Doubly charged silicon vacancy center, Si-N complexes, and photochromism in N and Si codoped diamond

Diamond samples containing silicon and nitrogen are shown to be heavily photochromic, with the dominant visible changes due to simultaneous change in total ${\mathrm{SiV}}^{0/-}$ concentration. The photochromism treatment is not capable of creating or destroying SiV defects, and thus we infer the presence of the optically inactive ${\mathrm{SiV}}^{2-}$ . We measure spectroscopic signatures we attribute to substitutional silicon in diamond, and identify a silicon-vacancy complex decorated with a nearest-neighbor nitrogen $\mathrm{SiVN}$, supported by theoretical calculations.

Figure 1

Point defect concentrations measured by EPR and IR in samples grown simultaneously and subsequently annealed for $1\mathrm{h}$ at high temperature under stabilizing pressure (see Table 1). All measurements taken with the sample in the UV-treated charge state (see text for details). Dashed lines are guides to the eye. Note that ${\mathrm{N}}_{\mathrm{s}}^{0/+}$ concentrations have been divided by 10 to increase legibility.

Figure 2

(a) IR absorption spectra of sample A (top) and sample F (bottom) in the two extremal charge states. The two regions give the defect-induced one-phonon absorption (left) and C-H stretch region (right): the intrinsic multiphonon absorption has been subtracted. (b) The one phonon of two as-grown samples grown under similar CVD conditions: both are nitrogen doped but silicon was added to the growth gasses of one. The primary difference is the feature at approximately $1340{\mathrm{cm}}^{-1}$ in the silicon-containing sample, which is tentatively assigned to substitutional silicon [22]. The remaining peaks are reported in studies of solely nitrogen-doped CVD material [24]. (c) Difference spectra between sample A (as-grown) and samples treated at the given temperatures. The change in the $1338{\mathrm{cm}}^{-1}$ mode is highlighted.

Figure 3

UV-vis absorption spectra of sample D (annealed at $2000{}^{\circ }\mathrm{C}$) measured at 80 K in two extremal charge states—see text for details. Strong ${\mathrm{SiV}}^{-}$ and ${\mathrm{SiV}}^0$ spectra are recorded in the UV-treated charge state, and are undetectable in the heated charge state. The charge transfer processes are reversible, i.e., no net $\mathrm{SiV}$ is created or destroyed in during the treatments: the dramatic loss of ${\mathrm{SiV}}^{-}$ and ${\mathrm{SiV}}^0$ between extremal states is therefore strong evidence for the existence of a third charge state of $\mathrm{SiV}$, which we identify as ${\mathrm{SiV}}^{2-}$ . Note that a broad ${\mathrm{SiV}}^0$ absorption continues beneath the ${\mathrm{SiV}}^{-}$ absorption, but the structure between $500–740\mathrm{nm}$ belongs to ${\mathrm{SiV}}^{-}$ . Inset: Representative transmission images of the sample in the two extremal charge states.

Figure 4

EPR spectra of (a) sample F (${}^{15}\mathrm{N}$ doped) and (b) sample G (${}^{14}\mathrm{N}$ doped) with $B\parallel \langle 111\rangle$. Experiment in black; simulation in red. Additional resonances in (a) are due to ${}^{15}{\mathrm{N}}_2{\mathrm{VH}}^0$. Arrows highlight resonances of the three inequivalent orientations of a $C_{1h}$ defect (relative populations 1:1:2), each of which is split by interaction with an $I=1/2$ ${}^{15}\mathrm{N}$ nucleus. (c) Angular variation ({1 1 0} plane) of measured EPR transition fields in sample G (circles) overlaid with a spin Hamiltonian simulation (Table 3). Transitions of a single ${\mathrm{SiVN}}^0$ orientation are highlighted to demonstrate the effect of the similar magnitude quadrupole and hyperfine interactions ($\mathrm{\Delta }{m}_S=\pm 1$; $\mathrm{\Delta }{m}_I=\pm 1$ transitions gain appreciable intensity) resulting in six transitions per orientation rather than the expected three.

Figure 5

EPR spectrum of sample G with $B\parallel \langle 111\rangle$. Experimental data in black; simulation in red. Additional panels show the ${}^{29}\mathrm{Si}$ hyperfines on each side of the primary spectrum. As expected, their intensity is $5\%$ of the primary spectrum. Simulation generated by EasySpin [71] using the spin Hamiltonian parameters given in Table 3.

Figure 6

(a) Schematic of the $\mathrm{SiVN}{}^{0/-}$ defect, highlighting the defect's {1 1 0} mirror plane and the ⟨3 1 3⟩ direction between the nitrogen and silicon atoms. (b) Defect-free region of the diamond lattice for comparison.

Figure 7

(a) Schematic of the $\mathrm{SiVN}{}^{0/-}$ defect, highlighting the defect's ⟨1 1 0⟩ mirror plane. (b) Calculated formation energies at varying chemical potentials ${\mu }_e$ for $\mathrm{SiV}$ and $\mathrm{SiVN}$ with reference to the intrinsic diamond valence band maximum. Transition levels include charge density offset and Madelung corrections. The calculated conduction band minimum was at $4.27\mathrm{eV}$. (c)–(e) Electron density on band gap states for (c) the donor state in ${\mathrm{N}}_{\mathrm{s}}^0$; (d) the state with an unpaired electron for ${\mathrm{SiVN}}^{2-}$; and (e) the same state for ${\mathrm{SiVN}}^0$ . Comparison of (d) and (e) highlights the additional electron in an N-C antibonding orbital in ${\mathrm{SiVN}}^{2-}$ . Isosurfaces depict a surface of constant spin density: a common spin density threshold was chosen for (d) and (e) to allow comparison in the same structure; a higher threshold was chosen for (c) due to the highly localized nature of the ${\mathrm{N}}_{\mathrm{s}}^0$ donor state.