Optical gap and optically active intragap defects in cubic BN

We report a comprehensive study on the optical properties of cubic boron nitride ($c$-BN) and its optically active defects. Using electron energy-loss spectroscopy (EELS) within a monochromated scanning transmission electron microscope (STEM) on the highest-quality crystals available, we demonstrate unequivocally that the optical-gap energy of $c$-BN slightly exceeds 10 eV. Further theoretical analysis in the framework of the Bethe-Salpeter equation of many-body perturbation theory supports this result. The spatial localization of defect-related emissions has been investigated using nanometric resolved cathodoluminescence (nano-CL) in a STEM. By high-temperature annealing a $c$-BN powder, we have promoted phase transitions in nanometric domains which have been detected by the appearance of specific hexagonal-phase signatures in both EELS and CL spectra. A high number of intragap optically active centers are known in $c$-BN, but the literature is rather scattered and hence has been summarized here. For several emission lines we have obtained nano-CL maps which show emission spot sizes as small as few tens of nanometers. Finally, by coupling nano-CL to a Hanbury-Brown-Twiss intensity interferometer, we have addressed individual spots in order to identify the possible presence of single-photon sources. The observed CL bunching effect is compatible with a limited set of single-photon emitters and it permits obtaining emission lifetimes of the order of the nanosecond.

Figure 1

EELS low-loss full spectrum of a thin region (about 80 nm thick) of a $c$-BN high-purity crystal. (a) Infrared spectral region showing a prominent peak corresponding to $c$-BN optical phonon modes. (b) Spectral region focusing on the onset of the response function. The weak signal in the lower-energy region corresponds to the signature of Cherenkov effects. (c) Complete loss function presented after removal of multiple scattering.

Figure 2

(a) The loss function, (b) the real and (c) the imaginary parts of the dielectric function calculated from the BSE (violet solid lines) and neglecting the electron-hole interactions in the GW approximation (green dashed lines). In (a) the calculated loss functions are compared with the measured spectrum where the zero-loss peak has been removed (black dashed line).

Figure 3

Band structure of $c$-BN from KS eigenvalues within the LDA (blue lines) and from the GWA in the ${\text{G}}_0{\text{W}}_0$ approach (black lines). The zero of the energy axis has been set to the top-valence maximum.

Figure 4

(a), (b) HAADF images of a micrometric particle from the thermally treated $c$-BN powder. The smaller rectangle in (b) indicates the region investigated by EELS hyperspectral imaging, the larger rectangle the one investigated by CL hyperspectral imaging. (c) EELS spectra extracted from regions 1 and 2 indicated in the intensity map in (d). The color scale in (d) represents the intensity of the 8.2-eV feature. (e) Typical CL signals occurring at the thermal treated $c$-BN powder. CL maps obtained integrating the luminescent signal in the 4.8–6.0 eV energy window.

Figure 5

(a), (d), (g) Examples of typical spectra associated with defects in $c$-BN. (b), (e), (h) HAADF images of the particles from which they have been extracted. (c), (f), (i) Corresponding CL emission intensity maps obtained by integrating the signal in the 10-meV window indicated by yellow regions in the spectra. For the spectra (a) and (g), the integration window correspond to zero-phonon lines. For the spectrum (d) a lower-energy phonon replica has been chosen in order to limit the overlap with the broad band centered at 2.5 eV.

Figure 6

(a) The PF-1 zero-phonon line at 3.57 eV and associated phonon replicas. Dashed curves are individual Gaussian components used to fit the phonon replicas, the red curve is the resulting global fit. In the inset we report a 3.62-eV emission which most likely arises from the same type of defect in a different local environment. (b) HAADF image of the $c$-BN particle from which the spectrum has been extracted and (c) CL intensity map of the this emission. (d) Autocorrelation function derived from HBT interferometry. An emission lifetime of $0.55\pm 0.09$ ns is obtained from the fit of the bunching peak.

Figure 7

(a) The $T$ defect zero-phonon line at 4.94 eV and associated phonon replica. Dashed curves are individual Gaussian components used to fit the phonon replicas, the red curve is the resulting global fit. (b) HAADF image of the $c$-BN particle from which the spectrum has been extracted and (c) CL intensity map of this emission from which several separated spots can be distinguished. (d) Autocorrelation function derived from HBT interferometry. An emission lifetime of $1.4\pm 0.1$ ns is obtained from the fit of the bunching peak.