Herstellung von spannungsoptimierten Silizium-Membranen durch den elektrochemischen Ätzstopp an pn-Übergängen

Vielen Sensoren in mikroelektromechanische Systemen (MEMS) liegen Siliziummembranen zu Grunde. Die Membranen werden mit unterschiedlichen Verfahren hergestellt. Zum Teil kann man Opferschichten benutzen. Will man jedoch Membranen aus kristallinem Silizium, so muß man auf Ätzstopptechniken zurückgreifen. Eine gängige Methode verwendet vergrabene p^(+)-Schichten. Die sehr hohe Dotierung des Siliziums führt jedoch zu unüberwindbaren Problemen bei der Schaltkreisintegration. Die Verwendung von pn-Übergängen beim elektrochemischen Ätzen als Ätzstopp scheint eine aussichtsreiche Lösung. Die Dotierstoffkonzentrationen sind unkritisch, da sie die Elektronik nicht beeinflussen. Die Siliziummembranen lassen sich mit den üblichen Standard-IC-Prozessen herstellen. Ein Ziel der vorliegenden Arbeit bestand darin, die physikalisch-chemische Erklärung des elektrochemischen Ätzstopps von Silizium in alkalischen Lösungen zu finden. Dabei sollten Effekte untersucht werden, die einen Einfluß auf das Ätzstoppverhalten haben, wozu insbesondere - die Verarmungszone unterhalb der p-Silizium/Elektrolyt-Phasengrenzfläche - sowie die Raumladungszone des in Sperrichtung geschalteten pn-Übergangs zählen. Ausgangspunkt sind die chemischen Ätzmechanismen des Siliziums respektive des Siliziumdioxids, wobei der Mechanismus des Siliziumätzens in alkalischen Lösungen und die elektrochemischen Effekte ausführlich beschrieben werden. Es zeigt sich ein starker Einfluß der Oberflächenelektronen im Leitungsband des Siliziums und der Hydroxidionen des Elektrolyten auf den Auflösungsmechanismus des Siliziums. Diese Erkenntnisse ermöglichen es, den mittels pn-Übergangs kontrollierten elektrochemischen Ätzstopp vollständig zu beschreiben. Es zeigte sich, daß die Dicke der Membran von den Dotierstoffkonzentrationen des p-Siliziumsubstrats und der n-Siliziumschicht, der Tiefe der n-Dotierung, der Temperatur im Ätzsystem und besonders von der anodischen Sperrspannung abhängig ist. Zur Verifizierung wurden Siliziummembrandicken mit Hilfe von IR-Ellipsometrie vermessen. Im Rahmen dieser Arbeit wurde auch der es Einflusses von starken Dotierungen im Silizium auf dessen elastische Eigenschaften untersucht, im speziellen bei µm-dünnen Membranen, die mittels elektrochemischen Ätzstopps hergestellt wurden. Dabei wird auf den Einfluß der hohen Dotierstoffkonzentrationen ein besonderes Augenmerk gelegt. Der Youngsche Modul läßt sich bei einem Zugversuch durch Messung der Ausdehnung ermitteln. Für eine zerstörungsfreie Messung der sehr dünnen pn-Siliziummembranen wird die Dehnung berührungslos mit Hilfe des Laser-Speckle Extensometers gemessen. Laserakustische Oberflächenwellen wurden eingesetzt, um gezielt den Einfluß der unterschiedlich starken Dotierung mit Phosphor auf den Elastizitätsmodul zu ermitteln.

3D Simultaneous Inversion for Seismic Structure and Local Hypocenters in Germany under Consideration of Seismic Anisotropy in the Upper Mantel

The main task of this work has been to investigate the effects of anisotropy onto the propagation of seismic waves along the Upper Mantle below Germany and adjacent areas. Refraction- and reflexion seismic experiments proved the existence of Upper Mantle anisotropy and its influence onto the propagation of Pn-waves. By the 3D tomographic investigations that have been done here for the crust and the upper mantle, considering the influence of anisotropy, a gap for the investigations in Europe has been closed. These investigations have been done with the SSH-Inversionprogram of Prof. Dr. M. Koch, which is able to compute simultaneously the seismic structure and hypocenters. For the investigation, a dataset has been available with recordings between the years 1975 to 2003 with a total of 60249 P- and 54212 S-phase records of 10028 seismic events. At the beginning, a precise analysis of the residuals (RES, the difference between calculated and observed arrivaltime) has been done which confirmed the existence of anisotropy for Pn-phases. The recognized sinusoidal distribution has been compensated by an extension of the SSH-program by an ellipse with a slow and rectangular fast axis with azimuth to correct the Pn-velocities. The azimuth of the fast axis has been fixed by the application of the simultaneous inversion at 25° - 27° with a variation of the velocities at +- 2.5 about an average value at 8 km/s. This new value differs from the old one at 35°, recognized in the initial residual analysis. This depends on the new computed hypocenters together with the structure. The application of the elliptical correction has resulted in a better fit of the vertical layered 1D-Model, compared to the results of preceding seismological experiments and 1D and 2D investigations. The optimal result of the 1D-inversion has been used as initial starting model for the 3D-inversions to compute the three dimensional picture of the seismic structure of the Crust and Upper Mantle. The simultaneous inversion has showed an optimization of the relocalization of the hypocenters and the reconstruction of the seismic structure in comparison to the geology and tectonic, as described by other investigations. The investigations for the seismic structure and the relocalization have been confirmed by several different tests. First, synthetic traveltime data are computed with an anisotropic variation and inverted with and without anisotropic correction. Further, tests with randomly disturbed hypocenters and traveltime data have been proceeded to verify the influence of the initial values onto the relocalization accuracy and onto the seismic structure and to test for a further improvement by the application of the anisotropic correction. Finally, the results of the work have been applied onto the Waldkirch earthquake in 2004 to compare the isotropic and the anisotropic relocalization with the initial optimal one to verify whether there is some improvement.

Optimizing Robust Quantum Gates in Open Quantum Systems

We are currently at the cusp of a revolution in quantum technology that relies not just on the passive use of quantum effects, but on their active control. At the forefront of this revolution is the implementation of a quantum computer. Encoding information in quantum states as “qubits” allows to use entanglement and quantum superposition to perform calculations that are infeasible on classical computers. The fundamental challenge in the realization of quantum computers is to avoid decoherence – the loss of quantum properties – due to unwanted interaction with the environment. This thesis addresses the problem of implementing entangling two-qubit quantum gates that are robust with respect to both decoherence and classical noise. It covers three aspects: the use of efficient numerical tools for the simulation and optimal control of open and closed quantum systems, the role of advanced optimization functionals in facilitating robustness, and the application of these techniques to two of the leading implementations of quantum computation, trapped atoms and superconducting circuits. After a review of the theoretical and numerical foundations, the central part of the thesis starts with the idea of using ensemble optimization to achieve robustness with respect to both classical fluctuations in the system parameters, and decoherence. For the example of a controlled phasegate implemented with trapped Rydberg atoms, this approach is demonstrated to yield a gate that is at least one order of magnitude more robust than the best known analytic scheme. Moreover this robustness is maintained even for gate durations significantly shorter than those obtained in the analytic scheme. Superconducting circuits are a particularly promising architecture for the implementation of a quantum computer. Their flexibility is demonstrated by performing optimizations for both diagonal and non-diagonal quantum gates. In order to achieve robustness with respect to decoherence, it is essential to implement quantum gates in the shortest possible amount of time. This may be facilitated by using an optimization functional that targets an arbitrary perfect entangler, based on a geometric theory of two-qubit gates. For the example of superconducting qubits, it is shown that this approach leads to significantly shorter gate durations, higher fidelities, and faster convergence than the optimization towards specific two-qubit gates. Performing optimization in Liouville space in order to properly take into account decoherence poses significant numerical challenges, as the dimension scales quadratically compared to Hilbert space. However, it can be shown that for a unitary target, the optimization only requires propagation of at most three states, instead of a full basis of Liouville space. Both for the example of trapped Rydberg atoms, and for superconducting qubits, the successful optimization of quantum gates is demonstrated, at a significantly reduced numerical cost than was previously thought possible. Together, the results of this thesis point towards a comprehensive framework for the optimization of robust quantum gates, paving the way for the future realization of quantum computers.