The B and N 1s core-level photoabsorption spectra are shown in Figure 1. Angle-dependent spectra were measured on the iBN thin-film by rotating the samples with respect to the incident x-ray beam linear polarization. The measurement geometry is illustrated in the inset to Figure 2.
Using the analysis methods described by Stšhr and Outka,[4] the hBN, cBN, and iBN B and N 1s (Figure 1(a) and (b)) NEXAFS spectra have been normalized to the background at 185 eV predicated on the fact that the secondary electron yield well below the continuum excitation threshold should be angle- independent. The most striking feature of the B 1s spectra is the well-defined peak near 192.0 eV measured from the hBN sample. Previous work[7,8] has identified the narrow feature at 192.0 eV to be an excitation from the 1s initial state to a quasi-bound pi* final state. Similar features have been seen in the spectra of adsorbed organic species with p-conjugated bonds.[4] The strong 1s-pi* empty-band resonance is absent in the cBN, as anticipated. The broad series of features above 194 eV in the cBN spectrum correspond to transitions to sigma* final states or shape resonances and and are consistent with the expected sp3 bonding of this material. The iBN B 1s spectrum exhibits a sharp peak that can be clearly identified as a pi* resonance. Thus, these measurements provide strong spectroscopic evidence that the iBN thin-film is comprised of sp2- bonded boron and nitrogen. Close examination of the broad pi* peak in the iBN sample shows that it is actually a manifold of four partially resolved components.[9]
The normal incidence N1s spectra for the three BN specimens are displayed in Figure 1(b). The pi* and sigma* portions of the spectra are less distinct than at the B 1s edge, but overall the same qualitative statements hold as for Figure 1(a): the pi* peak dominates the hBN and iBN spectra, but is absent in the N 1s spectrum measured from the cBN sample, which exhibits only the large, broad sigma* absorption.
Earlier work identified the pi* and sigma* resonances of pyrolitic BN and then determined the alignment of crystallites in hBN by examining the angular dependence of the B 1s -pi* features.[7,8] According to dipole selection rules, the matrix element should have a cosine-squared dependence on the angle between the bond and the electric field vector.[4] The B 1s spectra measured from the iBN sample at various incident polar angles q are shown in Figure 2. As can easily be seen, the pi* resonance intensity has a strong angle dependence.
The angle-dependent photoabsorption data of Figure 2 depicts a more intense pi* resonance near normal incidence (q = 0°) than near grazing incidence (q = 90°). This result is surprising because it suggests that the atomic p orbitals that constitute the p bonds and the associated BN c-axis lie preferentially in the plane of the film for this incoherent specimen. In order to quantitatively determine the orientation of the p bonds, integrated intensities were computed from non-linear least squares fits to the B 1s pi* spectra using the convolution of a Lorentzian lineshape with a fixed 65 meV Gaussian for each peak in the case of the NSLS data and 320 meV in case of the SSRL data. The Gaussian broadening reflects the core-hole lifetime, solid-state effects, and instrumental broadening contributions to the line shape. Details will be published elsewhere.[9]
The angular dependence of these integrated intensities is shown in Figure 3, along with two sinusoidal curves for comparison. An ideal fit to the cos2q angular dependence would indicate that the hexagonal planes lie orthogonal to the plane of the film. While this orientation seems unlikely, recently electron- diffraction measurements on hBN films fabricated by ion plating have indicated that the hexagonal planes are oriented near normal to the surface of the film.[10] The cos2(q - 20¡) curve actually fits the data better, as the figure shows. Azimuthal rotation of the iBN film has shown no evidence for in-plane preferential orientation, which eliminates the possibility that that the layer planes are simply tilted 20¡ away from the interface normal. While possible mosaic spread of the crystal can broaden the angular dependence of the intensity, it cannot explain the occurrence of the maximum at q ¹ 0¡. The simplest explanation of the intensity maximum at q » 20¡ is that some part of the film has differently oriented hexagonal planes.