DOS result problem?

[nus11]

Hi, The DOS I got after using the
split_dos from : http://cst-www.nrl.navy.mil/~erwin/vasptips/
Note there are some negative DOS and Integrated DOS remain at 10 when energy is above some values.
what might be the problem?
as following:
energy DOS integrated DOS
7.53260000 0.51730000 10.36000000
7.65360000 0.83100000 10.46000000
7.77560000 0.79680000 10.56000000
7.89760000 0.79480000 10.66000000
8.01960000 0.92080000 10.77000000
8.14060000 0.55750000 10.84000000
8.26260000 0.20320000 10.86000000
8.38460000 0.62670000 10.94000000
8.50660000 1.22500000 11.09000000
8.62760000 1.16700000 11.23000000
8.74960000 0.68300000 11.31000000
8.87160000 0.40250000 11.36000000
8.99360000 0.19830000 11.38000000
9.11460000 0.04932000 11.39000000
9.23660000 0.49390000 11.45000000
9.35860000 1.17800000 11.59000000
9.48060000 1.12100000 11.73000000
9.60260000 0.60360000 11.80000000
9.72360000 0.14750000 11.82000000
9.84560000 -0.05727000 11.81000000
9.96760000 0.09061000 11.83000000
10.08960000 0.59590000 11.90000000
10.21060000 0.83660000 12.00000000
10.33260000 0.38070000 12.05000000
10.45460000 0.09157000 12.06000000
10.57660000 0.83000000 12.16000000
10.69760000 2.33200000 12.44000000
10.81960000 3.68200000 12.89000000
10.94160000 4.02500000 13.38000000
11.06360000 3.31400000 13.78000000
11.18460000 2.48500000 14.09000000
11.30660000 1.95400000 14.33000000
11.42860000 1.38900000 14.49000000
11.55060000 0.98700000 14.61000000
11.67260000 0.97370000 14.73000000
11.79360000 1.18600000 14.88000000
11.91560000 1.08200000 15.01000000
12.03760000 0.72420000 15.10000000
12.15960000 0.51360000 15.16000000
12.28060000 0.72630000 15.25000000
12.40260000 0.80510000 15.35000000
12.52460000 0.83190000 15.45000000
12.64660000 1.58500000 15.64000000
12.76760000 1.66000000 15.84000000
12.88960000 0.84180000 15.95000000
13.01160000 0.35700000 15.99000000
13.13360000 0.11950000 16.00000000
13.25460000 -0.00824200 16.00000000
13.37660000 -0.01532000 16.00000000
13.49860000 -0.00309500 16.00000000
13.62060000 -0.00023630 16.00000000
13.74160000 -0.00000782 16.00000000
13.86360000 -0.00000012 16.00000000
13.98560000 -0.00000000 16.00000000
14.10760000 -0.00000000 16.00000000
14.22960000 -0.00000000 16.00000000
14.35060000 0.00000000 16.00000000
14.47260000 0.00000000 16.00000000
14.59460000 0.00000000 16.00000000
14.71660000 0.00000000 16.00000000
14.83760000 0.00000000 16.00000000
14.95960000 0.00000000 16.00000000

[admin]

1) the small negative DOS probably are due to the numerics of the BZ integration method you chose (ISMEAR). Please use the terahedron method to get accurate DOSs.
2) please have a look at the plot of your DOS. If the electronic structure has a gap, the integrated DOS remains constant in that energy range.

[nus11]

Thanks for your reply,admin! After add ISMEAR=-5 in INCAR
and also use the KPOINTS file like:

kpoints
15 ! 15 intersections
Line-mode
rec
0 0 0 ! gamma
0.333 0.333 0 ! k1

0.333 0.333 0 !k1
0.5 0.5 0 ! M

0.5 0.5 0 ! M
0.333 0.333 0 ! k1

0.333 0.333 0 !k1
0 0 0 ! gmama

However I got the error below:
The linear tetrahedron method can not be used with the KPOINTS file (generation of strings of k-points)

After referring to the manual, it said : The tetrahedron method requires an additional connection list for the tetrahedra....
ex:
Tetrahedra
1 0.183333333333333
6 1 2 3 4

How to determin the above numbers for different systems? And what i have done wrong! thanks

[angie]

You must not use generation of strings of k-points to determine a density of states. And, as you said: 'The tetrahedron method requires an additional connection list for the tetrahedra...' but only if you enter explicitly the list of k points. So, since you need a fine mesh of kpoints, as said in the manual, the easiest way is to use Automatic Generation (read 'Automatic k-mesh generation' in the manual) (a Monkhorst-pack should be fine). This way, you won't need 'extra numbers' in your KPOINTS file to use the tetrahedron method.

[admin]

please note that the terahedron method NEEDS a k-mesh which allows to construct tetrahedra. They should be of comparable volume. Therefore a k-mesh which uses only k-points along the high symmetry lines (as for a band structure plot) is completely inappropriate, as angie says.