WorksPhysics and Astronomy
Variational Calculations of the Excited States of the Charged NV-center in Diamond Using a Hybrid Functional
Abstract
The excited electronic states involved in the optical cycle preparation of a pure spin state of the negatively charged NV-defect in diamond are calculated using the HSE06 hybrid density functional and variational optimization of the orbitals. This includes the energy of the excited triplet as well as the two lowest singlet states with respect to the ground triplet state. In addition to the vertical excitation, the effect of structural relaxation is also estimated using analytical atomic forces. The lowering of the energy in the triplet excited state and the resulting zero-phonon line triplet excitation energy are both within 0.1 eV of the experimental estimates. An analogous relaxation in the lower energy singlet state using spin purified atomic forces is estimated to be 0.06 eV. These results, obtained with a hybrid density functional, improve on previously published results using local and semi-local functionals, which are known to underestimate the band gap. The good agreement with experimental estimates demonstrates how time-independent variational calculations of excited states using density functionals can give accurate results and, thereby, provide a powerful screening tool for identifying other defect systems as candidates for quantum technologies.
Introduction
Spin defects in solid-state materials have attracted significant attention in recent years due to their potential to enable quantum technologies operating near room temperature [1]. Among them, the negatively charged nitrogen-vacancy () center in diamond stands out as one of the most representative deep-level defects, offering unique advantages in quantum sensing, quantum communication, and quantum computing [2–6]. This prominence is largely attributed to its unique spin-optical properties, particularly the ability to initialize and read out its ground-state electron spin using optical fields, coupled with a remarkably long spin coherence time that persists under ambient conditions [7]. The cornerstone of these capabilities is the optical spin polarization cycle. The system, initially in a degenerate spin-triplet ground state , is optically excited to a corresponding triplet excited state . Subsequently, a spin-selective, non-radiative decay pathway allows the system to preferentially relax back into the sub level of the ground state, thereby achieving a high degree of spin polarization [2, 8].
Despite remarkable progress, accurately modeling the properties of such defects remains a critical challenge, particularly their excited states. A variety of advanced computational methods have been employed [9–18] to calculate the electronic states for fixed atomic structure. For instance, time-dependent density functional theory (TD-DFT) has been used to calculate vertical excitation energy values while considering quantum vibronic effects on the electronic energy levels [9]. Other advanced approaches treating electron correlation at a higher level include the GW approximation and the Bethe-Salpeter equation [10], parameterized effective Hubbard models [12, 16] and quantum embedding theories that solve a many-body Hamiltonian for the defect’s active space using highly accurate full configuration interaction quantum Monte Carlo [11, 18].
These efforts have led to a growing consensus on the nature of the excited electronic states and the energy involved in vertical excitations. Another important challenge is the self-consistent treatment of the atomic structure, as the defect relaxes in different ways in the excited states. While the aforementioned high-level methods provide an accurate description of the electronic wavefunctions, their prohibitive computational cost often prevents their direct use in calculations of full structural relaxations. Consequently, most of these studies are on excitation energy for a fixed atomic structure, typically one that is optimized for the electronic ground state using a less computationally demanding method such as density functional theory (DFT) [15] or based on an idealized crystal structure [18]. Such inconsistency between a high-level calculation of the electronic structure for an atomic structure obtained from a lower-level theory makes direct and accurate comparison with experimental values such as the zero phonon line (ZPL) difficult and complicates the validation of different theoretical approaches against one another [19].
A powerful alternative approach are time-independent variational calculations using density functionals. Different from conventional DFT calculations, convergence is reached on excited electronic states as higher-order stationary points on the electronic energy surface, thereby providing solutions to the Kohn-Sham equations corresponding to higher energy than the ground state [20–22]. In particular, the direct orbital optimization combined with the maximum overlap method (DO-MOM) has been shown to be successful [21, 23, 24]. It employs direct optimization to reduce computational cost and improve convergence rate [21, 24, 25]. The maximum overlap method can be used to guide the calculation to a specific target excited state [26], or a projection of the gradient is used to converge on a saddle point of a give order in the generalized mode following (GMF) method [22]. This strategy has proven to be efficient for systems that pose notable challenges to other methods, including molecules with quasi-degenerate orbitals, charge-transfer excitations, and calculations of potential energy surfaces near conical intersections where conventional SCF-based methods [27–29] as well as linear response TD-DFT methods often fail [30–32]. The variational nature of a solution corresponding to a stationary point on the electronic energy surface furthermore ensures compliance with the Hellmann-Feynman theorem, thereby granting access to nuclear forces for excited-state geometry optimizations and dynamics. Recent application of DO-MOM to the NV center using functionals within the local density approximation (LDA), generalized gradient approximation (GGA) and meta-GGA approximation has demonstrated promising results in that the ordering of the energy levels comes out to be right in each case, but the excitation energy has turned out to be strongly dependent on the chosen functional [20]. All these functionals underestimate the band gap of diamond and the energy of states within the gap can, therefore, also be expected to be underestimated. It is clearly desirable to further study this system using a higher-level functional that gives band gap consistent with experiment and, furthermore, use it to assess structural relaxation in the excited electronic states.
In this work, we apply the DO-MOM method in combination with the HSE06 hybrid functional [33, 34], which gives a band gap of diamond in close agreement with experimental data [35], to study the vertical and adiabatic excitations of the NV center. Structural relaxations are calculated at the same level of theory as the estimation of the excitation energy. Notably, we have used spin-purified atomic forces to optimize the structure of the singlet state. The calculated ZPL results and the relaxation energy in the state are found to be in excellent agreement with available experimental estimates. We also analyze the atomic displacements of the and states and compare them with known Jahn-Teller distortions to further validate the qualitative nature of the structural relaxations predicted by the calculations.
Model and Methods
The atomic structure of the NV center consists of a substitutional nitrogen atom adjacent to a vacancy, as shown in Figure 1(a). The system has three-fold symmetry axis and exhibits point group symmetry. Figure 1(b) illustrates the spin orbitals associated with the defect and associated energy levels within the band gap for the ground triplet state, . The bar over an orbital symbol indicates spin-down channel. The excited triplet state, , is formed by exciting an electron from the orbital to one of the degenerate and orbitals, which are illustrated in Figure 1(c). The energy levels and the optical cycle are illustrated in Figure 1(d) where the dashed lines indicate intersystem crossing (ISC) between singlet and triplet states.

Figure 1. (a) Atomic structure of the NV center and the threefold symmetry axis [111]. (b) The energy levels of spin orbitals within the band gap. A bar over the orbital symbol indicates spin down. (c) The orbitals and . (d) Schematic illustration of the optical cycle used for preparing the spin-pure ground state. Measured values of the zero phonon lines (ZPL) of triplets [36] and singlets [37] are indicated. Intersystem crossings (ISC) are shown as dashed lines. (e) The highest occupied orbitals of mixed states and . Right hand side illustrate the occupied spin orbitals of and .
The wave functions of singlet states are more complicated, but can be expressed as
These states cannot be represented by a single Slater determinant [38, 39]. Hence, simply carrying out DFT calculations with same occupation number in both spin channels, the wavefunctions would converge to two symmetry broken mixed states, and , which are linear combinations of pure spin states, , . After spin purification, the energy of the pure singlet states can be expressed as [40, 41]
The force acting on atoms in the spin-purified singlet states can be estimated by taking the partial derivatives of Eq. (3) with respect to the atomic positions. For example, the spin-purified force of the state is expressed in Eq. (5), where $f3f4$ can be calculated using the standard Hellmann–Feynman procedure
In the DO-MOM procedure, the targeted is expressed as a unitary transformation of a set of auxiliary orbitals , thus the total energy is a function of and
The stationary points for can be found by two steps:
Find a stationary point of with respect to . Analytical first-order partial derivative of can be calculated using exponential transformation of the unitary matrix [23]. Limited-memory version of the symmetric rank-one quasi-Newton algorithm is used to optimized the parameters [24].
Minimization of where conventional energy function minimization algorithms generalized for a wave function optimization can be employed. Detailed formalism and implementation of DO-MOM has previously been described [21, 23].
The cubic supercell contains 215 atoms. The optimized lattice constant of pure diamond at this level of theory is 3.546 , giving a box length of 10.639 . Only the point of the Brillouin zone is included. The projector augmented wave method is used in conjunction with a plane wave basis set with an energy cutoff at 600 eV. The atomic coordinates are optimized until the atomic forces are all below 0.01 eV/. To achieve the symmetrized structure relaxation of the state, the occupation numbers of the HOMO and HOMO+1 orbitals are manually set to 1/2. The calculations are carried out using the GPAW [42] software and the Atomic Simulation Environment suite [43].
Results and Discussions
The calculated relative energy of the four low-lying states of the center is illustrated in Figure 2 and the values are listed in Table I. Our results are compared there against other theoretical approaches [11, 15, 20] and reported experimental data [36, 44]. For clarity, we have only included theoretical data from three other works, including full configuration interaction quantum Monte Carlo with quantum embedding in DFT (FCIQMC-in-DFT) [11], beyond random phase approximation with quantum embedding in DFT (beyond-RPA) [15], and variational DFT calculations with the rSCAN exchange correlation functional [20]. These data are good representatives of the current state-of-the-art. The colored region in Figure 2 indicates the experimental estimate of and energy levels [44].

Figure 2. Energy levels of NV center. The solid lines denote vertical excitations and the dashed lines denote relaxed energy levels. The colored region indicate the experimental estimates of and energy levels. Ref. [11], Ref. [15], Ref. [20], Ref. [36, 44]. Ref. [36] reports a vertical excitation energy of 2.180 eV and the ZPL of 1.945 eV. Ref. [44] gives the experimental estimates of the singlet states.
| / eV | / eV | / eV | |
| FCIQMC-in-DFT | 0.583 | 1.573 | 1.98 |
| beyond-RPA | 0.56 | 1.76 | 2.00 |
| rSCAN | 0.62 | 1.80 | 2.06 |
| HSE06 | 0.72 → 0.60 | 2.01 | 2.29 → 2.01 |
| exp. | [0.35, 0.69] | [1.54, 1.88] | 2.180 → 1.945 |
Table I. Various estimates of the energy levels of the NV center, theoretical and experimental. On the left side of the arrows are values of the vertical excitations energy, while on the right side are values of the adiabatic excitation energy. FCIQMC-in-DFT stands for full configuration interaction quantum Monte Carlo with quantum embedding in DFT, beyond-RPA stands for beyond random phase approximation with quantum embedding in DFT, and rSCAN result is from variational DFT calculations with the rSCAN exchange correlation functional [20]. The numbers within brackets give the range in experimentally estimated values for the singlet states. Ref. [11], Ref. [15], Ref. [20], Ref. [36, 44]
For the transition from the ground state to the lowest-lying triplet excited state , a transition that corresponding to the primary optical absorption feature of the center, the HSE06 hybrid functional predicts a vertical excitation energy of 2.29 eV. This value can be compared with both experimental measurements and other theoretical results. Experimentally, the absorption peak associated with this transition is well-established to be 2.180 eV [36]. Among the four theoretical values listed, FCIQMC-in-DFT, beyond-RPA, and rSCAN underestimate the results by 0.20, 0.18 and 0.12 eV, respectively. We note, however, the reasons for these underestimations are different. For FCIQMC-in-DFT and beyond-RPA, while the treatments of electronic correlation are at high levels, their underestimations are most likely attributed to the limited size of the embedding cluster used in the calculations. The SCAN calculations are for a large periodic supercell, so there are negligible finite-size effects, but the meta-GGA nature of the SCAN functional leads to an underestimate of the band gap and thereby likely also the position of the excited energy level in the gap. Our HSE06 calculation addresses these limitations. It improves upon SCAN by reducing the self-interaction error through the introduction of Fock exchange, while also beying applicable to a large periodic supercell.
Following the vertical excitation calculation, a structural relaxation of the excited state was carried out to determine the minimum energy atomic structure. This relaxation process accounts for the atomic rearrangements that occur after electronic excitation, as the excited-state electron density distribution often stabilizes a slightly different atomic configuration. The relaxation lowers the energy to 2.01 eV. This relaxed energy corresponds to the theoretical ZPL energy of the transition, i.e. the energy of the pure electronic transition without coupling to lattice phonons, which is the most reliable benchmark for comparison with the experimental ZPL. This calculated ZPL energy deviates from the experimental value of 1.945 eV [36] by only 3%, a level of agreement that is considered excellent for electronic structure calculations of defect centers in solids. While both of the vertical and adiabatic energy are slightly overestimated in the HSE06 calculations, the relaxation energy of is estimated to be 0.28 eV, which is in good agreement to the experimental estimate of 0.235 eV [36].
We now turn to the singlet excited states, which are essential for an understanding of the full optical processing via inter-system crossing. Experimentally, the energy of the lower energy singlet state, , relative to the ground state has not been precisely determined as a single value but rather estimated to lie within a range of 0.35 to 0.69 eV, arising from the variability of different types of experiments. When evaluating the vertical excitation energy, the HSE06 calculation predicts a value of 0.72 eV for the state, which is slightly above the upper bound of the experimental range. However, after performing structural relaxation of the state using the spin-purified forces, the energy of the state is lowered to 0.60 eV, falling comfortably within the experimentally reported range.
The energy of the higher-lying singlet state, , can be calculated using Eq. (4). The structure is chosen to be the symmetrized geometry of the state, because a limitation arises that the structural relaxation process based on the derived Hellmann-Feynman forces fails to preserve the symmetry of the state. Consequently, this structural relaxation would disrupt the state itself. As shown in Figure 2, our results show that the state lies 2.01 eV above the ground state. This value is slightly higher than the upper bound of the experimentally reported range, but that is not unexpected because of the choice of using the symmetrized geometry of the state in the calculation of the state. There are three reasons for this choice. First, the state is inherently symmetrized, and Eq. (4) is also derived using representations of the group. A symmetry broken atomic configuration could induce significant changes in orbitals and result in convergence problems. Second, the state exhibits minimal distortion from the symmetrized configuration. The root mean square displacement of all atoms is only 0.07 Å, which is relatively small compared to the atomic displacements from to state listed in Table III. The corresponding energy change in state is 30.0 meV, another small value when compared to the state’s relaxation energy of 0.28 eV. Third, in many experimental studies of the optical cycle, the is populated via transitions from the state. Additionally, the transition from to does not involve ISC. As a result, the calculated energy of the can be considered to be in a relatively good agreement with the upper bound of experimentally reported values. Structural relaxation would most likely to reduce its energy and possibly bring it within the experimental range; however, as discussed earlier, further work is required to develop a structural relaxation methodology that maintains the symmetry of the state.
| N | C | C | C | C | C | C | |
| to / Å | 0.066 | 0.024 | 0.024 | 0.026 | 0.062 | 0.076 | 0.076 |
| to / Å | 0.019 | 0.005 | 0.012 | 0.005 | 0.035 | 0.035 | 0.084 |
Table II. Displacements of atoms during energy minimization in the excited states.
To further examine the relaxation effects, Figure 3 illustrates the atomic displacements from the state to the state and from the state to the state, respectively. The diamond lattice exhibits full symmetry, and the same applies to the and states. In contrast, spatial symmetry is spontaneously broken in both the and states. As illustrated in Figure 3 (a-Figure 3b), for the relaxation process from to state, the nitrogen atom is pulled closer to the vacancy, along with three adjacent carbon atoms. Meanwhile, the carbon atoms near the vacancy are pushed outward, moving in the direction opposite to the nitrogen atom. The displacements show strong character, while the anisotropy is rather subtle, indicating this relaxation is dominated by the intrinsic potential energy surface of the state without interacting with Jahn-Teller phonons [45] [19]. Figure 3(c,Figure 3d) show the atomic displacements between and state. The displacement pattern corresponds to

Figure 3. (a-b) Atomic displacements from to shown from two different viewpoints. The opaque and transparent atoms are from and , respectively. (c-d) The atomic displacements from to . The opaque and transparent atoms are from and , respectively. Structural relaxation is highly localized around the NV defect. Apart from the displaced atoms adjacent to the NV center, the atomic displacements of all other carbon atoms in the supercell are approximately one order of magnitude smaller and are omitted from the figure for clarity. The displacements are doubled for clearer visualization.
a typical -mode phonon, indicating that the structure relaxation between and are affected by Jahn-Teller effects more substantially. The analyses of displacements further strengthen our conclusion that the HSE06 calculation with DO-MOM method can correctly optimize the structure of these excited states, paving the way for further theoretical analyses on Jahn-Teller and non-adiabatic effects in the future.
Conclusion
In summary, the HSE06 hybrid functional is used in variational DO-MOM calculations of the excited-states of the center in diamond. Accurate estimates of the excitation energy are obtained for both excited triplet states and the lowest singlet state. The vertical excitation energy from the ground state to the excited triplet state is determined to be 2.29 eV. The relaxation energy is calculated to be 0.28 eV, and the resulting ZPL energy is then 2.01 eV. Both values are in close agreement with experimental measurements, with deviations within 0.1 eV. These values of the excitation energy are higher than those previously obtained using local and semi-local functionals, consistent with the well-known underestimation of band gaps by such functionals, while the HSE06 predicts a band gap that closely matches experimental measurements. By using spin-purified atomic forces, the optimized atomic structure of the singlet state is determined and the relaxation energy is estimated to be 0.06 eV. The energy of the state is then 0.60 eV above the ground state, which aligns well with experimental estimates. The analysis of atomic displacements also demonstrates that the structural relaxations obtained correctly capture the expected Jahn-Teller distortions in both and states, providing useful input and tools for deeper analyses of such effects using complementary theoretical frameworks that rely on the structures obtained with DFT. Our work highlights the potential of time-independent variational DFT methods for future studies of excited states of solid state quantum systems.
Acknowledgments
This work was supported by the National Key R&D Program of China under Grant No. 2021YFA1400500, National Science Foundation of China under grant no. 12334003, and by the Icelandic Research Fund under grant no. 2410644. We thank the High Performance
Computing Platform of Peking University for computational resources.
References
- [35]M. Maciaszek, V. Žalandaukas, R. Silk inis, A. Alkauskas, and L. Razinkovas, The application of the scan density functional to color centers in diamond, The Journal of Chemical Physics 159 (2023).DOI
- [36]G. Davies and M. Hamer, Optical studies of the 1.945 ev vibronic band in diamond, Proceedings of the Royal Society of London. A. Mathematical and Physical Sciences 348, 285 (1976).DOI
- [37]L. Rogers, S. Armstrong, M. Sellars, and N. Manson, Infrared emission of the nv centre in diamond: Zeeman and uniaxial stress studies, New Journal of Physics 10, 103024 (2008).DOI
- [38]A. Lenef and S. Rand, Electronic structure of the n-v center in diamond: Theory, Physical Review B 53, 13441 (1996).pubmed.ncbi.nlm.nih.gov/9983089
- [39]J. Goss, R. Jones, S. Breuer, P. Briddon, and S. Öberg, The twelve-line 1.682 ev luminescence center in diamond and the vacancy-silicon complex, Physical review letters 77, 3041 (1996).pubmed.ncbi.nlm.nih.gov/10062116
- [40]T. Ziegler, A. Rauk, and E. J. Baerends, On the calculation of multiplet energies by the hartree-fock-slater method, Theoretica chimica acta 43, 261 (1977).DOI
- [41]U. Von Barth, Local-density theory of multiplet structure, Physical Review A 20, 1693 (1979).DOI
- [42]J. J. Mortensen, A. H. Larsen, M. Kuisma, A. V. Ivanov, A. Taghizadeh, A. Peterson, A. Haldar, A. O. Dohn, C. Schäfer, E. Ö. Jónsson, et al., Gpaw: An open python package for electronic structure calculations, The Journal of Chemical Physics 160 (2024).
- [43]A. H. Larsen, J. J. Mortensen, J. Blomqvist, I. E. Castelli, R. Christensen, M. Dulak, J. Friis, M. N. Groves, B. Hammer, C. Hargus, et al., The atomic simulation environment—a python library for working with atoms, Journal of Physics: Condensed Matter 29, 273002 (2017).DOI
- [44]S. Wolf, I. Meirzada, G. Haim, and N. Bar-Gill, Nitrogen-vacancy singlet-manifold ionization energy, 19, 034076.
- [45]T. A. Abtew, Y. Sun, B.-C. Shih, P. Dev, S. Zhang, and P. Zhang, Dynamic jahn-teller effect in the nv-center in diamond, Physical review letters 107, 146403 (2011).pubmed.ncbi.nlm.nih.gov/22107219