what is background charge // RE: [Wien] how to calculate the ion-doped semiconductor

[Jorissen Kevin]

By changing Z, you effectively change the element.  It's not Nd anymore.
In some cases people use tricks like this : eg. in ELNES sometimes people use a "Z+1"-atom to mimic the EELS core excitation (the explanation is that due to the excitation, the core is deshielded, and hence about "one electron more attractive" to electrons - which is then said to amount to the same thing as having a heavier nucleus - adding a proton = removing an electron).
However, by lowering the atomic number Z in case.struct, you have turned Nd into La, which is a *lighter* core with a weaker potential.  Z defines the core charge!
Let's see what you've been doing...
If you refreshed case.inst (structgen or instgen) after lowering Z, then after removing three electrons from case.inst you actually have a La+3 starting density, and case.in2 has been prepared by lstart to keep the +3 in all iterations.  In this case, in each iteration, mixer will find the total electronic charge of the unit cell to be 3 e smaller than the positive charge of the nuclei, and it will renormalize because you did not specify a -3 charge in case.inm (check the different charges written by mixer in case.scfm before and after renormalization).  Such a calculation may converge (I've done them by accident), but be careful, it doesn't mean the results are useful!
If you did not refresh case.inst (i.e. it's still from the Nd epoch), then after removing three valence electrons, you essentially end up with a La atom.  The 0.0 background charge in mixer is now okay.  The calculation is expected to run fine, you have calculated a system doped with neutral La.  However, if you want to mimic a Nd+3, I'd expect to use a Z+3 rather than a Z-3.
 
If I understand you correctly, I'd recommend to start again in a fresh directory, leave Z=60 in the struct-file (check case.inst), choose whether or not to remove three valence electrons from case.inst (this is only the *starting point* - but it may determine where you end up, if this is a 'difficult' calculation), then go into case.in2 and check that the number of valence electrons is what you'd expect (3 smaller than for a neutral system - depending on what you do in case.inst, you may have to edit this number manually), then in case.inm choose a background charge of -3.0, which will keep mixer from renormalizing your electron charge density to match the nuclear charge.  Then do run_lapw and whatever analysis you feel like.
 
People often get confused by the "background charge" in mixer.  Background charge is just a trick.  It is *not* added to the density of your system.  It does not contribute to the potential.  It's not there at all ...  except at the very end of each iteration, after mixer has added the charge of valence and core electrons (from lapw2 and lcore, respectively).  It then checks whether this charge is equal (except for sign, of course) to the sum of the charges of all nuclei.  TO THIS ELECTRON CHARGE SUM IT ADDS THE BACKGROUND CHARGE.  Then, if it finds the two numbers (electrons vs. nuclei) to be different, it will multiply the electron charge density at every point with a constant factor which makes them equal again.  If you want the system to remain charged, you sort of fool mixer by telling him there's a uniform background he should add to his sums, and this will make mixer happy and keep it from messing with your charged system.
This is the *only* place in wien where this background charge is considered explicitly (which is why the number is in case.inm, of course, and not in other input files as well) - though, obviously, implicitly it influences everything :-).
 
 
Hope this helps,
 
Kevin.
 
 
 
 

	-----Oorspronkelijk bericht----- 
	Van: Yushan Wang [mailto:wys at UDel.Edu] 
	Verzonden: vr 7/9/2004 7:21 
	Aan: wien at zeus.theochem.tuwien.ac.at 
	CC: 
	Onderwerp: RE: [Wien] how to calculate the ion-doped semiconductor
	
	

	thanks, I did as follows, and it seems that the SCF is running smoothly:
	for Nd (+3),
	
	I set the Z in the .struct is 57, not 60, and then set the number of
	electron in .inst  in one 4f orbital from 3 to 0.
	however,
	here, I still set the background charge as 0, is is ok?
	
	yushan
	
	_______________________________________________
	Wien mailing list
	Wien at zeus.theochem.tuwien.ac.at
	http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
	
	

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/ms-tnef
Size: 7130 bytes
Desc: not available
Url : http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20040709/ebf48986/attachment.bin