计算化学公社

标题: 求助：VASP优化表面吸附小分子不收敛 [打印本页]

作者
Author: suhaiweias 时间: 2024-5-17 13:04
标题: 求助：VASP优化表面吸附小分子不收敛
大家好，我想请问一下：我之前用一个INCAR优化好晶胞和表面，后来像在表面吸附CO2等小分子，发现怎么也不收敛了。尝试了很多了方法，还是没有效果，请大佬们帮忙给点建议

INCAR:

!Initialisation:
NWRITE = 2       (Medium-level output)
ISTART = 0       (Read existing wavefunction; otherwise 0 for a new start)
INIWAV = 1       (Random initial wavefunction; otherwise)
ICORELEVEL = 1 (Print core levels)
ICHARG = 1 (11 Non-self-consistent: GGA/LDA band structures)

!Parallelisation:
LPLANE = .TRUE. (Real space distribution; supercells)
NCORE = 4    (Max is no. nodes; don't set for hybrids)
KPAR = 1    (Divides k-grid into separate groups)

!Electronic Relaxation:
!!PREC  = Accurate  (Precision level)
LREAL = Auto    (Projection operators: automatic)
!!LREAL = .FALSE. (Accurate forces for phonon calculations)
ALGO  = Normal    (SCF minimisation algorithm; 38/48 combo)
IALGO = 38
ENCUT = 450.00 eV (Plane-wave cutoff)
NELM  = 300       (Max SCF steps)
NELMIN = 5       (Min SCF steps)
EDIFF = 1E-05    (SCF energy convergence; eV)
!!ISPIN =  1       (Closed shell DFT)
ISPIN = 2    (Spin polarised DFT)
GGA    =  PE    (Perdew-Burke-Ernzerhof exchange-correlation)

!Ionic Relaxation:
EDIFFG =  -0.05 (Ionic convergence; eV/AA)
NSW =    300    (Max ionic steps)
NBLOCK =    1    (Update XDATCAR every X steps)
IBRION =    2    (Algorithm: 0-MD; 1-Quasi-New; 2-CG)
ISIF =    2    (Stress/relaxation: 2-Ions, 3-Shape/Ions/V, 7-Vol)
ISYM =    0    (Symmetry: 0=none; 2=GGA; 3=hybrids)
ISMEAR =    0    (0-Gaussian smearing)
                  (metal: 1 or 2)
                  (semiconductors/insulators: -5; for large cell (k-points <=2 : 0))
SIGMA  =    0.05  (Smearing value in eV)
                  (metals:0.2 (keep T*S < 1 meV/atom, otherwise reduce SIGMA))

#### DFT + U ####
LDAU = .TRUE.  (Activate DFT+U)
LDAUTYPE = 2    (Dudarev; only U-J matters)
LDAUL =  -1  2  -1  (Orbitals for each species)
LDAUU = 0  6.2  0  (U for each species)
LDAUJ = 0  0  0 (J for each species)
LMAXMIX =  4    (Mixing cut-off; 4-d, 6-f)

材料是CuO和WO3的一种固溶体，POSCAR顺序是Cu W O，晶胞和表面优化均能够正常收敛。查看OSZICAR时，每次电子步只跑到50多步就停了，一个离子步都没跑完。后来尝试过改ALGO，添加MAGMOM，但好像还是没什么效果...麻烦各位

作者
Author: jiangning198511 时间: 2024-5-17 16:19

suhaiweias 发表于 2024-5-17 13:07
菜鸟新手一个，请大家路过帮忙看看

你这个DFT+U写的好像有问题，应该包括所有元素，你这个输入好像只包括表面原子而没有吸附分子

作者
Author: 白皛白 时间: 2024-5-17 17:12

suhaiweias 发表于 2024-5-17 13:07
菜鸟新手一个，请大家路过帮忙看看

INCAR里好像加了自旋极化却没有设置磁矩的初猜，以个人经验来说，使用ISPIN = 2 时一定要给出比较合理的MAGMOM。
另外也提供一个方法，这个方法貌似使用的挺普遍的。
step 1，INCAR中设置LWAVE = .T. ，然后看OSZICAR里SCF步骤最接近收敛的那步收敛到什么程度了。我举个例子，比如收敛到0.2 E-03了，那你就设置EDIFF = 1E-03，然后做结构优化直到收敛。
step 2 ，INCAR中设置ISTART = 1;ICARGE = 0；EDIFF = 1E-5。cp CONTCAR POSCAR，再做一次结构优化。（如果EDIFF = 1E-5收敛不了就EDIFF = 1E-4，重复step1和2渐渐严格SCF的收敛阈值直到你想要的水平）
使用此法时，请仔细检查粗优化的最终的结构是否合理再进行下一步。
不过一切方法的前提是，你的初猜的结构不会太不合理，如果初猜结构就不合理的话，什么方法与参数都不会奏效。
也可以参考这个帖子http://bbs.keinsci.com/thread-11126-1-1.html

作者
Author: suhaiweias 时间: 2024-5-17 17:49

jiangning198511 发表于 2024-5-17 16:19
你这个DFT+U写的好像有问题，应该包括所有元素，你这个输入好像只包括表面原子而没有吸附分子

哦不好意思，谢谢您的答复。这个是最开始优化表面的INCAR。最新版吸附CO的INCAR在这，DFT+U和POTCAR都有对应POSCAR的元素顺序。另外，我的Cu没有+U，只对W加U了。后来尝试过Cu也加U（根据文献+7 eV），结果连晶格常数优化都不收敛。

!Initialisation:
NWRITE = 2       (Medium-level output)
ISTART = 0       (Read existing wavefunction; otherwise 0 for a new start)
INIWAV = 1       (Random initial wavefunction; otherwise)
ICORELEVEL = 1 (Print core levels)
!!NELECT = 352  (No. of electrons: charged cells; be careful)
!!NBANDS = 35 (No. of bands)
!!LVTOT = .TRUE.
!!LVHAR = .TRUE.
!!LELF = .TRUE.

!Parallelisation:
LPLANE = .TRUE. (Real space distribution; supercells)
NCORE = 12    (Max is no. nodes; don't set for hybrids)
KPAR = 1    (Divides k-grid into separate groups)

!Electronic Relaxation:
!!PREC  = Accurate  (Precision level)
LREAL = Auto    (Projection operators: automatic)
!!LREAL = .FALSE. (Accurate forces for phonon calculations)
ALGO  = Fast    (SCF minimisation algorithm; 38/48 combo)
IALGO = 48
ENCUT = 450.00 eV (Plane-wave cutoff)
NELM  = 120       (Max SCF steps)
NELMIN = 5       (Min SCF steps)
EDIFF = 1E-05    (SCF energy convergence; eV)
!!ISPIN =  1       (Closed shell DFT)
ISPIN = 2    (Spin polarised DFT)
GGA    =  PE    (Perdew-Burke-Ernzerhof exchange-correlation)
NPAR = 5

!Ionic Relaxation:
EDIFFG =  -0.05 (Ionic convergence; eV/AA)
NSW =    300    (Max ionic steps)
NBLOCK =    1    (Update XDATCAR every X steps)
IBRION =    2    (Algorithm: 0-MD; 1-Quasi-New; 2-CG)
ISIF =    2    (Stress/relaxation: 2-Ions, 3-Shape/Ions/V, 7-Vol)
ISYM =    0    (Symmetry: 0=none; 2=GGA; 3=hybrids)
ISMEAR =    0    (0-Gaussian smearing)
                  (metal: 1 or 2)
                  (semiconductors/insulators: -5; for large cell (k-points <=2 : 0))
SIGMA  =    0.05  (Smearing value in eV)
                  (metals:0.2 (keep T*S < 1 meV/atom, otherwise reduce SIGMA))

!Miscellaneous:
LORBIT = 11    (PAW radii for projected DOS)
!!LORBIT  = 1    (Supply radii for projected DOS)
!!RWIGS =  1.5 1.5 (Radii for each atomic species)
!!LOPTICS = .TRUE. (Dielectric functions)
!!NEDOS    =  1000 (DOSCAR points)
!!LVTOT = .TRUE. (Electrostatic potential)
!!LELF = .TRUE. (Localisation function)
!!NGX    = 500    (FFT grid mesh density; nice plots)
!!NGY    = 500    (FFT grid mesh density)
!!NGZ    = 500    (FFT grid mesh density)

!Magnetic Calculations:
MAGMOM  = 24*0 24*5 97*0    (Initial magnetic momoment on each ion)
!!NUPDOWN = -1    (Enforce spin multiplet)
!!LSORBIT = .TRUE.  (Spin-orbit coupling)

#### DFT + U ####
LDAU = .TRUE.  (Activate DFT+U)
LDAUTYPE = 2    (Dudarev; only U-J matters)
LDAUL = -1  2  -1  -1 (Orbitals for each species)
LDAUU = 0 6.2  0  0 (U for each species)
LDAUJ = 0  0  0  0 (J for each species)
LMAXMIX =  4    (Mixing cut-off; 4-d, 6-f)

作者
Author: suhaiweias 时间: 2024-5-17 17:58

白皛白发表于 2024-5-17 17:12
INCAR里好像加了自旋极化却没有设置磁矩的初猜，以个人经验来说，使用ISPIN = 2 时一定要给出比较合理的M ...

好的谢谢，之前我注意到OSZICAR里的电子步会收敛到E0，可是马上就会升回去E2或E3，如下图所示。
然后我试过直接将ICHARG=1删掉，然后设置初始磁矩，添加了MAGMOM。这是我昨天晚上跑的最新INCAR，还是不可以收敛

。我之前试过设置EDIFF = 1E-3，也不收敛，能量达到负的好几万... 链接我会好好去学习一下，感谢

欢迎光临计算化学公社 (http://bbs.keinsci.com/)