碩士
國立成功大學
生理學研究所
107
Lapatinib (LAP) and sorafenib (SOR) are multi-targeted, small molecules which belong to a family of tyrosine kinase inhibitors (TKIs). They have been recently approved for the ...treatment of a variety of malignant cancers. They are recognized to influence the signaling pathway of tumor cell proliferation, and to inhibit cell surface kinases, thus decreasing the angiogenesis. However, some TKIs have been notably reported to cause a significant prolongation of electrocardiographic QTc interval. The prolongation of QTc intervals may cause the tosade de pointes tachyarrhythmia, which may result in syncope or sudden death. Previous studies have shown that changes in the amplitude and kinetics of the slowly activating delayed-rectifier K+ current IK(S) and erg-mediated K+ current IK(erg) can potentially alter the QT interval duration. Therefore, in this study, we wanted to evaluate the possible effects of LAP or SOR on ionic currents including IK(S), IK(erg), inwardly rectifying K+ current IK(IR), voltage-gated Na+ current (INa) in heart-derived H9C2 cells and in neonatal rat ventricular cardiomyocytes by using the patch-clamp technique. Findings from these results have shown that the presence of LAP or SOR was able to suppress IK(S) amplitude as well as to alter the IK(S) activation and deactivation time courses. The presence of LAP can also shift the activation curve of IK(S) towards more depolarized potentials with no significant change in gating charge of the curve. Therefore, the presence of LAP or SOR can directly suppress the amplitudes of IK(S), IK(erg) , IK(IR) and INa with different potency observed in heart cells. Addition of LAP or SOR can also cause the prolongation of action potential duration. The targets for these ion channels in heart cells tend to be direct and could be of clinical relevance for this group of drugs, if similar findings occur in vivo.
博士
國立成功大學
航空太空工程學系碩博士班
98
The iterative learning control (ILC) theory is proposed for industry robotic systems under repeatable control environment. The traditional ILC is proposed to handle systems ...with uncertainties and repeat disturbances. It improves control system performance by repetition learning in a finite time interval. Theoretical analysis, design and applications have been widely exploited. In recent years, learning controller approaches to cope with the disadvantage of traditional ILC in association with other control method, such as adaptive learning technique, are proposed. In adaptive iterative learning control (AILC) schemes, the discontinuous control technique is utilized to deal with varying disturbances and improve response of the closed-loop control system. Due to actual physical limits, the high-frequency switching control is difficult to be realized and also leads to undesired control results. Based on the design concept of AILC, this dissertation devotes to develop a non-chattering learning control scheme to avoid serious control chattering. A simple control scheme to guarantee stability and robustness of the closed-loop system is proposed. Fast convergence of tracking error is achieved with the proposed adaptive learning law in the iteration domain. To deal the initial resetting condition problem in ILC, the error function with initial rectifying action is utilized for the initial error variations. For robotic manipulator tracking problems, tracking controllers not only compensate dynamic uncertainties in the joint-space but also suppress the influence of both kinematic and dynamic uncertainties in the task-space. The results of this dissertation obtained can be applied to motion control of a wide class of dynamic system with repetitive.
博士
國立成功大學
微電子工程研究所碩博士班
98
III-nitride semiconductors, with wide direct tunable band gap and unique polarization property, demonstrate outstanding potential to significantly advance solid-state ...electronic and optoelectronic technologies. In recent years, various devices have been developed using GaN-based materials grown on sapphire substrates. However, the large mismatches in lattice constant and thermal expansion coefficient limit the development of these devices by generating a significant amount of threading dislocations in a density of 1E9-1E11 cm-2. Dislocations have a deleterious effect on device performance and reliability of parasitic issue.
In this work, GaN-based electronic and optoelectronic devices were grown by metal-organic vapor phase epitaxy(MOVPE). The impact of the dislocation reduction and surface passivation on the performance of heterostructure high electron mobility transistors (HEMTs) and ultraviolet photodetectors (PDs) have been characterized and investigated. Initially, unique techniques to grow multi-MgxNy/GaN buffer or insert a low-temperature Al-containing intermediate layer between high-temperature GaN, as dislocation-reduction structure design in applications to electronic and optoelectronic devices, have been developed. By our methods, the dislocation density in GaN epitaxial layer can be dramatically reduced to the order of 1E8 cm-2 and the device performance can be consequently improved owing to the ameliorative crystalline quality. Additionally, low-dielectric-constant SiO2 film grown by photo-chemical vapor deposition or high-dielectric-constant Ta2O5 film grown by electron beam deposition have been attempted simultaneously as gate insulator and passivation layer for metal-oxide-semiconductor (MOS-)HEMT.
Subsequently, the in-situ grown cap layer, such as un-activated Mg-doped GaN and un-doped AlN, was applied to the fabrication of GaN-based optoelectronics. This unique approach has been proved to provide an effective surface passivation to GaN material. In terms of InGaN material, using triethylgallium (TEGa) instead of trimethylgallium(TMGa) alkyl sources can achieve smoother surface morphology and reduce carbon contamination, but being affected by nitrogen vacancies. Furthermore, TEGa-grown InGaN-based PDs with in-situ passivation have shown that outperformed the regular ones in device characteristics, such as lower dark current, higher detectivity and alleviated photoconductive gain associated with trapping effect. Under a -5 V bias, it was found that the UV to visible rejection ratio was 7.11E3 and 1.38E3 for InGaN-based PDs with un-activated Mg-doped GaN and un-doped AlN layer, respectively. Under the same bias, it was also found that the normalized detectivity (D*) was correspondingly determined as 1.00E13 and 2.56E11 cmHz0.5W-1, respectively.
Next, InGaN/GaN multiple quantum well (MQW) structure has been used in AlGaN/GaN HEMTs and PDs for the first time. The new idea to develop such devices to a reasonable level of optoelectronic integrated circuit application is to be in connection with fundamental light-emitting MQW structure. The use of MQW in the active region of PD offers a flexibility to tune the detection edge and the possibility of increasing photo-response. On the other hand, it results in a very sharp rise of the conduction band after inserting MQW between two AlGaN/GaN hetero-junctions, which is able to obtain a larger main peak transconductance due to better carrier confinement and provides an efficient access to the satellite peak transconductance. It exhibits broader gate voltage swing and higher current-carrying capability as a result of incorporating MQW into multi-channel transistor.
Finally, GaN-based transistors on lattice-matched heterostructure have been accomplished. A promising new approach to improve the performance is to replace the AlGaN with InAlN as barrier layer. It can provide superior polarization-induced charges together with lower strain in InAlN/GaN HEMT, specifically for InAlN lattice matched to GaN, which is achieved by spontaneous polarization only. Consequently, degradation and reliability problems that relate to strain induced defects and strain relaxation could be eliminated. In addition, the effectiveness of SiN surface passivation, with ex-situ grown by plasma-enhanced chemical-vapor deposition or in-situ grown by MOVPE on such devices, has been investigated as well. It was found that the formation of in-situ SiN enhanced the drain current in larger extent while it reveals a better immunity against current collapse by using ex-situ SiN passivation. Therefore, InAlN/AlN/GaN HEMT by dual SiN passivation was introduced to take advantage of combining ex-situ and in-situ SiN passivation and the characteristics of a maximum drain current of 1188 mA/mm and a peak transconductance of 414 mS/mm with a 1 μm gate length were obtained. The current-gain cut-off frequency (fT) and maximum frequency of oscillation (fmax) was 29.4 and 35.3 GHz, respectively. It demonstrates a great potential of our devices to low-noise, high-temperature, high-frequency, and high-power applications.
碩士
國立中興大學
獸醫學系暨研究所
97
Dermatitis is a commonly encountered disease in dogs and its treatment often involves antimicrobial therapy. It is caused by a wide range of factors including bacterial, fungal, ...allergic, parasitic and atopic etiologic agents. Moreover, many studies have reported the frequent isolation of multidrug-resistant strains from canine dermatitis patients and failure in treatment causes problems in small animal practice. Thus, the purpose of this study is to identify the common bacterial and fungal isolates from canine dermatitis patients and to assess the frequency of their resistance to commonly used antibiotics. There were 26 bacterial isolates obtained from various canine dermatitis patients from NCHU Veterinary Teaching Hospital. The isolates were identified using Biolog® microbial ID system, and were subjected to antibiotic sensitivity tests. Among the bacterial isolates S. intermedius B (26.92%) is the most abundant, other isolates were S. aureus ss aureus (23.08%), Pseudomonas aeruginosa (11.54%), other Staphylococci spp. (19.23%), and other bacteria (19.23%). Eighteen out of the 26 isolates showed resistance to at least one or more type of antibiotic. Furthermore, we were able to identify 15 fungal isolates belonging to 11 different species, namely Madurella grisea (2/15), Aspergillus fumigatus (2/15), Aspergillus sp. (2/15), Trichophyton tonsurans (2/15), Microsporum gallinae (1/15), Microsporum nanum (1/15), Penicillium sp. (1/15), Streptomyces sp. (1/15), Malbranchea sp. (1/15), Candida guillermondii (1/15), and Cladosporium sp. (1/15). In this study, we have also shown that the use of universal primers ITS1-ITS4 along with RFLP using HaeIII yielded specific banding patterns and thus could serve as a tool for the differentiation of various fungal species causing dermatitis.
碩士
國立成功大學
資訊工程學系
104
Packet classification plays an important role in router. It provides many service, such as: Virtual Private Network(VPN)、 Quality of Service (QoS) and Fire Wall(FW). However, ...packet classification has some major challenges, such as supporting large rule set, sustaining high performance and so on. Many algorithms were used to solve these problems include decision tree based algorithms, such as HiCuts, HyperCuts and EffiCuts and decomposition based algorithms, such as RFC. These algorithms have good performance on memory access. However, they suffer from duplication rules. Therefore, these algorithms need more memory to store these duplication of rules.
In this thesis, we propose an algorithm, named Maintree with Segmentation Table (MST), which uses the partition of the EffiCuts algorithm to group the ruleset into subsets. And these subsets will be constructed as trees. Then, a tree called main-tree, is chosen form these trees and links to the related nodes in the other trees to avoid traversing these other trees from scratch. We also construct the lookup table in the bucket to speedup the linear search in the buckets. The lookup table is used to decrease number of rules that need to be searched for the input headers. Besides, the rules have different priorities. If the input packet matched a rule, the priority of the matched rule is recorded. When the rules in a bucket are ready to be compared, we check if the priority of the matched rule found previously is already higher than the local maximum priority of the bucket in order to avoid comparing the rules in the bucket entirely. Hence, the priority helps us using less memory access on the searching of the rules. Compared with the well-known method such as EffiCuts, MLT has less memory usage and memory access. The average of memory usage is 31 bytes per rule in MLT and the average of memory usage is 267 bytes per rule in EffiCuts. The average of memory access is 33 per packet in MLT and the average of memory access is 53 per packet in EffiCuts.
We report the results of low temperature resistivity and magnetization measurements on polycrystalline samples of four Ni-Sb compounds, Ni3 Sb, Ni5 Sb2, NiSb, and NiSb2. Resistivity measurements ...revealed that these compounds exhibit a metallic type of electrical conductivity. Temperature dependences of the resistivities were well fitted by the generalized Bloch--Griineisen formula v~ith an exponent of n ~- 3, indicating that the s--d interband scattering is the dominant scatter- ing mechanism. The magnetic susceptibilities of NisSb2, NiSb, and NiSb2 are almost independent of temperature (above 150 K), exhibiting Pauli paramagnetic behavior. The temperature dependences of the susceptibilities were fitted using the Curie-Weiss law. Ni3Sb was found to have a paramagnetic-ferromagnetic phase transition at 229 K. First-principles calculations have been performed to investigate the electronic structures and physical properties of these Ni-Sb alloys. The calculation of the band structure predicted that Ni3Sb, NisSb2, NiSb, and NiSb2 have char- acteristics of metal, and the ground state of Ni3Sb is ferromagnetic. The electrical and magnetic properties observed experimentally are consistent with that predicted by the first-principle electronic structure calculations.
Follicular mucinosis (FM) is characterized histologically by mucinous degeneration of the outer root sheath of the hair follicle and sebaceous gland, accompanied by inflammatory infiltrate. It can ...occur as a primary idiopathic disorder or in association with benign or malignant diseases, most notably mycosis fungoides. In addition, it also can be found incidentally on histology. We describe an unusual case of follicular mucinosis in a 19-year-old man with acute lymphoblastic leukemia (ALL). One month after bone marrow transplantation, he developed cutaneous graft-versus-host disease (GVHD) involving the palms and soles, which was followed 12 days later by the appearance of multiple erythematous follicular papules and plaques on his face, auricles, and postauricular area. Biopsy of follicular plaque revealed changes of follicular mucinosis without evidence of graft-versus-host disease or leukemia cutis. The follicular rash was associated with prominent peripheral eosinophilia. The rash and eosinophilia resolved after 2 and 4 weeks, respectively. In conclusion, we report a case of FM occurring as a transient reaction during the course of cutaneous GVHD following bone marrow transplantation for ALL. Awareness of this condition may avoid undue concern that the rash might represent a manifestation of GVHD, cutaneous relapse of the hematological malignancy, or a drug allergy.
The magnetic properties and magnetocaloric effect(MCE) in EuTi1-xCoxO3(x = 0, 0.025, 0.05, 0.075, 0.1) compounds have been investigated. When the Ti^4+ ions were substituted by Co2+ions, the delicate ...balance was changed between antiferromagnetic(AFM) and ferromagnetic(FM) phases in the EuTiO3 compound. In EuTi1-xCoxO3 system, a giant reversible MCE and large refrigerant capacity(RC) were observed without hysteresis. The values of -△SM^max were evaluated to be around 10 J·kg^-1·K^-1 for EuTi0.95Co0.05O3 under a magnetic field change of 10 kOe. The giant reversible MCE and large RC suggests that EuTi1-xCoxO3 series could be considered as good candidate materials for low-temperature and low-field magnetic refrigerant.