Optical properties of a diamond NV color center from capped embedded multiconfigurational correlated wavefunction theory
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | npj Computational Materials |
| Published | 2026-02-06 |
| Authors | John Mark P. Martirez |
| DOI | 10.1038/s41524-026-01987-1 |
| Citations | 3 |
Capped embedding with NEVPT2 reproduced the triplet and singlet vertical excitation energies of the negatively charged nitrogen-vacancy center with errors below 0.1 eV. A selected 40-atom capped cluster gave 2.25 eV for the triplet transition, 0.07 eV above the experimental 2.18 eV value. The same model gave a 0.35 eV singlet-triplet energy within the reported 0.32 to 0.41 eV window.
Accuracy against optical references
Section titled “Accuracy against optical references”The paper reports calculations rather than new optical measurements. Its references include visible absorption at liquid nitrogen temperature after 2 MeV electron irradiation at room temperature and annealing. Other data use fluorescence at 4 K with 532 nm excitation, infrared absorption at 10 K with a 35 mW pump, and intersystem crossing rates.
All entries use the DFT-HSE06 embedding potential, the apvdz basis, and the CAS(6e,8o) active space.
| Transition | Calculated vertical energy | Experimental reference | Comparison |
|---|---|---|---|
| Triplet, (^{3}A_2) to (^{3}E) | 2.25 eV | 2.18 eV | 0.07 eV higher |
| Singlet, (^{1}E) to (^{1}A_1) | 1.28 eV | 1.26 eV | 0.02 eV higher |
| (^{1}A_1) and (^{3}E) crossing | 0.35 eV | 0.32 to 0.41 eV | Within reported window |
These calculated values are vertical excitations with fixed nuclei. They are not zero phonon lines, which include structural relaxation after excitation. The measured zero phonon lines were 1.95 eV for the triplet transition and 1.19 eV for the singlet transition. Vertical and relaxed energies must therefore remain separate during comparison.
Why multiconfigurational states require a correlated treatment
Section titled “Why multiconfigurational states require a correlated treatment”The negatively charged nitrogen-vacancy center has local (C_{3v}) symmetry and four low energy states used in the paper. The triplet (^{3}A_2) ground state is nondegenerate and has a doubly degenerate (^{3}E) excited state. The singlet (^{1}E) ground state is doubly degenerate, while the (^{1}A_1) excited state is nondegenerate.
State-averaged CASSCF with six electrons in eight orbitals captured static correlation. The triplet ground configuration accounted for 96% of the wavefunction, while the triplet excited configurations accounted for 93% of electron correlation. The singlet states used four and three leading configurations with similar weights.
NEVPT2 then added dynamic correlation. The singlet excited state included a double excitation with a squared coefficient of about 0.28. The authors conclude that methods limited to single excitations, including time-dependent DFT, will give incorrect predictions for this state.
Calculated triplet partners differed by 0.36 meV, while the singlet partners differed by 0.23 meV. Neglecting spin-orbit coupling, the authors found that the triplet-to-singlet crossing is not a simple spin flip.
Building the capped embedded model
Section titled “Building the capped embedded model”Periodic DFT calculations used VASP 6.4 with spin polarization and a uniform positive background for negatively charged supercells. The defective cells were (C_{62}N^{-}), (C_{126}N^{-}), and (C_{158}N^{-}). The original cells contained 64, 128, or 160 atoms. In each case, one carbon was removed and an adjacent carbon was replaced by nitrogen.
Structures were relaxed with (r^2)SCAN-L. That functional predicted a 3.5677 Å lattice constant, close to the room temperature measured value of 3.56712 ± 0.00005 Å. Its 4.22 ± 0.02 eV band gap underpredicted the measured 5.480 ± 0.004 eV gap.
For pristine C8 diamond, HSE06 gave 5.30 ± 0.02 eV, while G0W0 with DFT-PBE gave 5.50 ± 0.02 eV. Electronic structures and embedding potentials used HSE06 because of its band gap result. Its orbital energy difference was 2.38 eV, which the authors treated as a reasonable upper bound to the triplet vertical excitation.
The (C_{15}N) cluster contained three vacancy-adjacent carbons, three nitrogen-bound carbons, and nine additional nearest neighbors. Twelve F atoms capped terminal bonds, twelve O atoms capped bridge bonds, and twelve B atoms capped the (C_{47}) environment. The auxiliary fragment combined the (F_{12}O_{12}B_{12}) capping set. Together, the cluster and its caps formed a 40-atom nonperiodic model.
The embedding potential matched the fragment densities to the periodic HSE06 reference. The ground-state triplet potential and structure were retained for both spins, keeping the Hamiltonian fixed for direct energy comparisons.
State-averaged CASSCF included three states per spin without imposed symmetry. NEVPT2 used apvdz basis sets. An apvtz basis for carbon and nitrogen, with apvdz for fluorine and oxygen, changed energies by about 0.01 eV.
The triplet excitation fell 0.07 eV when the active space expanded from CAS(4e,6o) to CAS(6e,8o). The larger space added the nitrogen lone pair and its correlating orbital. Expansion to CAS(10e,12o) lowered it by another 0.01 to 0.02 eV. This supported CAS(6e,8o) as the practical optimum. The full CAS(12e,14o) reference could not support the subsequent MRPT2 calculation because of memory requirements.
Convergence with cluster and supercell size
Section titled “Convergence with cluster and supercell size”The same 40-atom embedded cluster was carved from each of the three periodic supercells. Its triplet excitation stayed at 2.25 eV for every supercell origin. Singlet excitation varied from 1.27 to 1.28 eV, and crossing energy varied from 0.35 to 0.37 eV.
All entries below use SS-NEVPT2, CAS(6e,8o), apvdz, and the DFT-HSE06 embedding potential.
| Cluster and supercell origin | Triplet energy | Singlet energy | Crossing energy |
|---|---|---|---|
| ([C_{15}NF_{12}O_{12}]^{-}) from (C_{62}N^{-}) | 2.25 eV | 1.28 eV | 0.35 eV |
| ([C_{15}NF_{12}O_{12}]^{-}) from (C_{126}N^{-}) | 2.25 eV | 1.27 eV | 0.37 eV |
| ([C_{15}NF_{12}O_{12}]^{-}) from (C_{158}N^{-}) | 2.25 eV | 1.28 eV | 0.36 eV |
| ([C_{21}NF_{18}O_{9}]^{-}) from (C_{126}N^{-}) | 2.24 eV | 1.22 eV | 0.44 eV |
For the smallest cluster, changing the supercell origin changed vertical excitation energies by at most 0.01 eV. The crossing energy changed by at most 0.02 eV.
The final cluster calculation is nonperiodic and contains no interaction with charged defect images. The authors attribute this weak dependence to the potential’s local dielectric response. They report no more than 0.05 eV cluster-size dependence for vertical excitation energies and 0.07 eV for crossing.
Prior periodic approaches in the paper required 255-atom GW+BSE and 511-atom CI-cRPA supercells. For the triplet transition, cited GW+BSE gave 2.32 eV and CI-cRPA gave 2.05 eV. Other cited embedding methods used supercells containing 215 or 511 atoms and required orbital localization. A cluster-only calculation needed 85 carbon atoms for 0.2 eV convergence against a 33 carbon cluster.
Bare capped clusters showed nonmonotonic triplet behavior as carbon was added. The singlet transition appeared converged at ([C_{24}NF_{18}O_{12}]^{-}) but remained 0.07 eV below the embedded ([C_{21}NF_{18}O_{9}]^{-}) benchmark. The authors conclude that no computationally viable bare cluster matched the embedded accuracy with the required basis and active space.
Engineering meaning and calculation limits
Section titled “Engineering meaning and calculation limits”The calculations address optical-state engineering for negatively charged nitrogen-vacancy qubits. The visible triplet transition is used for optical excitation and spin-dependent fluorescence. The singlet manifold provides the shelving route through intersystem crossing, which affects optical readout.
The selected calculations give a 2.25 eV vertical triplet excitation and a 0.35 eV crossing energy. These values describe electronic energy differences, not coherence, optical linewidth, or intersystem crossing rate. The measured zero phonon lines, 1.95 and 1.19 eV, remain the relaxed optical energies. Device-level comparisons of triplet and singlet states should retain multireference treatment because single-determinant methods miss the singlet configurations.
Vertical excitation calculations keep the nuclei fixed, so they do not provide relaxed excited-state geometries or phonon processes. The same embedding potential and triplet ground-state structure were used for singlet calculations. This assumes the environment’s dielectric screening does not change strongly during excitation or spin crossover. Spin-orbit coupling was omitted from the spin crossover interpretation.
The authors limit reliable embedding use to defects that remain electronically localized within a few atoms. For ionization potentials, electron affinities, and formation energies between systems with different excess charges, the authors say electrostatic corrections remain necessary. The supported use is comparative optical-level modeling of localized NV states, rather than bulk charge-state energetics.
Materials and support
Section titled “Materials and support”For nitrogen-vacancy quantum-device studies, single-crystal or polycrystalline diamond is the relevant host family, although nitrogen-vacancy fabrication is not specified in the catalog facts. The paper needs a crystalline host with a localized substitutional nitrogen site next to a vacancy. 6CCVD’s diamond materials catalog offers MPCVD-grown single-crystal and polycrystalline diamond plates and wafers up to 125 mm, while engineering support covers custom fabrication.
View Original Abstract
Diamond defects are among the most promising qubits. Modeling their properties through accurate quantum mechanical simulations can further their development into robust units of information. We use the recently developed capped density functional embedding theory (capped-DFET) with the multiconfigurational n-electron valence second-order perturbation theory to characterize the electronic excitation energies for different spin manifolds of the well-characterized negatively charged substitutional N defect adjacent to a vacancy (VC) in diamond (NCVC−). We successfully reproduce vertical excitation energies for both triplet and singlet states of NCVC− with errors < 0.1 eV. Unlike other embedding methods, capped-DFET exhibits robust predictions that are approximately independent of the embedded cluster size: it only requires a cluster to contain the defect atoms and their nearest neighbors (as small as a 40-atom capped cluster). Furthermore, our method is free from slowly converging Coulomb interactions between charged defects, and thus also only weakly dependent on supercell size.
High Concentration Quantum Grade Single Crystal Diamond
View Specs →Products
View Specs →
Single Crystal Cvd Diamond Plate
View Specs →Citing papers we cover
Section titled “Citing papers we cover”- A Chemist’sGuide to Solid-State Qubits - DiamondColor Centers as Embedded Vacancy-Centered Molecules
- Advances in Quantum Defect Embedding Theory