The surface properties after plasma etching of TiW alloy using the chemistries of BCl$_{3}$ and SF$_{6}$ gases with varying mixing ratio have been investigated using XPS(X-ray photoelectron spectrocopy). The elements existed on the etched sampled have been extracted with BCL$_{3}$/SF$_{6}$ ratio and their chemical binding states have also been analysed. It was confirmed that the thickness of native oxide formed on the TiW films is thinner than 10nm by using Ar sputtering. At the same time, the roughness of etched surface has been esamnied using AFM (atomic force microscopy). on the basis of the basis of this results, the relations between the caanges of oxygen contents detected by XPS and the rouhness of etched surface have been discussed. And the etch rate and etched profile of Tiw films have been examined and the changes of the etch rate and etched prfile have been discussed with XPS results. From XPS results, the role of passivation layer consisted of Ti-S compound with XPS results. From XPS results, the role of passivation layer consisted of Ti-S compound has been proposed. Ti-S compound seems to make a role of passivation layer that surpresses Ti-O formation.ion.

A 23GHz 1-stage LNA was designed by using new topology of coupled line type with EEsof softwares and modified by using MPIE numerical analysis. The parallel coupled filter-type matching sections give impedance matching and DC blocking simultaneously, and have small discontinuities. This matching scheme has simple structure in the design process and give small error. The EFT chip was directly attached to the ground metal. The designed LAN gives 6.2dB gain and 2.5dB noise figure without considering the loss of connectors. Through these results, it was verified that our design process, matching schemes and fabrication technologies was valied for developing 20GHz-band LNA.

Broad-band RAM's (Radar absorbing materials) are designed by multi-layered lossy dielectrics. The depth, the relative permittivity and the loss tangent of each layer are optimized in order to meet the required reflective power over the specified frequency range using a genetic algorithm. The reflection coefficients are calculated by the continued fraction method. A new population model of the partial initialization method during iterations is applied for the multi-modal functions to enhance the performance of the genetic algorithm. The optimal RAN's are designed by setting the relative permittivity and the loss tangent of the dielectrics as a funtion of the frequency over 5~20GHz.

In the design of microwave multi-stage amplifiers composed of N amplifier modules, the variation of performances of amplifier as various connection length between modules has been studied. In addition, the methods, equations and conditions for the maximum gain or the most flat gain are presented. The results of sensitivity analysis for the connection length showed that the small change in phase of input, output reflection coefficients(S$_{11}$, S$_{22}$) of module itself is the most important in determination of connection length for the most flat gain. Th egain flatness of 2-module amplifier of which connection length between modules had been determined by presented methods was the best one out of performances with various arbitrary connectio lengths.s.

Intervalence subband absroption of normally incident infrared radiation in p-type InGaAs-InAlAs coupled quantum well (CQW) is theoretically investigated by the multiband effective mass formalism. By solving a 4*4 luttinger-kohn hamitonian, we calculate valence subband structures, intervalence subband transition matrix elements, and absorption coefficient spectrum in the CQW which consists of a wider well, a thinner well and a barreir between them. Using the flexible design parameters given to the valence band CQW structure, we show that the absorption coefficient profile can be tailored. For a carefully designed CQW, theabsorption coefficient cn be made to maintain a large value over a wider wavelength range of incident infrared radiation compared with that shown in intersubband absorption in usual single quantum well.

In this study, two types of the direct digital frequency synthesizers (DDFS) designed and implemented using 1.0.mu.m CMOS gatearray(SOG) technolgoies are interoduced. To analize the effect of the number of phase bits(L), address data bits(A), and DAC bits (D) on the output spectrums of the DDFSs, the NCO-based BCD-DDFS composed of L=24, A=14, and D=8, and the improved binary-DDFS composed of L=24, A=8, and D=10 have been studied. The chips have been designed with and without a noise shapper to reduce spurious noises due to phase truncation and reduced sine ROM in output spectrum.

Design and performance of principal four ICs for the 10-Gb/s optical communication system are presented. AlGaAs/GaAs HBTs are basic devices to implement a laser diode driver, apre-amplifier, and a limiting amplifier, and GaInP/GaAs HBTs are used for an AGC amplifier. We fbricated 11.5-GHz LD driver, a pre-amplifier, and a limiting amplifier, an dGaInP/GaAs HBTs are used for an AGC amplifier. We fabricated LD deriver, 10.5 GHz pre amplifier, 7.2 GHz AGC amplifier, and 10.3 GHz limiting amplifier, optimized circuit design and the stabilized MMIC fabrication process.

By introducing a compressive-strained quanternary InGaAsP quantum-wells instead of a conventional ternary InGaAs quantum-wells in 1.55.mu.m DFB-LD, the lasing performances canb e improved and the problems caused by the thickness non-uniformity and the compositional abruptness among the hetero-interpaces canb e relaxed. In this paper, we investigated an iptimum InGaAsP/InGaAsP multiple-quantum-well(MQW) structure as an active layer in a direct-modulated 1.55.mu. DFB-LD from the view point of threshold current, chirping charcteristics, and resonance frequency. The optimum compressive-strained MQW structure was revealed as InGaAsP/InGaAsP structure with strain amount of about 1.2%, number of wells $N_{w}$ of 7, well width $L_{w}$ of 58.agns.. The threshold current density J of 500A/c $m^{2}$, the linewidth enhancement factor a of 1.8, and differential resonance frequency of d $f_{r}$/d(I-I)$^{1}$2/=2GHz/(mA)$^{1}$2/(atI=2 $I_{th}$) were expected in 1.55.mu.m .gamma./4-shifted DFB-LD with the cavity length of 400.mu.m long and kL value of 1.25. These values are considerably improved ones compared to those of 1.55um DFB-LD with InGaAs/InGaAsP MQW which have enhancement factor and the resonance frequence frequency by the detuning of lasing wavelength and gain-peak wavelength. It was found that the linewidth enhancement factor of 20% and differential resonance frequency of 35% without the degradation of the threshold current density could be enhanced in the range of -15nm~-20nm detuning which can be realized by controlling the thickness and Incomposition of InGaAsP well. well.and Incomposition of InGaAsP well. well.

With continued miniaturization and development of new devices, the highly nonuniform oxidation of three-dimensional non-planar silicon structures plays an increasingly important role. In this paper, the three-dimensional finite element numerical simulator. Grwoth of oxide is a coupled process of diffusion of oxidant and deformation of oxide. Because boundaries of oxide are moved in each time step and LOCOS structure is formed three-dimensional shape of sruface, it is necessary to develope an efficient node control algorithm that can locally generate and eliminate the node. Therefore we have developed the optimized three-dimensional mesh generator which is cpable of refining and eliminating the meshes at the moving boundary of oxide, and hve developed three-dimensional finite element oxidation solver.

Inductively coupled Cl$_{2}$ plasma has been studied to etch Pt thin films, which hardly form volatile compound with any reactive gas at normal process temperature. Low etch rate and residue problems are frequently observed. For higher etch rate, high density plasma and higher process temperature is adopted observed. For higher etch rate, high density plasma and higher process temperature is adopted and thus SiO$_{2}$ is used as for patterning mask instead of photoresist. The effect of O$_{2}$ or Ar addition to Cl$_{2}$ was investigated, and the chamber pressure, gas flow rate, surce RF power and bias RF power are also varied to check their effects on etch rate and selectivity. The major etching mechanism is the physical sputtering, but the ion assisted chemical raction is also found to be a big factor. The proposs can be optimized to obtain the etch rate of Pt up to 200nm/min and selectivity to SiO$_{2}$ at 2.0 or more. Patterning of submicron Pt lines are successfully demonstrated.

Steady-state I-V characteristics of a resonant tunneling diode (RTD) is calculated by numerical analysis using quantum liouville equation based on wigner function which is derived from density operator. Modifications to the conventional discrete model are made to calculate more accurate quantum correlations. It is pointed out that we must include inelastic processes and the resistivity of the contacting layers to get a much more credible potential which can be theoretically obtained from the simple screening theory. The effects of spatially-varying effective mass is also checked briefly.

We constructed a very accurate fiber bragg grating sensor system using an unbalanced mach-zehnder interferometer which converts the measurand-induced bragg wavelength shift to phase change of th einterference signal. With a shielded reference grating, output phase drift which orginates from the thermal drift on the interferometer was compensated. Mesured accuracy of the constructed system was 0.03 .deg. C and 0.26.mu. strain for temperature and strain measuremtn, respectively. It is over 300 times better resolution compared to the conventional optical spectrum analyzer which has 0.1nm wavelength resolution capability.

The surface characteristics of polymers for liquid crystal alignment are studied by optical second-harmonic genertion (SHG) tecnique. Using SHG technique, the LC monolayers on rubbed polymer have already been studied. But, in this paper, the SH signals of polymer were observed and the orientational distribution of oriented polymer was studied. Te SHG experiments for side-chain type and main-chain type polymers are carried out as a function of rubbing strength. The orentational distribution of surface molecules of polymers is compared with the LC pretilt angle measured by the crystal rotation method.

We determine the dispersion profile of a dispersion decreasing fiber(DDF) for optimum pulse compression from a trade-off between high pulse compression and low pedestal power/short DDF length. We find that the optimum vlaue of the exponential dispersion decreasing factor .alpha. is 0.95 and that the corresponding optimum fiber length is 1.5 times of the initial soliton period. Passing through the dDF, ~10 times of pulse comparession ratio can be achieved without significant increase in pedestal power. To compress relatively broad pulses using a given DDF optimized at a specific pulse width, we also detemrine the optimum input pulse amplitude, as a function of input pulse width.