First-Principles Calculations of Internal Conversion Processes in Spin Defects
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | Physical Review Letters |
| Published | 2026-09-14 |
| Authors | Anonymous, Yu Jin, Giulia Galli |
| DOI | 10.1103/y6z6-h16k |
The framework predicts an 8.52 GHz internal-conversion rate for the NV⁻ singlet transition at 78 K, corresponding to a 117.4 ps lifetime. The result agrees with femtosecond transient-absorption lifetimes of 102.3(3) ps and 92.1(5) ps from two samples. It also identifies a previously missed nonradiative contribution to the triplet lifetime of the kk-VV⁰ center in 4H-SiC.
Rate calculation with many-body states
Section titled “Rate calculation with many-body states”Internal conversion is a spin-conserving transition between electronic states through multiphonon exchange. The paper calculates its temperature-dependent rate with Fermi’s golden rule at first order in the electron-phonon interaction. The initial vibrational population follows the equilibrium distribution at the specified temperature. Final vibrational states are restricted by total energy conservation.
LR-TDDFT with hybrid functionals supplies many-body excited-state wavefunctions. This treatment captures multi-configurational effects that single-particle Kohn-Sham orbitals miss. Nonadiabatic coupling vectors are obtained analytically through an extended Lagrangian formulation in the WEST code. Analytical derivatives avoid finite-difference wavefunction derivatives and allow every phonon mode to enter the rate.
The displaced harmonic oscillator approximation represents vibrational overlap through temperature-dependent spectral functions. A generating function then evaluates the line shape and enforces energy conservation. The rate is read from that line shape at the adiabatic energy difference between the relaxed initial and final states.
Quantum ESPRESSO supplied the DFT calculations with a 60 Ry cutoff and norm-conserving Vanderbilt pseudopotentials. WEST handled TDDFT and analytical couplings. Phonopy generated dynamical matrices, and PyPL calculated line shapes.
NV⁻ singlet decay matches transient absorption
Section titled “NV⁻ singlet decay matches transient absorption”The ¹A₁ to ¹E transition of NV⁻ in diamond has an infrared zero-phonon line near 1.19 eV. Its measured integrated emission intensity is three orders of magnitude below the visible ³E to ³A₂ emission. The authors used a 215-atom supercell and DDH hybrid functional. TDDFT gave an adiabatic energy difference of 1.397 eV, while PBE gave qualitatively similar rates.
The calculated rates are of order 10¹ GHz from 0 to 400 K. At 78 K, the all-mode rate is 8.52 GHz and the calculated lifetime is 117.4 ps. The predicted radiative rate is 0.53 MHz. That separation makes internal conversion the main decay route for the ¹A₁ state.
At 4 K, the all-mode calculation gives 8.51 GHz. Acosta et al. used phase-shift measurements of infrared fluorescence and reported a lifetime of 0.9(5) ns. The paper notes a large relative error in that estimate. Ulbricht et al. used femtosecond transient-absorption pump-probe spectroscopy on two samples. Their inferred lifetimes at 78 K were 102.3(3) ps and 92.1(5) ps. These values agree with the computed 117.4 ps and support internal conversion as the main nonradiative decay mechanism.
All phonon modes affect the rate
Section titled “All phonon modes affect the rate”The defect-localized vibrational mode near 170 meV has strong nonadiabatic character. It might therefore appear to control conversion. At 78 K, using only that mode gives 4.37 GHz. Including every phonon mode gives 8.52 GHz. The localized mode therefore accounts for roughly half of the calculated rate.
The accepting-mode approximation gives 2.49 GHz at 300 K. A prior Kohn-Sham orbital calculation gives 0.048 GHz for the same approximation at that temperature. The paper attributes this large difference to the many-body character of the excited states. Single-mode and accepting-mode treatments can underestimate nonradiative rates by orders of magnitude. No single phonon mode dominates every transition studied.
Triplet decay differs between diamond and SiC
Section titled “Triplet decay differs between diamond and SiC”The NV⁻ triplet calculation used a 398-atom supercell, DDH, and an adiabatic energy difference of 2.112 eV. Internal conversion remains below 5 MHz from 0 to 400 K. The radiative rate is 83.3 MHz. The authors therefore find a negligible internal-conversion contribution to the ³E to ³A₂ decay.
The same nominal triplet pathway in the kk-VV⁰ center of 4H-SiC gives a different result. Its 398-atom calculation used DDH and an adiabatic energy difference of 1.351 eV. Including internal conversion brings prior lifetime estimates closer to experiment. Earlier DFT and constrained DFT estimates were 36.8 ns and 28 ns, respectively. Experimental triplet-state lifetimes were about 15 ns. The calculation identifies internal conversion as the missing contribution that improves this agreement.
The ¹A₁ to ¹E transition in kk-VV⁰ remains a prediction. The calculated adiabatic energy difference is 1.132 eV. Rates are of order 10¹ GHz over 0 to 300 K, compared with a radiative rate of 0.32 MHz. No experimental zero-phonon line estimate or singlet lifetime is available for this transition.
Functional choice and model limits
Section titled “Functional choice and model limits”The results depend on the exchange-correlation functional. For the NV⁻ singlet transition, DDH and PBE rates differ by about a factor of two. The main cause is the adiabatic energy difference, 1.397 eV with DDH and 0.871 eV with PBE. Substituting the experimental 1.195 eV zero-phonon energy reduces the rate ratio to about 1.5.
The SiC triplet calculation is more sensitive. PBE places the a1 defect orbital below the valence-band maximum and gives it a delocalized character. DDH places that orbital in the band gap and gives it localized character. The authors attribute this difference to PBE errors in charge-transfer physics and excited-state geometry.
Other approximations limit the calculation. The model assumes displaced harmonic oscillators with no mode mixing. Dynamical matrices were computed at the PBE level to reduce cost. Excited-state phonons of the initial state were used for most transitions. For the SiC triplet transition, ground-state phonons replace excited-state phonons. Calculating the latter would require thousands of TDDFT force calculations for the broken-symmetry ³E geometry.
Where no linewidth data existed, the authors assigned a 0.18 meV smearing factor to the kk-VV⁰ singlet calculation. This choice transfers the NV⁻ singlet value to an unmeasured transition. The paper reports work in progress on machine-learned dynamical matrices and coupling surrogates.
Implications for defect engineering
Section titled “Implications for defect engineering”The results show why excited-state lifetime models for spin defects need state-specific electron-phonon data. In NV⁻ diamond, the singlet pathway is dominated by internal conversion, while the triplet pathway is controlled by radiative decay. Treating both pathways with one generic phonon model would miss that contrast.
The same conclusion applies to defect selection across hosts. A large excited-state energy separation does not by itself determine nonradiative lifetime. Coupling strength, relaxed geometry, state character, and the full phonon spectrum all affect the rate. These quantities matter when estimating temperature dependence, readout efficiency, and ODMR lineshapes.
The authors state that combining internal conversion with first-principles intersystem-crossing rates could support parameter-free predictions of complete optical cycles. This is a proposed use of the framework, not a validated device prediction. For engineers, the practical lesson is narrower. State-resolved lifetime calculations should be checked against all available phonon modes and the measured optical linewidth before use in defect selection.
Materials and support
Section titled “Materials and support”The modeled NV⁻ application requires a diamond host containing a nitrogen-vacancy defect, but the paper does not specify a device geometry or fabrication recipe. 6CCVD offers single-crystal and polycrystalline diamond grown by MPCVD, with plates and wafers up to 125 mm, which can be reviewed in the diamond materials catalog for host-material formats. Engineering support can discuss the defect requirements against available specifications.
View Original Abstract
Optically active spin defects are foundational for quantum technologies, yet common approximations underestimate their internal conversion (IC) rates by orders of magnitude. We propose a broad, predictive framework to compute IC rates that incorporates multiconfigurational effects via many-body wave functions in time-dependent density functional theory (TDDFT) and includes all-phonon-mode contributions via analytical nonadiabatic couplings. Our approach resolves discrepancies with experiment, achieving quantitative agreement for the NV^{-} center in diamond and identifying a previously overlooked nonradiative channel in the divacancy triplet lifetime in SiC.
High Concentration Quantum Grade Single Crystal Diamond
View Specs →Products
View Specs →
Single Crystal Cvd Diamond Plate
View Specs →References we cover
Section titled “References we cover”- First-Principles Framework for the Prediction of Intersystem Crossing Rates in Spin Defects - The Role of Electron Correlation
- Vibrationally resolved optical excitations of the nitrogen-vacancy center in diamond
- Photoluminescence spectra of point defects in semiconductors - Validation of first-principles calculations
- Vibrational and vibronic structure of isolated point defects - The nitrogen-vacancy center in diamond
- Theory of the optical spin-polarization loop of the nitrogen-vacancy center in diamond