Search

In recent years, the rise of machine learning has made it possible to further explore large data in various fields. In order to explore the attributes of loyalty of public transport travelers and divide these people into different clustering clusters, this paper uses K-means clustering algorithm (K-means) to cluster the holding time, recharge amount and swiping frequency of bus travelers. Then we use Kernel Density Estimation Algorithms (KDE) to analyze the density distribution of the data of holding time, recharge amount and swipe frequency, and display the results of the two algorithms in the way of data visualization. Finally, according to the results of data visualization, the loyalty of users is classified, which provides theoretical and data support for public transport companies to determine the development potential of users.

Some spacecraft components with special functions need to be tested the performance of unlocking and deploying in thermal vacuum tests. Such products not only have strict requirements on the mechanical interface of the fixture, but also sometimes need to simulate the temperature boundary of the installation interface. It means that, the temperature of the fixture can be controlled precisely. According to the test requirements of a typical component, the paper proposes a fixture design concept suitable for this type of products and a method is obtained that assess the cooling capacity of the fixture. At the same time, in order to eliminate the impact of thermal radiation from the product on the temperature control of the fixture, measure to enhance the cooling capacity of fixture unit are summarized, which are proved to be effective through simulation verification and guarantee the success of the test.

The hot-carrier lifetime evaluation method of a 90-V-level n-type lateral double-diffused MOS (LDMOS) is investigated in this brief. Our charge pumping experiments demonstrate that the on-resistance degradation of the device is induced by the combination effect of the hot-hole injection at field oxide region and the interface state generation at field oxide and accumulation regions. Thereby, the hot-carrier lifetime of the device was evaluated by the comprehensive model proposed by Moens, which considers the effects of both hot-hole injection and the interface state generation. However, our experiments found that multiple lifetimes were obtained based on the model proposed by Moens and some lifetimes have the risk of overestimation. To solve the problem, an improved hot-carrier lifetime evaluation method based on a novel semiempirical model is proposed. Further experiments and analyses illustrate that the proposed hot-carrier lifetime evaluation method can provide a definite lifetime that has no risk of overestimation for the investigated LDMOS and has improved the lifetime prediction precision compare to the method based on Hu’s model.

In this paper, the hot-carrier-induced current capability degradation of a 600 V lateral insulated gate bipolar transistor (LIGBT) on thick silicon on insulator (SOI) substrate is investigated. Our experiments found that, for the SOI-LIGBT, the worst stress condition is the maximum gate voltage ( V gmax ) condition and the current degradation is dominated by the damages in the channel region under the V gmax stress condition. However, further analyses show that the influence of channel region damages on the collector current degradation increases with the increase of measured collector voltage and is maximum in the current saturation region. Therefore, in our opinion, the hot-carrier-induced current capability degradation of the SOI-LIGBT should be evaluated by the degradation of saturation current under the V gmax stress condition. In addition, a novel SOI-LIGBT structure with an external p-type region was also proposed, which can alleviate the damage in the channel region by reducing the lateral electric field peak. Our experimental results demonstrate that the proposed structure could optimize the hot-carrier reliability effectively with the other characteristics maintained.

Recently, integrated passive devices have become increasingly popular; inductor realization, in particular, offers interesting high performance for RF modules and systems. In this paper, a development of differential inductor fabricated by integrated passive devices technology using a double air-bridge structure is presented. A study of the model development of the differential inductor is first demonstrated. In this model section, a segment box analysis method is applied to provide a clear presentation of the differential inductor. Compared with other work that only shows a brief description of the process, the integrated passive devices process used to fabricate the inductor in this study is elaborated on. Finally, a characterization of differential inductors with different physical layout parameters is illustrated based on inductance and quality factors, which provides a valuable reference for realizing high performance. The proposed work provides a good solution for the design, fabrication and practical application of RF modules and systems.

Seismic data interpolation is an effective way of recovering missing traces and obtaining enough information for subsequent processing. Unlike traditional methods, deep neural network (DNN)-based methods do not need to make assumptions because they can self-learn the relationship between sampled data and complete data using large training data sets and complete the interpolation with a small computational burden. However, current DNN-based approaches only focus on reducing the difference between the recovered and original data during training, which helps to improve the quality of the reconstructed seismic data as a whole, while ignoring the characteristics of the local structure. We have developed a novel seismic U-net interpolator (SUIT) algorithm based on the framework of the U-net DNN in combination with a texture loss, rather than only optimizing for reconstruction loss. Texture loss is proposed to ensure the accuracy of local structural information, which is calculated by a pretrained texture extraction neural network. Furthermore, we use a trade-off parameter to balance the reconstruction error and texture loss, and a practical technique for selecting the associated weighting parameter. The feasibility of our method is assessed via synthetic and field data examples. Numerical tests show that SUIT is robust in noisy environments and that the trained network can reconstruct irregularly or regularly sampled seismic data. Our proposed algorithm performed better than DNN-based approaches that only use reconstruction loss and the traditional low-rank matrix fitting method.

The reduced surface field (RESURF) technique is an effective way to improve the performance of lateral double diffused MOSFET (LDMOS). Thereby, multiple RESURF effect is desired for LDMOS. So far, the LDMOS with triple RESURF effect has been realized and proved to have excellent performance. In this paper, a three-dimensional varying density field plate, which can provide another RESURF effect, is applied on conventional triple RESURF LDMOS. In this way, a LDMOS with quadruple RESURF effect is presented for the first time, which is proved to have 31.2% larger current capability than conventional triple REUSRF LDMOS. Our simulation results show that the performance of the LDMOS with quadruple RESURF effect breaks the silicon limit.

Neuron is the basic structure and functional unit of the brain, its projection and connections with other neurons provide a basic physical infrastructure for neural signal storage, allocation, processing, and integration. Recent technique progresses allow for labeling and imaging specific neuronal populations at single axonal level across a whole mouse brain. However, digital reconstruction of these neuron individuals needs months of human labor or sometimes is even an impossible task. Here we developed a software tool, Advanced NeuroGPS-Tree (AdTree), to solve this problem. AdTree offers a special error screening system to achieve reconstruction close to the round truth. Moreover, AdTree equips with algorithms significantly reduces intense manual interferences and achieves high-level automated reconstruction. We demonstrated using AdTree to get almost complete reconstruction from basal dendrites to far-end axonal terminals on dense datasets that challenge state-of-the-art software tools, and even the manual reconstruction with commercial software, whose accuracy is well acknowledged. We also demonstrated about 5 times speed faster over commercial software in brain-wide reconstruction. In addition, AdTree is applicable on datasets collected with different types of light microscopies such as confocal, two-photon, and light-sheet and serial sectioning wide-field tomography. This tool will provide a starting point to reconstruct neuronal population for cell type, circuit, and neurocomputing studies at single cell level.

We present a highly efficient structural color filtering approach for large-area application, using a nanoporous anodic alumina (NAA) film overlaid with an aluminum (Al) layer on top of an optically thick Al substrate. The NAA film, consisting of a self-assembled nanopore array in a hexagonal lattice, is equivalent to a quasi-homogeneous medium according to effective medium theory. The proposed structure enables strong absorption at resonance owing to the Fabry–Perot resonance supported by the metal-dielectric-metal configuration and the plasmonic effect mediated by the top nanoporous Al layer. The reflection colors can be readily tuned by altering the NAA thickness that is determined by anodization time, thereby allowing the flexible creation of complicated color images on a single platform. By fabricating three samples with different NAA thicknesses in a large area of 2 cm × 2 cm, it is confirmed that the proposed color filtering scheme exhibits highly enhanced color purity and high reflection efficiency of up to 73%, which is superior to that generated by previously reported NAA-based approaches. The presented strategy can pave the way for the efficient fabrication of large-area color filtering devices for various potential applications, including color display devices, imaging sensors, structural color printing, and photovoltaic cells.

In this paper, the influences of sandwiched structure trench on the latch-up reliability of the LIGBT with thick SOI substrate are investigated. Based on the influence mechanisms, a novel device structure with trench grounded is proposed, which can enhance the latch-up voltage by 24% with the other performance maintained. Further studies illustrate that the latch-up voltage would decrease dramatically under high temperature, indicating that the latch-up reliability of LIGBT should be evaluated under high temperature to ensure it operation safely in the practical condition.

Backgroud and Purpose: Extranodal natural killer/T-cell lymphoma (ENKTL) is an uncommon, aggressive form of non-Hodgkin's lymphoma. Optimal therapeutic strategies have not been fully defined yet. We reported the outcome of P-Gemox for the treatment of chemotherapy-naïve, refractory ENKTL in 2013 ASH meeting at the first time. Although several clinical trials have demonstrated the efficacy and tolerability of P-Gemox, we collected and anlyzed the real-world data of P-Gemox in Chinese patients with ENKTL. Patients and methods All cases with ENKTL from 2008 to 2017 were analyzed retrospectively P-Gemox were as follows: gemcitabine 1000 mg/m2; day 1,8; oxaliplatin 130 mg/m2 day 1, pegaspargase 2000-2500 U/m2 im day1. The regimen was repeated every 3 weeks. For newly diagnosed stage I/II patients, P-Gemox regimen was administered as induction chemotherapy for 4-6cycles and followed by involved-field radiotherapy or with"sandwiched" radiotherapy/chemotherapy. Furthermore autologous haematopoietic stem cell transplantation (ASCT) was recommended to chemo-sensitive refractory/relapsed patients as consolidation. All patients come to hospital regularly. Results: 267 patients with ENKTL were treated at the 10 institutions, in the department of Oncology or hematology in China between 2008 and 2017. Patient characteristics are summarized in Table 1. All patients received P-Gemox regimen, 209 patients (78.2%) as first-line therapy and 46 patients (22.0%) as second-line therapy. The median number of treatment cycles was 4.6(1-8) in the first line, 3.3 (2-6) in the second line, 243 patients were evaluable for response. The overall response (OR) rate was 84.8% (206/243), with a complete remission (CR) rate of 59.3% (144/243) (Table2). After the median follow-up time was 43.5 months (range, 1.5-101 months). The median OS of whole group was not reached, the median PFS was 61.0 months. The OS and PFS of the newly diagnosed patients is superior to refractory and relapsed patients (Fig1 A, B, P50%) of rates of grade 1 and 2 adverse events were as follows: hypoproteinemia, 74.1%, neutropenia, 64.4%, fibrinogenopenia 56.2%, transaminase elevated, 52.8%. Grade 3 and 4 adverse reactions (>10%) were thrombocytopenia, 29.2%, neutropenia, 27.7%(Table 3) . Conclusion: P-Gemox regimen followed by EIFRT yielded promising results for patients with stage I/II ENKTL. While for advanced or refractory/relapsed patients, P-Gemox was still unsatisfied . ASCT consolidation for good responders may improve the longterm outcom. Toxicity was moderate and tolerable. Disclosures No relevant conflicts of interest to declare.

With characteristics of high efficiency, low energy consumption and zero emission, pure electric vehicle is an effective solution for the recent energy crisis and environmental issues. This paper develops a pure electric vehicle control strategy that will be used on ENRANGER G3 — a compact SUV model by Wei Chai Motor. A hardware in the loop (HIL) test platform was constructed accordingly. The platform includes those mathematical models of major electrical components and those HIL's hardware boards where the plant model is running. Control model was then tested comprehensively and verified on the real vehicle later. The testing results prove that HIL test can significantly reduce the development time and effort, while improve the quality of the control strategies at the same time.

Specific to the defect that when smith-predictor model mismatch, its control quality gets worse, a steam temperature control system model build on state observer that based on pole assignment was designed. The model makes several constant coefficients first-order damp elements approximate controlled object by using pole assignment, the observer output can feedback in advance to the system by PD controller, to overcome the effect of lagging link and uncertainty of the system, and differential effect can eliminate the dynamic deviation of the system. Meanwhile, simulate the optimization model combined with a second-order controlled object which has an integration element. Through the simulation experiment shows that the system has better dynamic performance and control quality build by this method, provides a way to eliminate the influence of lag link in the steam temperature control system.

Mechanical scribing groove approach has been utilized in the process sequence of high efficiency buried-contact silicon solar cells. The best cell efficiency was 18.47% for float zone wafers in the initial stage. Because of the possibility of using multiple blades at the same time, it has the potential to transfer to mass-scale production of solar cells with low cost and high efficiency.

The formation mechanism of low-spatial-frequency laser-induced periodic surface structure (LSFL) on single-crystalline silicon irradiated by single femtosecond-laser pulse (pulse duration =150 fs and central wavelength =800 nm) in air is investigated theoretically based on the interference theory of Sipe-Drude model and surface plasmon polariton (SPP). In order to account for transient intrapulse changes in the optical properties of the material due to the excitation of a dense electron-hole plasma, we model the maximum of the electron density as a function of laser fluence by solving the generally accepted two-temperature equation and Drude model. The results show that both theories are applicable to explaining the LSFL formation on the high-excited silicon. In the Sipe-Drude theory, the factor (k) is used to describe the efficacy with which the surface roughness at position k leads to inhomogeneous absorption of radiation. We find that the value of (k) first increases until reaching a maximum at an electron density of 61021 cm-3 and then decreases with the laser fluence increasing. When the incident laser fluence is 0.38 J/cm2, which is the threshold for excited plasma, the period reaches a minimum value in a small range of the top. Besides, the law of period is calculated according to the relationship between the (k) and period. In the SPP theory, the ripple period on the highly excited silicon increases with the laser fluence increasing. Comparing the scopes of application of two theories, the Sipe-Drude theory is found to be suitable for the analysis of more extensive periodic surface structures, while the SPP theory is applicable only for the case that laser fluence is close to the damage threshold. Moreover, our results are capable of explaining that the delay direction of periodic ripples are always perpendicular to the incident laser polarization direction by using the Sipe-Drude theory. When laser fluence approaches to the damage threshold, the LIPSS period is calculated sightly to be below the laser wavelength. It also reveals that the periodic surface structures are approximately the same in the different polarization directions with the increase of incident angle. Taking into account a single pulse, the corrugation period decreases with the increase of angle of incidence in the S polarization direction. And under different polarizations, with the increase of incident angle, the changes of the ripple period show an opposite trend. The obtained dependence provides a way to better control the properties of the periodic structures induced by femtosecond laser on the surface of a semiconductor material, which is of great significance for understanding the formation of periodic structure of silicon surface, caused by femtosecond laser, and its application in the field of silicon materials processing.

Multi-businesses and multi-tenants are operating on a uniform cloud network when the concept of the cloud computing is popular today, the bearing network and key technology will be required to realize security isolation in the data from all variety of businesses.

This paper investigates factors related to the distribution of medical insurance coverage in China, using information from an eight-province household survey of almost 16,000 individuals, conducted in 1989. Results of bivariate analyses show that medical insurance coverage, defined very broadly, varies considerably by individual and regional characteristics. Age, gender, education, occupation, employment sector, urbanization, level of industrial and commercial development, and province are all related to being insured or not. In addition, we find that the type of insurance program available to people varies by these same factors, and that the benefits provided by the seemingly uniform public and worker programs also vary, especially by province and degree of urban development. When the individual and regional variables are considered together in logistic regression analyses, the factors most strongly statistically related to the likelihood of being insured are where one works and where one lives. The distribution of insurance benefits in China appears to result in a pattern in which the rural and the poor, who are often at great risk of illness, are less likely to have medical insurance.

The “Housing with Limited Property Rights (HLPR)” has become a very serious problem confronted within Chinese urban development. This paper attempts to unveil the formation mechanism of the HLPR in Nanjing. It first provides the research background and then reviews the most important literature on this issue. Subsequently, this paper puts forward a conceptual framework which is mainly based on theory of space, power and resistance. After briefly introducing the general condition of HLPR in Nanjing, this paper analyzes the formation mechanism of HLPR in Xingwei Village, and the problems it caused. This paper identifies that, it is the practice of state power during the process of rural land expropriation, the extremely tense financial situation confronted with the villagers committee, and its political identity related to the financial condition, which mutually integrate and generate the illegal action of occupying collectively owned land to real estate development.

This paper investigates equity with respect to one component of welfare in China--the provision and use of health services. Based upon a large-scale survey of almost 16,000 individuals in eight provinces in China, we examine a sub-sample of working-age adults who have identified themselves as injured or ill during the four weeks prior to being interviewed. We found that, beyond the level of severity of the reported condition, very few individual-level factors are related to the use of services when ill or injured. Only gender (female) and employment in state-run enterprises are associated with higher patterns of use. These results suggest that China has achieved a very wide distribution of clinics and other services at the local level, and that they are widely used by those who identify need for them. It is rare to be over half an hour away by bike from some form of care and the majority of care appears to be reasonably inexpensive. This broad availability of services contrasts with recent reports from China stressing declining accessibility, and paints a picture of relatively equal access to health care.

The function of excited signal, in the generation, transmission and extraction process of measuring signal in the piezomagnetic type shape meter, was particularly analyzed, and its importance of signal quality to measuring accuracy of stripe shape was showed clearly. Aimming to the various disadvantages of traditional methodes on generating excited signal, a new method based on microcontroller was proposed. Following the method given, a exciting power system for the shape meter appling to the 1250 HC cold rolling mill of ANSHAN IRON AND STEEL GROUP CORPORATION was developmented. Industrial operation shows this exciting power system can provide the current-collecting device of shape meter and signal processing system with high-quality excited signals. The distortion degree of measuring signal is greatly reduced in the process of transmission and extraction, sequentially, the measuring accuracy of stripe shape is significantly improved. This method offers a new way for exciting circuit design, and has much reference to produt design relating to excited signal.

Research and application on shape meter are introduced, especially in China. A self-made MPIBPST(Multi-Roll-Piece Inner Bore Piezomagnetic Sensor Type) shape meter roll for cold-rolled steel strip was applied in to the industrial test. MPIBPST means that one shape meter roll is made up of some roll-piceses, there are two bores in each roll picesce, and each inner bore contains one piezomagnetic sensor. In order to combine with the work of the shape meter roll and display the real-time shape, a shape measurement system based on TMS320F2812 is designed and developed. Measurement principle on shape, hardware and software design are described in detail. The running in the industrial field for about six months shows that shape meter has the advantages of signal stability and credibility, stronger anti-interference ability, high-precision, and meets the measuring requirements on the industrial field. The successful application of the MPIBPST shape meter lays a solid foundation for the further research on AFC (Automatic Flatness Control System) in China. The design on this system also has a reference value for similar work.

The partners of construction project is a new kind of construction project management method, the ideal project partners selected is an important topic that the project owners construction . In this paper, the improved TOPSIS method and the principle of linear programming are introduced to built the model of selecting the partners for the project owners, the comprehensive evaluation systems are designed, then ,according to the above evaluation systems, the improved TOPSIS way was applied to screening the project partners , linear programming and entropy principle are used to optimize the project partners, which are used to build the three stages model selection. The use of that party Contractor selection method can be optimized to achieve the project parties to which the project owners select the partners . The above methods are applied to optimize to select the partners for project owners, the project partners can achieve "win-win" results, the harmonious management of the project can be realized.

The gray fuzzy predictive theory is introduced to engineering costs forecasting in implementation phase, which has been drawn a gray fuzzy predictive model, which is used to estimate the budget costs work scheduled of the unfinished engineering, then the budget costs work scheduled are optimized and adjusted, the buffer management mechanism of engineering in implementation stage is designed ,which is applied to be a timely dynamics warning of engineering costs control. The gray fuzzy predictive model is combined with the buffer management mechanism of engineering costs control ,which can be used to early dynamic warn and control in the whole construction implementation process, engineering costs are ensured to be effectively controlled from start to finish. At last ,the whole control objectives of engineering costs can be achieved .The gray fuzzy predictive model of engineering costs and the buffer management mechanism of engineering costs in implementation stage can provide a new kind of way to forecast the budget costs work scheduled of unfinished engineering of the implementation stage, which can provide an important guiding significance for construction enterprises cost management practice.

The economic relationship among the owners and the contractors of construction project is a kind of principal-agent relationship of strict economic meaning. After analyzing the principle-agent relationship formed in construction project bidding, which produce the problems and the reasons ,the constraints conditions of design on the mechanism of construction project bidding are further analyzed, then ,the incentive mechanism of construction project bidding are planned based on bid price and the type of contract selection .At the same time ,the supervision and restriction mechanism of construction project bidding ,such as the reputation mechanism is designed which encourage the economic objectives of the bidders to be close to the economic targets of the owners so that the probability of the risk of moral hazard and adverse selection are reduced. The system of construction market will gradually be promoted, the harmonious development of construction management will be achieved in the future.

With the implementation of national “Township Electricity Program” and “Sino-German Financial Cooperation Solar Energy Program in Qinghai, over 160 stand-alone PV, wind/PV and PV/diesel hybrid stations have already been constructed in Qinghai province by the beginning of 2007. In order to know operation situation of built PV stations in Qinghai better, Qinghai Brightness Engineering Co., Ltd., cooperated with Deutsche Gesellschaft fuer Technische Zusammenarbeit, selected 12 stations of different types with typical characteristics in different areas of Qinghai province and installed the real time monitoring system to analyze the operation situation of the stations.

To solve the power supply problem in non-electricity areas in our nation, the national government launched the biggest scale electrification program up to now the “Township Electricity Program” in 2001. This program implemented in seven provinces like Tibet, Qinghai, Yunnan, Gansu etc. And over 600 stand-alone PV and wind/PV hybrid stations have been totally constructed with MW total installed capacities. The lead-acid batteries are largely used as the power storage equipments. But with such a great number of lead-acid batteries the recycling will become a great problem after expiration of their performance lifetime. This paper takes the “Township Electricity Program” in Gansu province as an example and puts forward some recycling suggestion for the lead-acid batteries by the results from a survey.

In this paper we use neural network and fuzzy theory for the study on modeling and multilayer intelligent control system Design of urban expressway. First, we develop an extended version of the METANET traffic flow model to describe the evolution of the traffic flows in the urban expressway. Then the simplified model suited for traffic control approach is described by neural network. Second, we propose a multilayer intelligent control structure based on these models and in conjunction with the characteristics of traffic systems, i.e. we divide the control problems of urban expressway traffic system into organization layer, adaptation layer, optimization layer and direct control layer, trying to solve the problem of traditional control systems. Finally we give an example of simulation.

Traditional automatic hardware testing deals mainly with stateless systems. With the rapidly increasing complexity of digital systems, software is more and more engaged in hardware testing. At present, such testing is still mainly carried out manually by engineers, consuming a lot of time and effort. In this paper, we propose an automatic hardware test system that can substantially save engineers' time in testing. It can download the test cases to the target system one by one, receive test output, verify it, and generate log files. The test case generation platform is also discussed, which supports multiple test sets.

During carrying out a project on high efficiency crystalline silicon solar cells, different materials (CZ and FZ wafers with different resistivities) and different structures were experimented upon. For improving solar cell performance, the technology of emitter passivation, AR surface texture and coating, HCl cleaning of the quartz tube, etc., were used. The best efficiency obtained to date is 18.2%.

Cadmium selenide thin film transistors were fabricated using normal evaporation and sputtering technique in multiple pumpdown of vacuum systems. The device structure, materials of each layer and deposition conditions are depicted. Characteristics of CdSe TFTs were studied. The results indicate that stable CdSe TFTs with good characteristics can be obtained by use of simple vacuum deposition processes and adding indium thin layer in the insulator/semiconductor interface.

An AlN photoconductor for X-ray detection has been fabricated, and its response to X-ray irradiation intensity is studied. The photoconductor has a very low leakage current, less than 0.1 nA at an applied voltage of 100 V in the absence of X-ray irradiation. The photocurrent measurement results clearly reveal that the photocurrent is proportional to the square root of the X-ray irradiation intensity, and under relatively high irradiation the photocurrent can reach values one order of magnitude larger than the dark current when a voltage of 100 V is applied across the AlN photoconductor. By using the ABC model the dependence of the photocurrent on the X-ray irradiation intensity is analyzed, and a reasonable interpretation of the physical mechanism is obtained.

A principle for absolute determination of an object’ position by means of LLMC with a CCD camera and the method for finding the error from the instrument and the CCD chip are described in this paper. The observational power and application of the LLMC with a CCD are also discussed.

An X-band four-way combined GaN solid-state power amplifier module is fabricated based on a self-developed AlGaN/GaN HEMT with 2.5-mm gate width technology on SiC substrate. The module consists of an AlGaN/GaN HEMT, Wilkinson power hybrids, a DC-bias circuit and microstrip matching circuits. For the stability of the amplifier module, special RC networks at the input and output, a resistor between the DC power supply and a transistor gate at the input and 3λ/4 Wilkinson power hybrids are used for the cancellation of low frequency self-oscillation and crosstalk of each amplifier. Under Vds = 27 V, Vgs = −4.0 V, CW operating conditions at 8 GHz, the amplifier module exhibits a line gain of 5 dB with a power added efficiency of 17.9%, and an output power of 42.93 dBm; the power gain compression is 2 dB. For a four-way combined solid-state amplifier, the power combining efficiency is 67.5%. It is concluded that the reduction in combining efficiency results from the non-identical GaN HMET, the loss of the hybrid coupler and the circuit fabricating errors of each one-way amplifier.

Nonpolar a-plane [110] GaN has been grown on r-plane [102] sapphire by MOCVD, and investigated by high resolution X-ray diffraction and atomic force microscopy. As opposed to the c-direction, this particular orientation is non-polar, and it avoids polarization charge, the associated screening charge and the consequent band bending. Both low-temperature GaN buffer and high-temperature AlN buffer are used for a-plane GaN growth on r-plane sapphire, and the triangular pits and pleat morphology come forth with different buffers, the possible reasons for which are discussed. The triangular and pleat direction are also investigated. A novel modulate buffer is used for a-plane GaN growth on r-plane sapphire, and with this technique, the crystal quality has been greatly improved.