about ismear

Message

zhuanghl · #1 Post by **zhuanghl** » Tue Jan 09, 2007 9:17 am

Dear everyone,

As said in the guide book ismear should be set to Zero for semiconductor or insulator system.But for metallic system it is better to be set as 1 or 2.Well,if I don't know whether a system is metallic or semiconductor,or some dopants may change metallic system to semiconductor,how should I set the value of Ismear at first.

Many thanks!

#2 Post by **admin** » Wed Jan 10, 2007 4:12 pm

do a single electronic scf run without ionic relaxation (ISMEAR=-5) first and have a look at the electronic structure of that run.

zhuanghl · #3 Post by **zhuanghl** » Fri Jan 12, 2007 6:14 am

when the system is so large
can we still use ISMEAR=-5?

#4 Post by **admin** » Fri Jan 12, 2007 2:53 pm

you have to take enough k-points to cover the BZ with tetrahedra, than you can use ISMEAR=-5 of course.

simly · #5 Post by **simly** » Tue Mar 27, 2007 5:54 am

For accurate bulk relaxation (for insulator and semiconductor), is it appropriate to use ISMEAR=-5 for relaxation in the first step if there are enough k points to cover the BZ with tetrahedra?

It says in the manual to use either 0 or 1.

#6 Post by **admin** » Tue Mar 27, 2007 12:35 pm

relaxations with the tetrahedron (Bloechl) method should only be avoided if the system is metallic , or if the cell is so large that the number of k-points is not sufficiently high to construct the tetrahedra. So, if you have enough k-points, use ISMEAR=-5