Electron energy-loss spectra calculations and experiments as a tool for the identification of a lamellar C3N4 compound

In order to identify a poorly crystallized lamellar C3N4 compound using electron energy-loss spectroscopy, we first concentrate on model compounds, namely, graphite at the carbon K edge and hexagonal boron nitride at both K edges. We show that extremely good agreement can be obtained between experimental spectra and ab initio calculations using the 5WIEN2k program [P. Blaha, K. Schwarz, P. Sorantin, and S. B. Trickey, Comput. Phys. Commun. 59, 399 (1990)]. Because the calculations are based on the density functional theory considering independent quasiparticles, they are generally believed to be inadequate to rigorously describe the excitation process. Nevertheless, we demonstrate that the spectra, when divided into two regions, above and below the ionization energy, can be simulated over a 50 eV energy range. In these materials, the first region must be simulated taking into account the electronic relaxation around a full core hole. For the second part starting usually 10-12 eV above the threshold, only a very small or no core hole at all should be introduced for the relaxation. Furthermore, we show that an empirical formula is sufficient but necessary to take into account the lifetime of the electron in the excited state. After discussion of these results on model compounds, we clearly identify the limits and benefits of these comparisons and apply them to an unknown structure. The experimental spectra were recorded at low temperature in order to avoid beam damages. They are compared with those calculated for various structures having compositions close to C3N4. From these core-loss studies, as well as from the corresponding low-loss ones, the structure of our sample is shown to be locally very close to that of the (C3N3)(2)(NH)(3) lamellar compound described by Kawaguchi and Noziak [Chem. Mater.7, 257 (1995)]. This structure can also be viewed as a defective hexagonal C3N4 lamellar structure (with ABA stacking), rendering this compound most suitable for a further high-pressure conversion to a superhard 3D C3N4 phase.