Nanoscale lithography is the key component of the semiconductor device fabrication process. For the sub-10 nm node device, the conventional deep ultraviolet (DUV) photolithography approach is limited by the diffraction nature of light even with the help of double or multiple patterning. The upcoming extreme ultraviolet (EUV) photolithography can overcome this resolution limit by using very short wavelength (13.5nm) light. Because of the prohibitive cost of the tool and the photomask, the EUV lithography is only suitable for high volume manufacturing of high value. Several alternative lithography technologies are proposed to address the cost issue of EUV such as directed self-assembly (DSA), nanoimprint lithography (NIL), scanning probe lithography, maskless plasmonic photolithography, optical maskless lithography, multiple electron-beam lithography, etc.<div><br></div><div>Electron-beam lithography (EBL) utilizes a focused electron beam to write patterns dot by dot on the silicon wafer. The beam size can be sub-nanometers and the resolution is limited by the resist not the beam size. However, the major drawback of EBL is its low throughput. The throughput can be increased by using large current but at the cost of large beam size. This is because the interaction between electrons in the pathway of the electron beam. To address the trade-off between resolution and throughput of EBL, the multiple electron-beam lithography was proposed to use an array of electron-beams. Each beam has a not very large beam current to maintain good resolution but the total current can be very high to improve the throughput. One of the major challenges is how to create a uniform array of electron beamlets with large brightness.<br></div><div><br></div><div>This dissertation shows a novel low-cost high-throughput multiple electron-beam lithography approach that uses plasmonic enhanced photoemission beamlets as the electron beam source. This technology uses a novel device to excite and focus surface electromagnetic and electron waves to generate millions of parallel electron beamlets from photoemission. The device consists of an array of plasmonic lenses which generate electrons and electrostatic micro-lenses which guide the electrons and focus them into beams. Each of the electron beamlets can be independently controlled. During lithography, a fast spatial optical modulator will dynamically project light onto the plasmonic lenses individually to control the switching and brightness of electron beamlets without the need of a complicated beamlet-blanking array and addressable circuits. The incident photons are first converted into surface electromagnetic and electron waves by plasmonic lens and then concentrated into a diffraction-unlimited spot to excite the local electrons above their vacuum levels. Meanwhile, the electrostatic micro-lens will extract the excited electrons to form a finely focused beamlet, which can be rastered across a wafer to perform lithography. The scalable plasmonic enhanced photoemission electron-beam sources are designed and fabricated. An array of micro-scale electrostatic electron lenses are designed and fabricated using typical micro-electro-mechanical system (MEMS) fabrication method. The working distance (WD) defined as the gap from the electron lens to the underneath silicon wafer is regulated using a gap control system. A vacuum system is designed and constructed to host the multiple electron-beam system. Using this demo system, the resolution of the electron beams is confirmed to be better than 30 nm from the lithography results done on poly methyl methacrylate (PMMA) and hydrogen silsesquioxane (HSQ) resists. According to simulation results, the electron beam spot size can be further optimized to be better than 10 nm.<br></div><div><br></div><div>This scheme of high-throughput electron-beam lithography with multiple plasmonic enhanced photoemission beamlets has the potential to be an alternative approach for the sub-10 nm node lithography. Because of its maskless nature, it is cost effective and especially suitable for low volume manufacturing and prototype demonstration.<br></div><div><br></div><div><br></div>

Nous présentons un modèle théorique et numérique pour simuler la diffraction d’ondes électromagnétiques par des lentilles artificielles métalliques. Le premier chapitre présente les radars anti-collision dans le contexte automobile, le système d’antenne est composé d’une source primaire ponctuelle et d’une lentille artificielle. Cette dernière est réalisée de façon très simple en assemblant des lames métalliques minces sur des morceaux de mousse. Une méthode approchée permet d’obtenir rapidement le champ rayonné à travers une lentille par une source ponctuelle à l’aide des concepts d’optique géométrique et d’optique physique. Dans le second chapitre, deux variantes de la méthode modale sont proposées pour l’étude de la diffraction par des réseaux de lames parfaitement conductrices infiniment minces, une dite classique, décrit le champ à l’intérieur des guides parfaitement conducteurs à l’aide des modes de ces derniers, et l’autre considère que les guides forment un milieu inhomogène par morceaux. Les parois des guides sont vues comme des matériaux d’épaisseur très fine et très conducteurs. Numériquement, cet artifice est possible grâce à la technique de résolution spatiale adaptative aussi appelée formulation paramétrique. Dans le chapitre 3, l’ensemble des techniques présentées précédemment est appliqué aux lentilles. Un modèle numérique et électromagnétique est présenté où la lentille métallique est vue comme un empilement de réseaux lamellaires. Le champ global est obtenu en raccordant les modes de chaque couche. Une autre extension qui permet de modéliser des objets non périodiques est introduite : il s’agit d’un changement de coordonnées complexes qui produit des conditions aux limites absorbantes aux bords du domaine de calcul. Dans le chapitre 4, l’ensemble des techniques numériques développées plus haut est mis en oeuvre sur des cas concrets de lentilles artificielles et des comparaisons avec le modèle simplifié du chapitre 1 sont effectuées. Le chapitre 5 est également consacré à l’étude de lentilles. Cependant le domaine de longueur d’onde envisagé n’est plus le même puisqu’on passe dans le domaine optique. La notion de métal perd le sens qu’on lui donne habituellement. Le métal est caractérisé par une permittivité complexe dont la partie réelle peut être négative. Des modes nouveaux apparaissent. La méthode d’analyse retenue est encore une méthode modale. Pour tenir compte des profils d’entrée et de sortie de la lentille, on effectue un changement de coordonnée grâce auquel ces derniers deviennent des surfaces de coordonnées. / We present a theoretical and numerical model to simulate the scattering of electromagnetic waves by artificial metallic lenses. The first chapter introduces the anti-collision radar in the automotive context. The antenna system is composed of a primary point source and an artificial lens. The latter is achieved very simply by assembling thin metal strips on pieces of foam. The field radiated through a lens by a point source can be quickly obtained using the concepts of geometrical optics and optical physics. In the second chapter, two different modal method are proposed for the study of diffraction by arrays of perfectly conducting infinitely thin blades. The first one describes the field inside the perfectly conducting guides by using their exact modes. The second one considers that the guides are piecewise homogeneous media. The walls of the guides are seen as very thin and highly conducting materials. Numerically, this trick is possible thanks to the technique of adaptive spatial resolution. In chapter 3, all the techniques presented above are applied to lenses. A numerical and electromagnetic model is presented where the lens is seen as a stack of strip gratings. The overall field is obtained by matching the modes of each layer. Another extension that allows to model non-periodic objects is introduced : it consists in a complex change of coordinates that produces absorbing boundary conditions at the edges of the computational domain. In Chapter 4, all the numerical techniques developed above are implemented on specific cases of artificial lenses and comparisons with the approximate model of Chapter 1 are performed. Chapter 5 is devoted to the study of lenses in the optical domain. The concept of metal looses its usual meaning. The metal is characterized by a complex permittivity whose real part can be negative. New modes appear. The analytical method is still a modal method.To account for input and out profile of the lens, a change of coordinates is introduced so that the input and output surface of the lens become surface of coordinates.