VASP 计算能带和 DOS，高能态计算结果与文献不符

北大-陶豫

如题，我最近学习使用 VASP，发现我算出的能带和 DOS 常常与文献不符。例如我计算立方钙钛矿 CsCaBr3，相关输入文件如下：POSCAR：

1. CsCaBr3 https://materialsproject.org/materials/mp-30056/ Perovskite CN: Cs=12, Ca=8
   
2. 1.0
   
3. 5.772608 0.000000 0.000000
   
4. 0.000000 5.772608 0.000000
   
5. 0.000000 0.000000 5.772608
   
6. Cs Ca Br
   
7. 1 1 3
   
8. direct
   
9. 0.000000 0.000000 0.000000 Cs
   
10. 0.500000 0.500000 0.500000 Ca
    
11. 0.500000 0.000000 0.500000 Br
    
12. 0.000000 0.500000 0.500000 Br
    
13. 0.500000 0.500000 0.000000 Br
    

复制代码

INCAR_DOS：

1. Global Parameters
   
2. ISTART =  1            (Read existing wavefunction, if there)
   
3. ISPIN  =  1            (Non-Spin polarised DFT)
   
4. # ICHARG =  11         (Non-self-consistent: GGA/LDA band structures)
   
5. LREAL  = .FALSE.       (Projection operators: automatic)
   
6. ENCUT  = 1200          (Cut-off energy for plane wave basis set, in eV)
   
7. # PREC   =  Accurate   (Precision level: Normal or Accurate, set Accurate when perform structure lattice relaxation calculation)
   
8. LWAVE  = .TRUE.        (Write WAVECAR or not)
   
9. LCHARG = .TRUE.        (Write CHGCAR or not)
   
10. ADDGRID= .TRUE.        (Increase grid, helps GGA convergence)
    
11. # LVTOT  = .TRUE.      (Write total electrostatic potential into LOCPOT or not)
    
12. # LVHAR  = .TRUE.      (Write ionic + Hartree electrostatic potential into LOCPOT or not)
    
13. # NELECT =             (No. of electrons: charged cells, be careful)
    
14. # LPLANE = .TRUE.      (Real space distribution, supercells)
    
15. # NWRITE = 2           (Medium-level output)
    
16. # KPAR   = 2           (Divides k-grid into separate groups)
    
17. # NGXF    = 300        (FFT grid mesh density for nice charge/potential plots)
    
18. # NGYF    = 300        (FFT grid mesh density for nice charge/potential plots)
    
19. # NGZF    = 300        (FFT grid mesh density for nice charge/potential plots)
    
20. 
    
21. Static Calculation
    
22. ISMEAR =  -5           (tethdl smearing method)
    
23. SIGMA  =  0.05         (please check the width of the smearing)
    
24. LORBIT =  11           (PAW radii for projected DOS)
    
25. NEDOS  =  2001         (DOSCAR points)
    
26. NELM   =  60           (Max electronic SCF steps)
    
27. EDIFF  =  1E-08        (SCF energy convergence, in eV)
    
28. <blockquote>EMAX = 20

复制代码

NCORE = 12


KPOINTS_DOS：

1. K-Spacing Value to Generate K-Mesh: 0.025
   
2. 0
   
3. Monkhorst-Pack
   
4.    7   7   7
   
5. 0.0  0.0  0.0
   

复制代码

INCAR_BAND：

1. Global Parameters
   
2. ISTART =  1            (Read existing wavefunction, if there)
   
3. ISPIN  =  1            (Non-Spin polarised DFT)
   
4. ICHARG =  11         (Non-self-consistent: GGA/LDA band structures)
   
5. LREAL  = .FALSE.       (Projection operators: automatic)
   
6. ENCUT  =  1200          (Cut-off energy for plane wave basis set, in eV)
   
7. # PREC   =  Accurate   (Precision level: Normal or Accurate, set Accurate when perform structure lattice relaxation calculation)
   
8. LWAVE  = .TRUE.        (Write WAVECAR or not)
   
9. LCHARG = .TRUE.        (Write CHGCAR or not)
   
10. ADDGRID= .TRUE.        (Increase grid, helps GGA convergence)
    
11. # LVTOT  = .TRUE.      (Write total electrostatic potential into LOCPOT or not)
    
12. # LVHAR  = .TRUE.      (Write ionic + Hartree electrostatic potential into LOCPOT or not)
    
13. # NELECT =             (No. of electrons: charged cells, be careful)
    
14. # LPLANE = .TRUE.      (Real space distribution, supercells)
    
15. # NWRITE = 2           (Medium-level output)
    
16. # KPAR   = 2           (Divides k-grid into separate groups)
    
17. # NGXF    = 300        (FFT grid mesh density for nice charge/potential plots)
    
18. # NGYF    = 300        (FFT grid mesh density for nice charge/potential plots)
    
19. # NGZF    = 300        (FFT grid mesh density for nice charge/potential plots)
    
20. 
    
21. Static Calculation
    
22. ISMEAR =  0            (gaussian smearing method)
    
23. SIGMA  =  0.05         (please check the width of the smearing)
    
24. LORBIT =  11           (PAW radii for projected DOS)
    
25. NEDOS  =  2001         (DOSCAR points)
    
26. NELM   =  60           (Max electronic SCF steps)
    
27. EDIFF  =  1E-08        (SCF energy convergence, in eV)
    
28. EMAX = 20
    
29. EMIN = -20
    
30. 
    
31. NCORE = 12
    
32. 
    

复制代码

KPOINTS_BAND：

1. K-Path Generated by VASPKIT.
   
2.    20
   
3. Line-Mode
   
4. Reciprocal
   
5.    0.0000000000   0.0000000000   0.0000000000     GAMMA
   
6.    0.0000000000   0.5000000000   0.0000000000     X
   
7. 
   
8.    0.0000000000   0.5000000000   0.0000000000     X
   
9.    0.5000000000   0.5000000000   0.0000000000     M
   
10. 
    
11.    0.5000000000   0.5000000000   0.0000000000     M
    
12.    0.0000000000   0.0000000000   0.0000000000     GAMMA
    
13. 
    
14.    0.0000000000   0.0000000000   0.0000000000     GAMMA
    
15.    0.5000000000   0.5000000000   0.5000000000     R
    
16. 
    
17.    0.5000000000   0.5000000000   0.5000000000     R
    
18.    0.0000000000   0.5000000000   0.0000000000     X
    
19. 
    
20.    0.5000000000   0.5000000000   0.5000000000     R
    
21.    0.5000000000   0.5000000000   0.0000000000     M
    

复制代码

计算能带时所用的电荷密度使用了计算DOS时算出的密度。
一开始计算的时候没有带 ENCUT = 1200，EMAX = 20，EMIN = -20 这三个参数，发现占据态的计算基本正确但未占据态有很多没算出来，我猜测是基组不够大的原因，因此加了这三个参数，但是仍然算不出较高的未占据态。materialsproject 给出的计算结果（见https://materialsproject.org/materials/mp-30056/）如下：

而我计算的结果如下：

我的计算结果与 materialsproject 给出的计算结果相比，占据态的计算基本正确，但未占据态有很多没算出来。
我想请问一下，我的计算流程有没有什么问题？我的结果有很多态没算出来，仍然是可以接受的吗？如果想把更多的态也算出来，就像 materialsproject 给出的计算结果那样，应该如何操作？谢谢！

hebrewsnabla

NBANDS=40