Molcas Forum
Support and discussions for Molcas and OpenMolcas users and developers
You are not logged in.
- Topics: Active | Unanswered
Announcement
Please note: The forum's URL has changed. The new URL is: https://molcasforum.univie.ac.at. Please update your bookmarks!
You can choose an avatar and change the default style by going to "Profile" → "Personality" or "Display".Pages: 1
#1 2016-11-26 16:05:01
- wull
- Member
- Registered: 2016-11-15
- Posts: 5
the combination of ECP and AMFI
Dear admin,
As far as I know, for the atomic mean-field (AMFI), Douglas-Kroll (DOUG) hamiltonian must be employed. However the pseudopotential for metals already contain the spin-orbit operators. So, when I combine the AMFI and the ECP to conduct the SO-calculations, the values is about twice compared with the experimental value. We have tested that the splitting values (7F0-7F6) for the Eu3+ ion is ~9400cm-1 when the combination of AMFI and ECP is used, whereas the experimental value is only ~4900cm-1. But when I choose the all-electron basis , the value is ~5900cm-1. The splittings are shown below.What can I do to solve this problem?
ps: ECP is necessary for my system to consider the relativistic effect.
For AMFI+ECP
Eigenvalues of complex Hamiltonian:
-----------------------------------
Total energy (au) Rel lowest level(eV) D:o, cm**(-1)
1 -707.00041557 0.000000 0.000
2 -706.99837646 0.055487 447.535
3 -706.99837646 0.055487 447.535
4 -706.99837646 0.055487 447.535
5 -706.99429822 0.166462 1342.604
6 -706.99429822 0.166462 1342.604
7 -706.99429822 0.166462 1342.604
8 -706.99429822 0.166462 1342.604
9 -706.99429822 0.166462 1342.604
10 -706.98818087 0.332923 2685.208
11 -706.98818087 0.332923 2685.208
12 -706.98818087 0.332923 2685.208
13 -706.98818087 0.332923 2685.208
14 -706.98818087 0.332923 2685.208
15 -706.98818087 0.332923 2685.208
16 -706.98818087 0.332923 2685.208
17 -706.98002440 0.554872 4475.346
18 -706.98002440 0.554872 4475.346
19 -706.98002440 0.554872 4475.346
20 -706.98002440 0.554872 4475.346
21 -706.98002440 0.554872 4475.346
22 -706.98002440 0.554872 4475.346
23 -706.98002440 0.554872 4475.346
24 -706.98002440 0.554872 4475.346
25 -706.98002440 0.554872 4475.346
26 -706.96982881 0.832308 6713.019
27 -706.96982881 0.832308 6713.019
28 -706.96982881 0.832308 6713.019
29 -706.96982881 0.832308 6713.019
30 -706.96982881 0.832308 6713.019
31 -706.96982881 0.832308 6713.019
32 -706.96982881 0.832308 6713.019
33 -706.96982881 0.832308 6713.019
34 -706.96982881 0.832308 6713.019
35 -706.96982881 0.832308 6713.019
36 -706.96982881 0.832308 6713.019
37 -706.95759410 1.165232 9398.227
38 -706.95759410 1.165232 9398.227
39 -706.95759410 1.165232 9398.227
40 -706.95759410 1.165232 9398.227
41 -706.95759410 1.165232 9398.227
42 -706.95759410 1.165232 9398.227
43 -706.95759410 1.165232 9398.227
44 -706.95759410 1.165232 9398.227
45 -706.95759410 1.165232 9398.227
46 -706.95759410 1.165232 9398.227
47 -706.95759410 1.165232 9398.227
48 -706.95759410 1.165232 9398.227
49 -706.95759410 1.165232 9398.227For all electron
Eigenvalues of complex Hamiltonian:
-----------------------------------
(Shifted by EVAC (a.u.) = -9000.0)
Relative EVac(au) Rel lowest level(eV) D:o, cm**(-1)
1 -748.70813691 0.000000 0.000
2 -748.70684249 0.035223 284.093
3 -748.70684249 0.035223 284.093
4 -748.70684249 0.035223 284.093
5 -748.70425364 0.105669 852.280
6 -748.70425364 0.105669 852.280
7 -748.70425364 0.105669 852.280
8 -748.70425364 0.105669 852.280
9 -748.70425364 0.105669 852.280
10 -748.70037037 0.211338 1704.560
11 -748.70037037 0.211338 1704.560
12 -748.70037037 0.211338 1704.560
13 -748.70037037 0.211338 1704.560
14 -748.70037037 0.211338 1704.560
15 -748.70037037 0.211338 1704.560
16 -748.70037037 0.211338 1704.560
17 -748.69519267 0.352231 2840.933
18 -748.69519267 0.352231 2840.933
19 -748.69519267 0.352231 2840.933
20 -748.69519267 0.352231 2840.933
21 -748.69519267 0.352231 2840.933
22 -748.69519267 0.352231 2840.933
23 -748.69519267 0.352231 2840.933
24 -748.69519267 0.352231 2840.933
25 -748.69519267 0.352231 2840.933
26 -748.68872055 0.528346 4261.400
27 -748.68872055 0.528346 4261.400
28 -748.68872055 0.528346 4261.400
29 -748.68872055 0.528346 4261.400
30 -748.68872055 0.528346 4261.400
31 -748.68872055 0.528346 4261.400
32 -748.68872055 0.528346 4261.400
33 -748.68872055 0.528346 4261.400
34 -748.68872055 0.528346 4261.400
35 -748.68872055 0.528346 4261.400
36 -748.68872055 0.528346 4261.400
37 -748.68095400 0.739685 5965.960
38 -748.68095400 0.739685 5965.960
39 -748.68095400 0.739685 5965.960
40 -748.68095400 0.739685 5965.960
41 -748.68095400 0.739685 5965.960
42 -748.68095400 0.739685 5965.960
43 -748.68095400 0.739685 5965.960
44 -748.68095400 0.739685 5965.960
45 -748.68095400 0.739685 5965.960
46 -748.68095400 0.739685 5965.960
47 -748.68095400 0.739685 5965.960
48 -748.68095400 0.739685 5965.960
49 -748.68095400 0.739685 5965.960Last edited by wull (2016-11-26 16:30:58)
Offline
#2 2017-02-23 17:14:53
- nikolay
- Member
- From: Stuttgart
- Registered: 2016-03-21
- Posts: 54
Re: the combination of ECP and AMFI
Hi wull,
currently with Molcas it is only possible to run SO-calculations with all-electron basis set via AMFI.
If you care about the relativistic effects, the good choice is also use ANO-RCC or some other basis set with the -DK suffix together with the Douglas-Kroll or X2C option.
At the moment ECP's in Molcas do not include the spin-orbit potential part, therefore there is no SOC contribution from the core electrons.
In principle it is possible to implement the SOC part of the ECP (it has the same form as the scalar relativistic part) and combine it together with AMFI in some fashion.
Is there an expert to comment on that?
I also wonder what is the situation with combining ECPs and AMFI?
Does the calculation simply lacks the SOC at sites with ECPs or the results are spoiled?
Offline
#3 2017-02-28 13:49:44
- wull
- Member
- Registered: 2016-11-15
- Posts: 5
Re: the combination of ECP and AMFI
The current basis set libraries in Molcas for Eu are only /Eu.ECP.Dolg.7s6p5d.5s4p3d.11e-MWB. and /Eu.ECP.Dolg.7s6p5d.5s4p3d.10e-MWB. These two do not contain the f shell of Eu.
Therefore, we use the energy-consistent scalar-relativistic WB-adjusted 28-electron-core pseudopotential associated with the ECP28MWB_ANO (14s13p10d8f6g)/[6s6p5d4f3g] basis set for the Eu instead.
Unfortunately, we encounter the above mentioned problem. But, when I turn to molpro, the obtained splitting values are comparable with the experimental values. I don't what makes the difference.
Hi wull,
currently with Molcas it is only possible to run SO-calculations with all-electron basis set via AMFI.
If you care about the relativistic effects, the good choice is also use ANO-RCC or some other basis set with the -DK suffix together with the Douglas-Kroll or X2C option.At the moment ECP's in Molcas do not include the spin-orbit potential part, therefore there is no SOC contribution from the core electrons.
In principle it is possible to implement the SOC part of the ECP (it has the same form as the scalar relativistic part) and combine it together with AMFI in some fashion.
Is there an expert to comment on that?
I also wonder what is the situation with combining ECPs and AMFI?
Does the calculation simply lacks the SOC at sites with ECPs or the results are spoiled?
Offline
#4 2017-02-28 15:21:15
- nikolay
- Member
- From: Stuttgart
- Registered: 2016-03-21
- Posts: 54
Re: the combination of ECP and AMFI
The current basis set libraries in Molcas for Eu are only /Eu.ECP.Dolg.7s6p5d.5s4p3d.11e-MWB. and /Eu.ECP.Dolg.7s6p5d.5s4p3d.10e-MWB. These two do not contain the f shell of Eu.
There are other basis sets too, molcas help basis Eu gives
Recommended basis sets for Eu
Eu.ANO-RCC-MB Eu.ANO-RCC...6s5p3d1f.
Eu.ANO-RCC-VDZ Eu.ANO-RCC...7s6p4d2f.
Eu.ANO-RCC-VDZP Eu.ANO-RCC...7s6p4d2f1g.
Eu.ANO-RCC-VTZP Eu.ANO-RCC...8s7p5d3f2g1h.
Eu.ANO-RCC-VQZP Eu.ANO-RCC...9s8p6d4f3g2h.
Other basis sets for Eu
Eu.ANO-DK3.Tsuchiya.27s23p15d10f.6s4p2d1f.
Eu.ano-rcc.Roos.25s22p15d11f4g2h.12s11p8d7f4g2h.
Eu.ECP.Seijo.14s10p9d8f3g.2s1p1d1f1g.17e-CG-AIMP.
Eu.ECP.Seijo.14s10p10d8f3g.2s1p2d1f1g.27e-CG-AIMP.
Eu.ECP.Seijo.14s10p10d8f3g.3s2p2d1f1g.35e-CG-AIMP.
Eu.ECP.Tsuchiya.14s10p9d8f.2s1p1d1f.17e-DK3-AIMP.
Eu.ECP.Stoll.7s6p5d.5s4p3d.11e-MWB.
Eu.ECP.Stoll.7s6p5d.5s4p3d.10e-MWB.
Eu.ECP.Tsuchiya.14s10p9d8f.2s1p1d1f.17e-DK3-AIMP.
Eu.ECP.Dolg.7s6p5d.5s4p3d.11e-MWB.
Eu.ECP.Dolg.7s6p5d.5s4p3d.10e-MWB.But, when I turn to molpro, the obtained splitting values are comparable with the experimental values. I don't know what makes the difference.
As I said in my previous message ECPs consist of two parts: scalar relativistic and SOC.
In Molpro both are implemented, in Molcas there is only the first one (correct me if I'm wrong).
As far as I know, the only way to get SOC with Molcas at the moment is to use either Eu.ANO-RCC or Eu.ANO-DK3.
Offline
#5 2017-03-01 09:06:06
- wull
- Member
- Registered: 2016-11-15
- Posts: 5
Re: the combination of ECP and AMFI
Thank you very much for your suggestion. I have obtained the right split values by using either Eu.ANO-RCC or Eu.ANO-DK3 in Molcas. They both are all-electron basis set. The generation of the basis sets has been carried out without taking spin-orbit coupling into account.
However, the WB-adjusted 28-electron-core pseudopotential developed by M. Dolg et al has included the corresponding spin-orbit effects. The SO coupling will be calculated twice when this pseudopotential is used. Is this the reason why I got the strange split value before?
As far as I know, the only way to get SOC with Molcas at the moment is to use either Eu.ANO-RCC or Eu.ANO-DK3.
Offline
#6 2017-03-01 12:40:09
- nikolay
- Member
- From: Stuttgart
- Registered: 2016-03-21
- Posts: 54
Re: the combination of ECP and AMFI
Thank you very much for your suggestion. I have obtained the right split values by using either Eu.ANO-RCC or Eu.ANO-DK3 in Molcas. They both are all-electron basis set. The generation of the basis sets has been carried out without taking spin-orbit coupling into account.
I think with the posteriori perturbation SOC treatment as in Molcas or Molpro no special care is needed beyond scalar relativistic.
Using of ECP is only an approximation to the all-electron calculation with ANO/DK basis set.
However, the WB-adjusted 28-electron-core pseudopotential developed by M. Dolg et al has included the corresponding spin-orbit effects. The SO coupling will be calculated twice when this pseudopotential is used. Is this the reason why I got the strange split value before?
The original ECP28MWB potential taken, say, from here (Molpro format)
! Q=35., MEFIT, WB, Ref 9.
ECP,Eu,28,5,3; ! Lmax=5 for scalar, Lmax=3 for SOC
1; 2,1.000000,0.000000;
1; 2,23.471384,607.659331;
1; 2,16.772479,264.385476;
1; 2,13.981343,115.381375;
1; 2,23.962888,-49.400794;
1; 2,21.232458,-26.748273;
! Last three functions are for SOC
1; 2,16.772479,19.869243;
1; 2,13.981343,1.523881;
1; 2,23.962888,0.399191;
! References:
! [9] M. Dolg, H. Stoll, H. Preuss, J. Chem. Phys. 90, 1730 (1989).If you download it from the EMSL, you get (Molcas format, same for Molpro)
ECP
Eu nelec 28
Eu ul
2 1.000000000 0.000000000
Eu S
2 23.471384000 607.659331000
Eu P
2 16.772479000 264.385476000
Eu D
2 13.981343000 115.381375000
Eu F
2 23.962888000 -49.400794000
Eu G
2 21.232458000 -26.748273000
ENDwhich doesn't include the SOC part.
Moreover, at the moment there is no mechanism in Molcas to make use of that SOC part.
I hope someone else can comment on that.
Last edited by nikolay (2017-03-01 12:40:34)
Offline
Pages: 1