The results of cross polarized pump-probe experiments in preformed He plasma waveguides are reported. Pump and probe have same wavelength and duration of 800nm and 80fs respectively. Peak pump intensity is I[subscript guided] = 0.2X10¹⁸W/cm² ~1000 I[subscript probe]. Single shot probe spectra and mode profiles at the channel exit are discriminated from the pump with a polarization analyzer and captured at various relative time delays [Delta]t. Frequency-domain interference (FDI) between the probe and a weak depolarized component of the pump is observed for [scientific equation]. Although the depolarized component is nearly undetectable through measurement of pump leakage alone, FDI sensitively reveals its substantially non-Gaussian structure. The possible depolarization mechanisms are analyzed. When probe is positioned at the leading edge of the pump, [scientific equation], its spectrum suffers a blue shift not measurable in the transmitted pump itself. The evidence suggests the channel interior is fully ionized and the partially formed channel ends are the origin of both depolarization and blue shift. A robust, pulsed, differentially-pumped plasma channel generation cell for high intensity guiding experiments has been developed. The design includes an axicon lens, windows for transverse interferometry, and permits injection of one or two different gases (main gas plus high Z seed gas) with several millisecond injection times and simultaneous 0.1ms pressure sensing resolution. Very well formed plasma waveguides have been formed in helium as well as hydrogen, at repeatable and well controlled pressures up to 1000Torr, with very uniform interior density, rapid density drop at boundaries, and very low exterior density. The possible danger associated with the use of large amounts of hydrogen was considered and a complex safety system was designed, constructed and used. Extensive analysis of channel profile reconstruction through transverse interferometry was performed. This includes an intuitive, efficient reformulation and extension of the Phase Locked Loop (PLL) carrier fringe demodulation method. It is also demonstrated and explained how and under which conditions artificial fringe frequency multiplication can reduce demodulation distortions in both PLL and Fast Fourier Transform (FFT) methods. / text

Cette thèse se décompose en deux parties : i) l’étude du creusement d'un canal dans un plasma sous-dense (0.1nc<n<nc, nc étant la densité critique) de plusieurs centaines de microns par une impulsion laser de durée 1-10 ps et d'intensité 10^18 à 10^20 W/cm² ; ii) les mécanismes de saturation de la diffusion Raman arrière stimulée d'une impulsion laser de durée ps et d'intensité 10^14 à 10^16 W/cm². Le creusement d’un canal plasma par un laser très intense a fait l’objet d’une étude paramétrique à l’aide d’un code PIC (Particle In Cell) 2D. On obtient différents types de canaux en fonction des paramètres du laser et du plasma, reproduisant ainsi et élargissant des résultats précédent. De plus, la vitesse de creusement du canal a été mesurée, et des lois d’échelle ont été établies pour les plasmas homogènes. Elles sont ensuite appliquées à des plasmas inhomogènes, du type de ceux rencontrés lors de la fusion par confinement inertiel (FCI). Cela permet de prévoir l’énergie nécessaire pour creuser un canal jusqu’à la densité critique, étape importante de la FCI par allumage rapide. La saturation du Raman a été étudiée d'un point de vue numérique, pour déterminer si la cause de la saturation était due au déphasage ou à la croissance d'une onde satellite (« sideband »), en utilisant diverses approches. La première est de regarder les résultats de simulations Raman (donc électromagnétiques) à partir de codes cinétiques PIC et Vlasov. La deuxième, consiste à regarder ce qui se passe lorsque l'on initialise un plasma avec une fonction de distribution issue de la théorie adiabatique à l'aide d'un code Vlasov (donc dans une version purement électrostatique). Dans ce cas, on observe bien la croissance d'une onde satellite, dont le nombre d'onde dominant, ainsi que le taux de croissance sont en bon accord avec ce que l'on observe dans les simulations cinétiques. Au final, la saturation de l’onde plasma peut être causée par les deux mécanismes de saturation. / This thesis is divided in two parts : i) the laser channeling in hundreds of microns long under-dense plasmas (0.1nc<n<nc, nc being the critical density) of a laser pulse of intensity 10^18 to 10^20 W/cm² and duration 1-10 ps; ii) the saturation mechanisms of stimulated Raman back-scattering of a laser pulse of intensity 10^14 to 10^16 W/cm² and duration of about 1 ps. A parametric study was performed to study the channeling of a very intense laser pulse, using a 2D PIC (Particle In Cell) code. Various kinds of channels were obtained depending on the laser and plasma parameters, thereby reproducing and enlarging previous studies. Moreover, the channeling velocity was measured and scaling laws were established for homogeneous plasmas. They are then applied to inhomogeneous plasmas, similar to those encountered in inertial confinement fusion (ICF). It is then possible to estimate the energy necessary to channel to the critical density, an important step for the fast ignition scheme of ICF. Raman saturation was studied using numerical simulations, in order to determine if it is due to dephasing or to the growth of sidebands, using different approaches. The first is to study Raman simulations (electromagnetic) performed with kinetic PIC and Vlasov codes. The second, is to study the evolution of a plasma initialized with a distribution function after the adiabatic theory, using a Vlasov code (electrostatic). In this case, we observe the growth of a sideband, with dominant wavenumber and growth rate in good agreement with kinetic simulations. The saturation of the plasma wave can be caused by both saturation mechanisms.