This research theme was launched in the team some fifteen years ago. The aim was to study quantum confinement effects in Si or Ge confined nanostructures. Because of their indirect gap, these semiconductors are poor optical emitters. The reduction of the crystal size allows to relax the selection rules on the of wave vector. The quantum confinement of the charge carriers thus increases the probability of radiative transitions. Si nanocrystals (Si-nc) have been widely studied in the scientific community for nearly 15 years and are still the object of research because of their potential as optical sources. Between 2007 and 2010, the team contributed to the study of Si-nc confinement properties in multilayer systems, which allowed the state of the art to be reached in terms of Si-nc development. The control of this model system sets apart the work of the team in the field and has allowed the properties of both electronic and vibrational confinement to be studied.
To control Si-nc size, we inserted SiO layers between SiO2 barrier layers by preparing SiO/SiO2 multilayers by successive evaporation of SiO and SiO2. The relevance of the SiO2 layer is to limit the growth of silicon grains in the SiO layer. Annealing at 1100°C thus allowed the crystallization of silicon aggregates arising from the dissociation of SiO into Si and SiO2.
The films show intense photoluminescence intensity (visible to the naked eyes) in the visible part of the spectrum. We first demonstrated that photon emission is due to carrier confinement in nanometre-sized crystals. As predicted by the theory of quantum confinement, the PL energy is a decreasing function of the Si-nc size. The study of the dependency of the luminescence with temperature allowed us to explain the luminescence behaviour at low temperature. The variation of the photoluminescence intensity with temperature is unusual since it shows a maximum around 70 K. The increase in PL intensity when the temperature drops comes from the reduction of the role of thermally activated non-radiative centres. In contrast, the reduced PL intensity at the lowest temperatures is barely touched upon in the literature. We showed that this decrease is linked to an increased radiative lifetime at low temperature, which diminishes the luminescence yield. Furthermore, we demonstrated that the measurements of PL energy as a function of temperature gave a signature characteristic of bulk Si. The energy follows the Varshni law that is characteristic of crystalline silicon, on the condition that the energy linked to the confinement is added to the model. Similarly to intensity, energy also undergoes a specific evolution below 40 K. The energy dependency no longer follows Varshni's law and becomes higher than predicted. We explained this trend by a luminescence saturation phenomenon at low temperature, which is all the more important when the Si-nc are large. As the films show a size distribution where the width increases as a function of the Si-nc size, the luminescence at low temperature comes mainly from small nanocrystals. This explains the "abnormal" increase of PL energy at low temperature.
This study aimed to control the spontaneous light emission that emanates from Si-nc, by modifying the spatial distribution of the emission. Such a study has fundamental objectives, such as the understanding of the mechanisms of light-matter coupling, or the development of nano-emitters. It also enables the development of more concrete applications, including the optical injection of light into optical fibres. This point is all the more important in that it concerns silicon emitters developed with the help of CMOS technology.
Optical micro-cavities have been developed, where the active layer is made of Si-nc obtained by the preparation of an SiO/SiO2 multilayer. These Si-nc were inserted between two Bragg mirrors, developed with the help of Si/SiO2 doublets. The optical micro-cavity was fabricated monolithically during a unique evaporation process, followed by thermal annealing at 1100°C. The study of the influence of the active layer thickness was carried out using SiO/SiO2 multilayers that contained 18 to 21 doublets. An evolution of the micro-cavity transmission peak of the order of 4.5 nm per nm deposited was observed. Such an influence of the cavity thickness must thus be taken into account to centre the micro-cavity on the Si-nc emission peak.
With annealing, the thin layers obtained underwent modifications in their thickness and their optical refraction index due to a densification process. Mastery of these two parameters is crucial for the control of the forbidden optical band of Bragg's mirror. A systematic study of the optical properties of the different layers depending on the annealing temperature was carried out, using spectrometric ellipsometry, transmission spectroscopy in the UV-Visible part of the spectrum and X-ray reflectometry.
The micro-cavities were also analyzed by simulation. Using experimentally determined values of thickness, refraction index and absorption, the simulated spectra closely follow the experimental ones in terms of evolution of the micro-cavity resonance peak as a function of the annealing temperature. The shift of the resonance peak towards the blue, linked to thermal annealing of the micro-cavity, is of the order of a hundred nanometres. This is explained by densification of the layers because of annealing.
The figure above shows the photoluminescence of two multilayer reference samples and of a multilayer inserted in a micro-cavity. The first sample is a Si-nc multilayer. The second is identical except that its upper part is a silicon layer of a thickness equivalent to that contained in a Bragg's mirror above the micro-cavity. This allows absorption by the Si layer to be taken into account. Coupling between the cavity and the light emitters is clearly visible. Luminescence derived from the micro-cavity is thinner and more intense than that derived from Si-nc. This is particularly noticeable when the comparison is carried out using the Si-nc sample that has an upper layer supplemented with silicon. The selectivity angle is also clearly visible, with luminous emission lobes that are well-separated spatially. To confirm the weak Purcell type of linkage in this sort of system, analyses by time-resolved photoluminescence were carried out. The theory of weak linkage predicts a lowered time of decline of the photoluminescence in the case of emitters coupled to a micro-cavity. This was seen in our experiments of time-resolved photoluminescence.