Structural and optical properties of silicon nanocrystals embedded in silicon carbide: Comparison of single layers and multilayer structures

Abstract

The outstanding demonstration of quantum confinement in Si nanocrystals (Si NC) in a SiC matrix requires the fabrication of Si NC with a narrow size distribution. It is understood without controversy that this fabrication is a difficult exercise and that a multilayer (ML) structure is suitable for such fabrication only in a narrow parameter range. This parameter range is sought by varying both the stoichiometric SiC barrier thickness and the Si-rich SiC well thickness between 3 and 9 nm and comparing them to single layers (SL). The samples processed for this investigation were deposited by plasma-enhanced chemical vapor deposition (PECVD) and subsequently subjected to thermal annealing at 1000-1100 degrees C for crystal formation. Bulk information about the entire sample area and depth were obtained by structural and optical characterization methods: information about the mean Si NC size was determined from grazing incidence X-ray diffraction (GIXRD) measurements. Fourier-transform infrared spectroscopy (FTIR) was applied to gain insight into the structure of the Si-C network, and spectrophotometry measurements were performed to investigate the absorption coefficient and to estimate the bandgap E-04. All measurements showed that the influence of the ML structure on the Si NC size, on the Si-C network and on the absorption properties is subordinate to the influence of the overall Si content in the samples, which we identified as the key parameter for the structural and optical properties. We attribute this behavior to interdiffusion of the barrier and well layers. Because the produced Si NC are within the target size range of 2-4 nm for all layer thickness variations, we propose to use the Si content to adjust the Si NC size in future experiments

Formation of silicon nanocrystals in silicon carbide using flash lamp annealing

“…The former is crucial for the quantum confinement effect, and the latter causes a high defect density and therefore extremely short lifetimes, making a definitive proof of quantum confinement -to the best of the authors' knowledge -impossible until now. In a prior study, we showed that the Si NC size can be controlled by the Si content of the SixC1-x layers [12]. In the present work, we tried to reduce SiC crystallization by reducing the annealing time.…”

“…The three different layer compositions of x = 0.50, x = 0.63 and x = 0.77 used in this work were previously determined by Rutherford backscattering spectrometry with an accuracy of 1% [17]. The associated gas fluxes can be found in our previous work [12].…”

“…Additionally, the φ angle was varied between 0° and 360°. The GIXRD analysis is described in greater detail in our previous work [12,20].…”

“…Because L( ) represents the crystalline Si-C bonds whereas G( ) corresponds to the amorphous Si-C network, the relation of the peak areas yields an estimate for the Si-C crystalline fraction. This method is widely used in the literature [6,24,25] and was checked for applicability as well as described in detail for our measurements in a previous study [12]. To fit the Si-C mode well, one Lorentzian (L(ν)) and two Gaussian peaks (G1(ν) and G2(ν)) were used.…”

“…To fit the Si-C mode well, one Lorentzian (L(ν)) and two Gaussian peaks (G1(ν) and G2(ν)) were used. The data treatment is exactly the same as that described in our last paper [12] with one exception: Therein we assigned G2(ν) the peak 850 cm -1 to a second amorphous Si-C mode. Further measurements showed an increasing G2(ν) mode with increasing temperature, disproving the a-SiC assignment.…”