6.4 Open-shell O₂: SCF convergence, stability, and spin
These are unexecuted teaching inputs and starting models. Original diagrams are schematics, not calculated results. Validate version-specific syntax, licensed or authorized data, numerical convergence and the scientific model before using this workflow.
6.4.1 Model, units and provenance
Geometry in Å; electronic energy in hartree; vibrational wavenumbers in cm⁻¹. Check each printed field and keep thermal and standard-state terms distinct.
Shared inputs, conventions and evidence
6.4.2 Unexecuted inputs and explicit deltas
Use the accompanying instructions to identify the parent calculation and placement of every delta; a snippet is not automatically a standalone input. Preserve all blank-line and file-provenance requirements.
6.4.2.1 Input block 1
%chk=oxygen24.chk
%mem=2GB
%nprocshared=2
#p UB3LYP/6-31+G(d,p) Opt=Tight NoSymm
SCF=(Tight,XQC) Int=UltraFine
Triplet oxygen: initial unrestricted geometry optimization
0 3
O 0.000000 0.000000 -0.605000
O 0.000000 0.000000 0.605000
6.4.2.2 Input block 2
%chk=oxygen24.chk
%mem=2GB
%nprocshared=2
#p UB3LYP/6-31+G(d,p) Stable=Opt Geom=AllCheck Guess=Read
NoSymm SCF=(Tight,XQC) Int=UltraFine
6.4.2.3 Input block 3
%chk=oxygen24.chk
%mem=2GB
%nprocshared=2
#p UB3LYP/6-31+G(d,p) Opt=Tight Geom=AllCheck Guess=Read
NoSymm SCF=(Tight,XQC) Int=UltraFine
6.4.2.4 Input block 4
%chk=oxygen24.chk
%mem=2GB
%nprocshared=2
#p UB3LYP/6-31+G(d,p) Freq Geom=AllCheck Guess=Read
NoSymm SCF=(Tight,XQC) Int=UltraFine
6.4.3 Worked investigation
6.4.3.1 Question, intuition, and prerequisites
Can an SCF calculation converge to the wrong electronic solution? Yes. Convergence means an iterative equation met its stopping criteria. Stability asks whether the resulting determinant can lower its energy under tested orbital variations. Spin diagnostics ask another question: how well does an unrestricted determinant resemble the intended spin state? Triplet O₂ lets beginners separate these checks without a large system. Prerequisites are multiplicity 2S+1, alpha/beta electrons, geometry optimization, and checkpoint reuse.
6.4.3.2 Complete initial optimization
See input block 1 above.
6.4.3.3 Stability and conditional geometry loop
Run this after the initial optimization
See input block 2 above.
If stability optimization changes the solution, rerun the geometry optimization from that updated checkpoint
See input block 3 above.
Then repeat the stability job. When the geometry is converged on a stable solution and the stability check no longer changes it, run the frequency job
See input block 4 above.
Save a separate log for each job and preserve a backup of each important checkpoint before the next modifying step. The three deltas intentionally reuse the working checkpoint; their sequence is conditional, not a blind four-step script. Stable=Opt optimizes orbital/wavefunction stability, not nuclear geometry. If it finds a different electronic solution, the old geometry may no longer be stationary for that solution.
6.4.3.4 Numbered workflow
- Count electrons: neutral O₂ has 16. A triplet has \(S=1\), so the usual highest-MS determinant has nine alpha and seven beta electrons. Check the printed counts.
- At each converged SCF, record energy and
S**2before and after any reported spin annihilation. Keep the unannihilated diagnostic visible in the report. - Read the explicit stability conclusion. If orbitals are changed, compare energies at the same geometry before reoptimizing nuclei; do not confuse electronic relaxation with a structural energy difference.
- Iterate only as needed until electronic stability and geometry stationarity are both established. Verify that the final frequency calculation uses that same wavefunction, method, grid, and geometry.
- Inspect the single vibrational stretching mode of this linear diatomic. Optionally generate a spin-density cube with
cubegen 0 Spin=SCF oxygen24.fchk oxygen24_spin.cube 80 hafterformchk oxygen24.chk oxygen24.fchk.
6.4.3.5 Read and check
An ideal pure triplet has \(\langle S^2\rangle=S(S+1)=2.000\). Values appreciably larger than 2 indicate contamination by higher-spin components in the unrestricted determinant. There is no universal safe numeric threshold; report the value and judge its consequences for the intended property. In Kohn–Sham DFT, this determinant diagnostic should not be overinterpreted as an exact many-electron spin measurement. A value near 2 does not prove the wavefunction is stable; a stable wavefunction does not prove that the method or spin state is chemically correct.
A linear diatomic has \(3N-5=1\) vibrational mode. An O–O distance broadly around 1.1–1.4 Å is a useful beginner check near the usual triplet minimum, not a promised optimized result. The harmonic stretch should be real at a minimum. For an isolated homonuclear O₂ molecule, the ordinary electric-dipole harmonic IR intensity should vanish by symmetry; a nearly zero IR intensity does not mean the bond does not vibrate. The integrated spin density \(\rho_\alpha-\rho_\beta\) should approach \(N_\alpha-N_\beta=2\) with an adequate numerical grid.
Do not “fix” contamination by reporting only the post-annihilation value; printed annihilation does not automatically make all reported energies rigorously spin-pure. SCF=XQC helps difficult convergence but does not certify the correct root. Stable=Opt tests local determinant stability under its configured degrees of freedom, not every multiplicity or a global electronic minimum. Singlet oxygen is a qualitatively more delicate electronic-structure problem; changing 0 3 to 0 1 and subtracting two unrestricted energies is not automatically a reliable singlet–triplet gap. Seek an appropriate state-specific or multireference treatment for that question.
6.4.3.6 Exercise
EN exercise: Explain three logically distinct statements: “SCF converged”, “the determinant is stable under the tested variations”, and “the geometry is a minimum”. For each, identify one output check that supports it and one claim that it cannot establish. Compare an O₂ stretching calculation started at 1.00 and 1.50 Å, using separate checkpoints, and determine whether the same final triplet solution is reached.