstly: In most of the published articles they represent the seebeck coefficient as a function of the chemical potential. Usually in the x-axis they have zero value for the chemical potential; for example the case of Bi2Te3 in the main article. How do they get the zero value even though in the Bi2Te3.trace file there no zero value for the chemical potential. Do we need to substract a certain value from the column of the chemical potential?
>> I don't think so.  The values in case.trace should be partially
>> controlled by the "Fermilevel (Ry), deltae, ecut, number of valence
>> electrons" line in case.intrans [
>> http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg08117.html
>> ].  I suspect deltae, which should control the increment of the x-axis
>> values, was chosen such that 0 was included in the outputted chemical
>> potential values.
>>
>>> Secondly: in the case.intrance file there are two parameters that I can't understand the difference between them: ecut and efcut. As I understand that the chemical potential and the fermi level are synonym and they are equal to the fermi energy at 0 kelvin. So I think this parameters (ecut and efcut) should be synonym too.
>> In Table 1.1 of the BoltzTraP UserGuide there is:
>>
>> ecut gives the range around efermi in which the bands are included
>>
>> efcut. energy range of chemical potential around efermi that is used for
>> integrals (Eqs.(p12-p15))
>>
>> As it says, ecut is used to restrict what bands should be included in
>> the calculation and efcut is used to specify the chemical energy range
>> that should be used for the integration.
>>
>> It looks like ecut is used in src/bandana.F90 and efcut in
>> src/fermiintegrals.F90.
>>
>> You'll likely see that ecut changes the output in BANDANA block in
>> case.outputtrans, which should look something like:
>>
>> ==============  OUTPUT from BANDANA       ================
>> Fermi level changed to middle of gap. Ef = 0.05896
>> Egap:     0.122338 Energy range:    -0.141272 -     0.237481. Bands
>> range:  101 -  140
>> VBM:    -0.002205 CBM:     0.120133 Efermi:     0.058964
>>
>> ==============  End BANDANA           ====================
>>
>> On the other hand, efcut will likely change the chemical energy range
>> that you see outputted in case.trace.
>>
>> Say you have a total of 200 bands.  My understanding is ecut is like a
>> coarse adjustment to include just the 101 - 140 bands as shown in "Bands
>> range" above, while the efcut a finer adjustment to select say just the
>> 118 - 122 bands for the integration and to control the chemical energy
>> range outputted (such as for plotting [
>> http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg10961.html
>> , https://github.com/gsabo/BoltzTraP-Patches/tree/master/1.2.5 ]).
>>
>>> Also in the case.intrance file at the we should insert the value of the fermi level. How can we get this value? Do we have to put the value of the Fermi energy that we have in case.scf file?
>> Yes, you should be able to grep for it [
>> http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg11788.html
>> ,
>> http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/msg14185.html
>> ]; for example:
>>
>> grep :FER case.scf*