This thesis work brings a contribution to the topic of integration of active materials in silicon photonics for the realization of on chip light amplification and emission. The emphasis has been placed on materials prepared in thin layers that can be deposited on silicon substrates to produce light sources or amplifiers in the datacom&telecom wavelength windows (~1.3μm&1.55μm). The approach adopted favored the use of slot waveguides because of the enhanced overlap between the guided optical modes and the covering cladding materials: - Si/SiO2 and SiN/SiO2 slot waveguides and ring resonators based on these waveguides have led to propagation losses typically of the order of 1-4dB/cm and resonators quality factor of several tens of thousands for photonic structures covered by cladding active materials of refractive index 1,5. - Work carried out on the integration of active materials doped with Erbium was performed in the framework of two international collaborations (China and Finland). The first collaboration led to the demonstration of optical gain from an inverted rib waveguide geometry. An internal gain of the order of 25 dB was obtained by this approach for an optical pump power of 80 mW. A second collaboration focused on the integration of Al2O3 oxide in SiN slot waveguides fabricated within C2N. The questions related to integration of materials were studied initially. The main obtained result was the observation of a max net gain of 25 dB/cm in short slot waveguides for pump powers of the order of 50 mW at wavelength 1480 nm. - We explored a second way for demonstrating emitter / amplifier structures on a chip, exploiting the use of semiconductor carbon nanotubes (SCNTs). Our C2N team, in collaboration with CEA-Saclay, has developed a method for the preparation of thin layers rich in SCNTs that can be used as an active medium in a planar integration. Using this approach, we demonstrated that vertical pumping of SCNTs hybrid photonic structures could result in very strong on chip coupled photoluminescence (PL) (105/s), and that a significant increase of SCNTs emission in cavities (> 100) was obtained by the Purcell effect in air mode nanobeam cavities, associated with a spectral narrowing of the observed resonances as a function of the pump power. The work presented in this thesis contributes to the development of hybrid photonics on silicon exploiting SOI platform waveguiding properties and those of active materials (here, polymers doped with Erbium or carbon nanotubes).

Les travaux présentés dans ce manuscrit contribuent au développement d’une méthode de caractérisation diélectrique sans contact de tissus biologiques, au moyen de sondes inductives fonctionnant dans le domaine des radiofréquences (RF). Ces travaux s’inscrivent dans le contexte général du développement de méthodes de détection des tissus cancéreux à bas coût, ayant pour but de faciliter la mise en œuvre d’une politique de prévention massive. La méthode de caractérisation s’appuie sur la mesure des propriétés diélectriques (conductivité électrique et permittivité diélectrique) du tissu, qui sont fortement affectées par les modifications structurelles des tissus lors de l’apparition de pathologies cancéreuses. S’appuyant sur des travaux expérimentaux préalables qui ont montré expérimentalement la sensibilité de sondes RF sans contact à la permittivité complexe de milieux organiques, les travaux présentés dans cette thèse s’attachent à explorer de manière plus formelle la pertinence de telles méthodes pour la caractérisation de tissus cancéreux. Pour ce faire, nous avons étudié l’implantation d’une méthode semi-analytique DPSM pour modéliser les interactions intervenant en champ proche et à échelle mésoscopique entre une antenne inductive et le milieu à caractériser, et déterminé les paramètres permettant d’ajuster la sensibilité de l’antenne aux paramètres diélectriques du milieu. Ensuite, nous avons exploité une méthodologie d’estimation de la distribution spatiale des propriétés diélectriques du milieu à l’aide d’une approche à réseaux de neurones artificiels, permettant de détecter, localiser et estimer les propriétés d’une inclusion « tumorale » enfouie dans un « tissus sain », avec des erreurs d’estimation inférieures à 10% pour les configurations étudiées. La méthodologie développée doit ouvrir à la voie au développement de systèmes multicapteurs RF sans contact pertinents pour la détection à bas coût de lésions tumorales dans les tissus

With the rise of nanoelectronics, many novel technologies have emerged, holding the promise to replace or complement the traditional computing building block – the CMOS transistor. However, at the nanoscale, noise significantly affects the behavior of systems, inducing random fluctuations. It is thus natural to look for computing techniques which are intrinsically tolerant to noise, variability and errors, or even better, which take advantage of these. Among the possible solutions, one paradigm has emerged as particularly promising and disruptive: taking inspiration from biology. Indeed, our brain is able to perform computations – while consuming only 20 W – even though its components themselves exhibit stochastic behavior. Bio-inspired computing with stochastic nanodevices should prove to be particularly successful for cognitive tasks such as pattern recognition and classification. Mixing conventional electronic components with emerging technologies could allow performing such tasks at low energy cost. The focus of this thesis is a specific nanodevice, the magnetic tunnel junction. Because of its endurance, reliability and CMOS compatibility, this bistable system has emerged as the flagship device of spintronics. However, maintaining the stability of this device while reducing its size is a challenge. Unstable magnetic tunnel junctions – called superparamagnetic tunnel junctions – behave as stochastic oscillators. In this thesis, I investigated for the first time how to harness the random behavior of stochastic magnetic tunnel junctions, taking inspiration from biology. First, it is experimentally demonstrated that electrical noise can induce the synchronization of a junction to a weak voltage source. A theoretical model is developed and predicts that using noise could allow a hundred-fold energy gain over the synchronization of traditional dc-driven spin torque oscillators. This result opens the way to the low power hardware implementation of synchronization-based computing schemes which can perform tasks such as pattern recognition. Then, an analogy between superparamagnetic tunnel junctions and sensory neurons – which fire voltage pulses with random time intervals – is drawn. Pushing this analogy, it is numerically demonstrated that interconnected populations can perform computing tasks such as learning, coordinate transformations and sensory fusion. Such a system is realistically implementable and could allow for intelligent sensory processing at low energy cost. All these results suggest that the superparamagnetic tunnel junction is a promising building block for hardware implementations of bio-inspired computing.