From owner-chemistry@ccl.net Tue Nov 10 14:41:00 2020 From: "Lenka Kubickova kubickol%x%fzu.cz" To: CCL Subject: CCL:G: Question: "Low-spin" AFM complex in Gaussian Message-Id: <-54209-201110051451-6961-eoJWfQ1igQWhhmqYRQgSyQ _ server.ccl.net> X-Original-From: "Lenka Kubickova" Date: Tue, 10 Nov 2020 05:14:48 -0500 Sent to CCL by: "Lenka Kubickova" [kubickol*o*fzu.cz] Dear colleagues, I would like to ask you for help. I have been trying to calculate the energy of a binuclear Cr(II) organometallic complex, where the Cr(II) atoms are connected via hydride bridges, to determine the spin state and exchange coupling constant. In these complexes, Cr(II) can exist either in a "high-spin" state with 4 unpaired d-electrons (S = 2), or in a "low-spin" state with 2 paired and 2 unpaired electrons (S = 1). I have calculated the energies of the former for ferromagnetic (multiplicity M = 9) and antiferromagnetic (M = 1) coupling of Cr spins, even the ferromagnetic coupling of the latter (M = 5). However, every time I try running a calculation for the antiferromagnetic coupling of the Cr(II) with spins S = 1, the calculation converges to the antiferromagnetic high-spin solution with Cr spins S = 2. Please, do you have any idea how to persuade Gaussian to stick to the "low-spin" solution? I use unrestricted broken-symmetry DFT in Gaussian 16 and basically follow the instructions in https://gaussian.com/afc/ - start from experimental geometry (or optimized from the ferromagnetic case), divide the molecule into 4 fragments (2x(Cr+ligands), 2x(H- bridge)), create a guess, and use the created checkpoint in follow-up optimizations. The first rows of the guess look like this (leaving out the basis definitions for C, H, and Cr at the end): %chk=complex.chk %nprocshared=4 %mem=4GB # ub3lyp/gen pseudo=cards guess=(fragment=4,only) Unrestricted 18134o 0 1 1 3 1 -3 -1 1 -1 1 C(Fragment=2) 16.75200000 9.35300000 2.77500000 .. The follow-up optimization, which then converges to the "unwanted" high- spin solution: %chk=complex.chk %nproc=4 %mem=8GB #ub3lyp/gen opt freq scf=xqc guess=read geom=allcheck pseudo=cards .. Already applying the stable=opt calculation on the generated guess provides the high-spin AFM solution. The above-mentioned input lines worked well with the ferromagnetic "low-spin" case, I don't understand why changing just a sign of the multiplicity of one fragment changes the output so dramatically... Moreover, we know from magnetic measurements that this complex is antiferromagnetic and the XRD structure is closer to the "low-spin" FM geometry, so the solution I am not able to get with the calculations sketched above might be indeed the right one. Please, I would be really grateful if anyone has any idea how to solve this trouble... Many thanks and best wishes of good health to everyone Lenka Kubickova kubickol]*[fzu.cz From owner-chemistry@ccl.net Tue Nov 10 18:16:00 2020 From: "Kathrin Helen Hopmann kathrin.hopmann .. uit.no" To: CCL Subject: CCL:G: Question: "Low-spin" AFM complex in Gaussian Message-Id: <-54210-201110164636-31401-UxupLO3N6hm1JKw0HcotPg=server.ccl.net> X-Original-From: Kathrin Helen Hopmann Content-Language: en-US Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset="us-ascii" Date: Tue, 10 Nov 2020 21:46:28 +0000 MIME-Version: 1.0 Sent to CCL by: Kathrin Helen Hopmann [kathrin.hopmann_+_uit.no] Hi Lenka, I have never used the approach you describe below, but a general comment on spin states: A pure functional may be more likely to give you a 'low spin' solution. If you manage to get the low spin solution with another functional, you could use it as a starting guess for B3LYP (if you need to use B3LYP). But it could very well be that B3LYP will always converge to the high spin case. Kathrin -----Original Message----- > From: owner-chemistry+kathrin.hopmann==uit.no+*+ccl.net On Behalf Of Lenka Kubickova kubickol%x%fzu.cz Sent: tirsdag 10. november 2020 11:15 To: Kathrin Helen Hopmann Subject: CCL:G: Question: "Low-spin" AFM complex in Gaussian Sent to CCL by: "Lenka Kubickova" [kubickol*o*fzu.cz] Dear colleagues, I would like to ask you for help. I have been trying to calculate the energy of a binuclear Cr(II) organometallic complex, where the Cr(II) atoms are connected via hydride bridges, to determine the spin state and exchange coupling constant. In these complexes, Cr(II) can exist either in a "high-spin" state with 4 unpaired d-electrons (S = 2), or in a "low-spin" state with 2 paired and 2 unpaired electrons (S = 1). I have calculated the energies of the former for ferromagnetic (multiplicity M = 9) and antiferromagnetic (M = 1) coupling of Cr spins, even the ferromagnetic coupling of the latter (M = 5). However, every time I try running a calculation for the antiferromagnetic coupling of the Cr(II) with spins S = 1, the calculation converges to the antiferromagnetic high-spin solution with Cr spins S = 2. Please, do you have any idea how to persuade Gaussian to stick to the "low-spin" solution? I use unrestricted broken-symmetry DFT in Gaussian 16 and basically follow the instructions in https://gaussian.com/afc/ - start from experimental geometry (or optimized from the ferromagnetic case), divide the molecule into 4 fragments (2x(Cr+ligands), 2x(H- bridge)), create a guess, and use the created checkpoint in follow-up optimizations. The first rows of the guess look like this (leaving out the basis definitions for C, H, and Cr at the end): %chk=complex.chk %nprocshared=4 %mem=4GB # ub3lyp/gen pseudo=cards guess=(fragment=4,only) Unrestricted 18134o 0 1 1 3 1 -3 -1 1 -1 1 C(Fragment=2) 16.75200000 9.35300000 2.77500000 .. The follow-up optimization, which then converges to the "unwanted" high- spin solution: %chk=complex.chk %nproc=4 %mem=8GB #ub3lyp/gen opt freq scf=xqc guess=read geom=allcheck pseudo=cards .. Already applying the stable=opt calculation on the generated guess provides the high-spin AFM solution. The above-mentioned input lines worked well with the ferromagnetic "low-spin" case, I don't understand why changing just a sign of the multiplicity of one fragment changes the output so dramatically... Moreover, we know from magnetic measurements that this complex is antiferromagnetic and the XRD structure is closer to the "low-spin" FM geometry, so the solution I am not able to get with the calculations sketched above might be indeed the right one. Please, I would be really grateful if anyone has any idea how to solve this trouble... Many thanks and best wishes of good health to everyone Lenka Kubickova kubickol ~ fzu.czhttp://www.ccl.net/cgi-bin/ccl/send_ccl_messagehttp://www.ccl.net/chemistry/sub_unsub.shtmlhttp://www.ccl.net/spammers.txt