No description available.

Pipelining is the major organizational technique by which computers can execute several instructions simultaneously to reach higher single-processor performance. Branches are recognized as a major impediment to achieve the maximum performance of pipelining and superscalar processors due to stalls caused by unresolved branches. Branch prediction is an effective strategy to reduce the branch penalty via predicting, prefetching and executing the speculative instructions before the branch is resolved. A branch target buffer (BTB)[13] can reduce the performance caused by branches via predicting the direction of the branch and caching information about the branch. While prediction is incorrect, the processor requires flushing the speculative instructions, undoing the effects of the improperly initiated speculative execution and resuming on the correct path. These flushing and refilling degrade significantly processor performance. In this thesis we propose a mechanism, Double Level Branch Buffer, which can reduce the branch penalty and performance loss caused from incorrect prediction. We try to cache the information of branch about both taken and not taken direction. The pipeline will degrade the dependence upon branch prediction accuracy by utilizing this mechanism.

Existing edge and core routers in DiffServ networks require an effective scheduling mechanism. In this thesis, we design and implement a DiffServ scheduler on Linux platform to provide QoS for different PHB requirements. We first modify the PDD model proposed by Dovrolis, and then develop two new scheduling algorithms. The first algorithm is referred to as Priority Queue with Quantum (PQWQ) and the second one is referred to as Average Delay Queue (ADQ). PQWQ can provide lower delay for EF traffic than Deficit Round Robin (DRR), and higher network utilization than Priority Queue (PQ) with EF Token Bucket. In addition, PQWQ can guarantee a minimum bandwidth for AF and Default PHBs and avoid starvation in case of low priority PHBs. The second scheduler, ADQ, is designed to provide different levels of delay for AF classes. The average delays of the four AF classes can be proportional by adjusting the Delay Differentiation Parameter (DDP). This proportional scheme may allow the higher priority class to send packets more quickly, and therefore achieve higher QoS. Finally, we implement the two schedulers, PQWQ and ADQ, on Linux platform. We adopt share buffer scheme for AF PHB. Share buffer management can effectively improve the buffer utilization and avoid the unnecessary packet dropping due to the unfair buffer allocation. From the experimental results, we can observe that the new DiffServ schedulers not only provide lower delay and higher bandwidth utilization for EF PHB, but also achieve proportional delay among different AF classes.

Veeco Applied EPI 930 molecular beam epitaxy system equipped with a radio frequency plasma assisted nitrogen source has been introduced and the growth procedure and some specialized measurements are also described. The GaN thin films grown by RF-MBE have been talked about nitridation, low temperature GaN buffer layer, and GaN epilayer. The nitridation is a necessary for grown GaN on c-sapphire. From the result of the HRXRD symmetric (002) rocking curve, the magnitude of the FWHM of the GaN films without nitridation was larger than the films with nitridation. The growth temperature of the LT GaN buffer layer was the major factor on the quality of GaN epilayer which grown on the almost without nitridated sapphire substrate. The growth condition of high growth temperature, thin, low growth rate, and low N/Ga ratio of the LT GaN buffer layer can improve the sequent GaN epilayer quality. On the other hand, in the N/Ga flux ratio of GaN epitaxy layer experiment, we have found three interesting results. First, the narrowest peak width of PL spectrum appeared in a slight Ga-rich condition. Second, the smallest of HRXRD FWHM appeared in the nearly stoichiometry condition. Third, the highest electron mobility and less overall dislocations appeared in a slight N-rich condition. Finally, we report the results about AlGaN/GaN heterostructure grown by metalorganic chemical vapor deposition. The piezoelectric effect on the Alx-£_In£_Ga1-xN/GaN heterostructures was investigated and we found that a little In atom in the spacer (£_< 0.01 %) will substantially reduce the strain at interface due to the much larger size of the In atom in comparison to Al and Ga atoms. The electric field at the interface is reduced one order of magnitude smaller than that of the heterostructure without In atom. Two SdH oscillations beat each other due to the population of the lowest two subbands was been seen. Another two SdH oscillations beating have been observed in modulation-doped AlxGa1-xN/GaN heterostructures caused by the finite zero-filed spin splitting due to the inversion-asymmetry-induced bulk k3 term.

With increasing demands on environmental protection in recent years, the Government agency concerned has recently proposed the strategies for shore protection and management, which aim for prevention and mitigation of coastal disaster and reduction in coastal erosion, as well as the creation of an environment with focus on landscape, ecology and community recreation. Soft and quasi-natural approach will be implemented to restore the glory of a stable coast. Based on the consideration of disaster prevention, this study investigates the beach profile changes, which include beach berm erosion and bar formation resulting from storm waves with different return periods. The SBEACH model is used to estimate the beach changes subject to variable conditions of beach berm width, medium sand grain diameter, beach slope and design water level etc. Regression analysis is then applied to establish a relationship between the storm beach buffer width and relevant physical parameters. Prior to this, the results of large wave tank tests on beach profile changes conducted by Coastal Engineering Research Center in the United States are used to calibrate the two main parameters K and £` used in SBEACH model. Beach profile changes can now be estimated systematically using a set of modified K and £` values. After having performed a series numerical studies, we may conclude that: (1) With storms of different return periods but identical non-dimensional fall velocity (H0/£sT), berm erosion increases and the location of the bar becomes further offshore as storm return period increases ; (2) With different sand grain sizes subject to identical storm wave conditions, beach berm erosion increases as grain size increased, but shoreline retreat decreases; and location of bar is further offshore for a beach consisting smaller sand grains; (3) Under the same storm return period and sand grain diameter (i.e., similar non-dimensional fall velocity), berm erosion increases as storm intensity and design water level increase, but shoreline retreat decreases and bar is located nearer; and vice versa; (4) from a series of calculations based on different sand grains and storm beach buffer width, it is found that larger buffer is required for beach with smaller grain size, in order to absorb the storm wave energy.

none

Growing environmental concerns and increasing demand for energy have stimulated extensive interest in electrical energy storage. Li-air batteries are appealing in this regard as they offer much higher energy density than the current Li-ion batteries, but the nonaqueous Li-air batteries suffer from poor cycle life arising from electrolyte decomposition and clogging of the air electrode by insoluble discharge products. Interestingly, hybrid Li-air batteries in which a solid electrolyte separates the lithium-metal anode in an aprotic electrolyte from the air electrode in an aqueous catholyte could overcome these problems. Lots of efforts have been made on developing efficient bifunctional catalysts to lower the overpotential and improve the stability of hybrid Li-air batteries, but the cycle life is still limited. This dissertation focuses on the development of advanced cell configurations and high-performance catalysts for hybrid Li-air batteries. First, a buffer catholyte solution with a moderate pH, based on phosphoric acid and supporting salts, has been developed to keep the solid electrolyte stable and reduce the internal resistance and overpotential. With a high operating voltage and the utilization of all the three protons of phosphoric acid, the buffer catholyte enables a Li-air cell with high energy density. Further increase in power density has been realized by increasing the solid-electrolyte conductivity and operating temperature to 40 °C. The biggest challenge with Li-air cells is the large overpotentials associated with the oxygen reduction reaction (ORR) and oxygen evolution reaction (OER). Noble-metal-free NiCo₂O₄ nanoflakes directly grown onto a nickel foam (NCONF@Ni) has been found to exhibit high OER activity that is comparable to that of the expensive, noble-metal IrO₂ catalyst. Furthermore, a novel 3-D O- and N-doped carbon nanoweb (ON-CNW) has been developed as an inexpensive, metal-free catalyst for ORR. With a hybrid Li-air cell, the ON-CNW exhibits performance close to that of commercial Pt/C. In addition, a novel hybrid Li-air cell configuration with decoupled ORR and OER electrodes has been developed. The hybrid Li-air cell with decoupled ORR and OER electrodes eliminates the degradation of ORR catalysts and carbon support in the highly oxidizing charge process and leads to high efficiency with good cycle life. / text

Achieving the utmost reliability of power semiconductors is an ongoing challenge for the scientists and engineers in the packaging community of the industry and research institutions. Still the semiconductor and therefore the function of the power module could live longer, when only the bonding and joining technologies would be more stable against power and temperature cycling wearout. In particular, the conventional electrical connection to the top and bottom surface of the semiconductor is limiting the lifetime due to degradation. For both, the solder layer under the backside and the Al-wire on the topside of the die, it is necessary to develop new contact technologies, as the substitution of just one connection does not perform the required reliability of the module. In this work, different new technologies for power modules were evaluated and an own development is presented. Especially the new development is characterized by an outstanding reliability while keeping the design flexibility of the currently applied methods. To achieve that, the solder joints were replaced by Ag-sintered connections and Cu-wires were bonded as a substitute of Al-wires. This new approach is called DBB-Technology („Danfoss BondBuffer“) and is demonstrated in the example of a 1700V DBB power module. With the help of this technology sintering creates two joints: One between the bottom of die and the substrate and between the die and a thin Cu-foil, which is located on top. This Cu foil (BondBuffer) enables the Cu-bond process as top contact technology without damaging the semiconductor. The DBB Cu foil acts as an absorber for the high bond-forces. The detailed characterization of a DBB-covered semiconductor module reveals an extraordinary high reliability improvement, enhanced thermal impedance and upgraded electrical properties. / Leistungsmodule unterliegen in der Anwendung vielfältigen, kombinierten Beanspruchungen, die je nach Anwendung eine Herausforderung an unterschiedliche Verbindungsstellen im Modul darstellen. Die Betriebsdauer eines Leistungsmoduls wird im Wesentlichen von den halbleiternahen Aufbau- und Verbindungstechniken limitiert. Das geht z.B. aus umfangreichen Lastwechseluntersuchungen hervor, in denen als Fehlermechanismen die Zerrüttung des Lots unter dem Chip oder das Abheben des Aluminium-Bonddrahts vom Halbleiter identifiziert wurden. Die einzelnen Verbindungsschichten im Leistungsmodul bilden eine Funktionskette, die beim Ausfall nur eines Gliedes die gesamte Funktionalität verliert. Maßnahmen zur Optimierung einzelner Schichten, z. B. der Halbleiter-Substrat-Verbindung oder nur der oberseitigen Chipkontaktierung, bringen alleinstehend also nicht den gewünschten Erfolg. In dieser Arbeit werden unterschiedliche Aufbaukonzepte leistungselektronischer Module aus Fachveröffentlichungen verglichen, bevor das eigene Konzept beschrieben wird. Die Lösung basiert dabei auf innovativen und sehr robusten Fügetechnologien, die gezielt herkömmliche Verbindungen im Aufbau ersetzen. Das Ergebnis ist ein Leistungsmodul mit verbesserten thermischen, elektrischen und thermo-mechanischen Eigenschaften. Eine wesentliche Rolle spielt dabei das Silbersintern als Alternative zum Löten. Dank der Sintertechnik geht der Halbleiter eine hochfeste Verbindung mit dem Substrat ein. Darüber hinaus ermöglicht die Sintertechnologie das stoffschlüssige Verbinden einer dünnen Kupferfolie mit der oberen Halbleitermetallisierung. Diese Kupferfolie ist erforderlich, um das Cu-Drahtbonden für die oberseitige Kontaktierung der Halbleiter anzuwenden, ohne diesen aufgrund hoher Bondkräfte zu zerstören. Dank der vorteilhaften Materialeigenschaften ist die Cu-Bondverbindung deutlich leistungsfähiger als eine Al-Bondverbindung. Diese Kombination aus robusten Fügestellen trägt den Namen DBB-Technologie („Danfoss BondBuffer“) und soll zukünftig dank der Verfügbarkeit sinterbarer Halbleiter in hochzuverlässigen Leistungsmodulen angewendet werden.

No description available.

With the advances in process technology, comes the domination of interconnect in the overall propagation delay in modern VLSI designs. Hence, interconnect synthesis techniques, such as buffer insertion, wire sizing and layer assignment play critical roles in the successful timing closure for EDA tools. In this thesis, while our aim is to satisfy timing constraints, accounting for the overhead caused by these optimization techniques is of another primary concern. We utilized a Lagrangian relaxation method to minimize the usage of buffers and metal resources to meet the timing constraints. Compared with the previous work that extended traditional Van Ginneken’s algorithm, which allows for bumping up the wire from thin to thick given significant delay improvement, our approach achieved around 25% reduction in buffer + wire capacitance under the same timing budget.