La dimensionnalité d’un système affecte fortement ses propriétés physiques ; les transitions de phasequi s’y déroulent ainsi que le type d’ordre qui y apparaît dépendent de la dimension. Dans les systèmesde basse dimension, la cohérence s’avère plus difficile à établir car les fluctuations thermiques etquantiques y jouent un rôle plus important. Le fluide de Bose à deux dimensions est particulièrementintéressant car, même si un ordre total est exclu, un ordre résiduel à « quasi-longue » portée s’établit àbasse température. Deux ingrédients ont un effet significatif sur l’état du système : (i) la taille finie d’unsystème réel permet de retrouver une occupation macroscopique d’un état à une particule ; (ii) les interactionsentre particules conduisent à l’apparition d’un type non-conventionnel de transition de phasevers un état superfluide.Dans cette thèse, nous présentons une étude expérimentale du gaz de Bose bidimensionnel (2D) utilisantdeux types de paysages énergétiques pour piéger nos atomes. Dans la première partie, nous utilisonsla dépendance spatiale de certaines propriétés locales d’un gaz inhomogène pour caractériser l’étatdu système homogène équivalent. Nous extrayons son équation d’état des profils de densité et noustestons son comportement superfluide en mesurant le chauffage induit par le mouvement d’une perturbationlocale. Dans la deuxième partie, nous observons et caractérisons l’émergence d’une cohérencede phase étendue dans un gaz 2D homogène, en particulier via le passage de trois dimensions à deux(croisement dimensionnel). Nous étudions l’établissement dynamique de la cohérence par un passagerapide du croisement dimensionnel et nous observons des défauts topologiques dans l’état superfluidefinal. Nous comparons nos résultats avec les prédictions du mécanisme de Kibble–Zurek. / The dimensionality of a system strongly affects its physical properties; the phase transitions that takeplace and the type of order that arises depend on the dimension. In low dimensional systems phasecoherence proves more difficult to achieve as both thermal and quantum fluctuations play a strongerrole. The two-dimensional Bose fluid is of particular interest as even if full order is precluded, a residual"quasi-long" range order arises at low temperatures. Then two ingredients have a significant effecton the state of the system: (i) the finite size of a real system enables one to recover of a macroscopicoccupation of a single-particle state; (ii) the interactions between particles lead to the emergence of anon-conventional type of phase transition toward a superfluid state.In this thesis, we present an experimental study of the two-dimensional (2D) Bose gas using two differentenergy landscapes to trap our atoms. In the first part, we use the spatial dependence of somelocal properties of an inhomogeneous gas to characterize the state of the equivalent homogeneous system.We extract its equation of state with a high accuracy from the gas density profiles and test itssuperfluid behavior by measuring the heating induced by a moving local perturbation. In the secondpart, we observe and characterize the emergence of an extended phase coherence in a 2D homogeneousgas in particular via a 3D-to-2D dimensional crossover. We investigate the dynamical establishment ofthe coherence via a rapid crossing of the dimensional crossover and observe topological defects in thefinal superfluid state. We compare our findings with the predictions for the Kibble–Zurek mechanism.

The quest to achieve high resolution in ion mobility spectrometry (IMS) has continued to challenge scientist and engineers in the field of separation science. The low resolution presently attainable in IMS has continued to negatively impact its utility and acceptance. Until now, efforts to improve the resolution have mainly focused on better instrumentation and detection methods. However, since the resolution of IMS is diffusion limited, it makes sense to address this limitation in order to attain high resolution. This dissertation presents a new IMS technique, which utilizes a high electric field and opposing high gas flow velocity with the aim to improve resolution. This approach essentially reduces the residence time of ions in the analyzer. This new technique is called "counter-flow ion mobility analysis" (CIMA). Theoretical modeling of this new technique predicted that a resolution of over 1000 is possible, which is over one order of magnitude better than that of conventional IMS techniques currently used. A wind tunnel was designed and constructed to produce a plug gas flow profile that is needed for CIMA. The test region of the wind tunnel was used as the CIMA analyzer region and was constructed from power circuit boards, PCBs, (top and bottom walls) and conductive plastic side walls. An inclined electric field was created by applying suitable voltages to multiple electrode traces on the PCBs. This inclined field, when resolved into its x- and y-components, was used to oppose the counter-gas flow and transport the ions to the detector, respectively. The results obtained did not show an improvement over conventional IMS techniques because of a limitation in the voltage that could be applied to the analyzer region. However, the results predict that high resolution is possible if (1) the ratio of the electric fields in the horizontal (x direction) to the vertical (y direction) is within the range of 2--0.5, (2) very high electric field and high gas flow velocities are applied, and (3) wall effects in the counter-flow gas profile are eliminated. While the resolution obtained using the present instrumentation is far from what was predicted, the foundation for ultimately achieving high resolution has been laid. The use of a wind tunnel has made the instrumentation possible. As far as the author knows, this is the first time a wind tunnel has been used in chemical measurement instrumentation. Chapter 5 of this dissertation, reports a method developed for predicting the reduced mobility constants, of chemical compounds. This method uses a purely statistical regression analysis for a wide range of compounds which is different from similar methods that use a neural network. The calculated value for this method was 87.4% when calculated values were plotted against experimental K0 values, which was close to the value for the neural network method (i.e., 88.7%).