Nanoscale electron diffraction and plasmon spectroscopy of single- and few-layer boron nitride

Boron nitride (BN) sheets were exfoliated, and proof of the presence of single and double layers was obtained via electron diffraction and plasmon electron energy loss spectroscopy. A plasmon structure unique to mono- and bi-layer BN was established, and was accompanied by WIEN2K DFT calculations. The latter reproduced plasmon energies and general plasmon structure well; however, the detailed shape of the π-plasmon of the calculated pure BN spectra shows discrepancies with the experimental data. The theoretical models were then modified to include impurity atoms. Both oxygen and carbon impurities were considered, as well as different structures, including singular oxygen atoms and oxygen next to carbon atoms. These various configurations were obtained by analyzing atomic-resolution high-angle annular dark field (HAADF) images. Using these modified models, a π-plasmon structure close to the experimentally observed structure could be simulated.

Figure 1

(a) Low-magnification electron micrograph of a single-layer BN flake freely suspended over a circular hole in a TEM grid; (b) part of the flake at larger magnification, showing clean areas—on the order of 100 nm—surrounded by macromolecular deposits; (c) diffraction patterns of the single-layer area (∼0.3 μm in diameter) in panel b at various tilts (the blue diagonal line in the zero-tilt pattern is the line along which the intensity profile in Fig. 2c is taken); (d) diffraction patterns of double-layer area at various tilts.

Figure 2

(a) Full width at half maximum (FWHM) and (b) intensity of the diffraction spots in single- and double-layer BN as a function of tilt angle; (c) intensity profile of diffraction spots in single- and double-layer BN plotted along the line in Fig. 1c.

Figure 3

(a) Experimental, background (ZLP)-removed spectra of BN, with the electron beam penetrating bulk (solid, black), bilayer (dotted, green), and single-layer (dashed, red) BN, as well as aloof (dash-dotted, blue) to single layer perpendicular to the sheets. The 7.5-eV peak is arrowed (gray). (b) WIEN2K calculations for the in-plane plasmon component of bulk (solid, black) and double- (dashed, green) and single- (dashed, red) layer BN; (c) WIEN2K calculations for the same BN configuration as in panel b, but with an added out-of-plane plasmon component. The layer repeat distance for the separated-layer calculations is 60 Å; the number of k-points is 3000. The topmost inset shows an experimental (purple) and calculated (black) plasmon spectrum for single-layer graphene for comparison (Ref. 28). The $y$-axis assigns the intensity of the energy loss function in arbitrary linear units (the units size in panels b and c is the same).

Figure 4

(a) HAADF image of terraces (layer numbers assigned) formed in multilayer BN as the e-beam “burns” material away; image size, 10 nm; (b) double-layer and (c) single-layer (image size, 3 nm) with hole forming during acquisition of an EEL spectrum image. The frame in panel b corresponds to the area of pixels from which the bilayer spectrum in Fig. 3a is taken; the frames in panel c show the areas from which the spectra of the single layer in penetrating and aloof geometry in Fig. 3a are extracted. Panels b and c are HAADF images acquired simultaneously with the spectrum images.

Figure 5

(a) Models for the WIEN2K 2D calculations used to calculate (b) the in-plane and (c) the in-plane plus out-of-plane plasmon component of single-layer BN with various amounts of oxygen. The parameters for the 8-, 16-, and 32-atom supercell calculations are 15, 25, and 35 Å vacuum layer and 1000, 500, and 100 k-points, respectively. The $y$-axis assigns the energy loss intensity in arbitrary, but the same, units in panels b and c.

Figure 6

(a) Top: 1.7 × 1.3-nm${}^2$ HAADF image of single-layer BN after deconvolution with an electron probe function (Ref. 29); bottom: image with the model structure overlaid after calibration of the HAADF contrast (N: gray, B: light blue, O: dark blue, C: yellow); (b)WIEN2K calculations of EEL spectrum according to the model structure in panel a (black solid curve). A 32-atom supercell was used, including a 25-Å vacuum layer, and the k-point number was 50. Overlaid gray curve shows background (ZLP)-extracted, experimental EEL spectra; (c) histograms of B (red/dark gray), C (black), N (green/gray), and O (violet/light gray) atoms obtained from a 5 × 5-nm${}^2$ area.