A transmission type phase shift attenuator has a poor reflection characteristics at an output port. And that is controlled by the current, its dynamic rnage decreases due to the current limitation. In this paper, to avoid such disadvantages, a reflection type low phase shift attenuator has been designed and measured. As a result, at a center frequency (1855MHz), the sreflection type low pahse shift attenuator has an attenuation of 30dB, within the limit of 3 phase shift and less than -17dB reflection characteristics at both input and output ports. And it demonstrates the peformance of the reflection type low phase shift attenuator is better than the transmission type phase shift attenuator with the same measurement specifications.

A rigorous analysis method of electromagentic scattering from frequency selective surface with gridded square loop elements in case of oblique incident and arbitrary polarization is presented, which uses the roof-top subdomain basis function. The frequency response and polarization characteristics of the reflected wave and the transmitted wave for various widths of the grid and the conductor square loop, and for the various gaps between the grid and the conductor square loop, and for the various gaps between the grid and square loop is investigated. To confirm the validity of presented method, frequency selective surfaces with gridded square loop elements are fabricated with honeycomb structures, calculate dvalues for the frequency response of the reflected wave and the transmitted wave for arbitrary incident angle and polarization are compared with measured values.

Waveguide bandpass filters using E-plane or H-plane discontinuities have been used in many satellite communication systems. Since the transmitter and receiver of such systems tend to be getting smaller, components such as bandpass filters must also be reduced in size. In this paper, in order to reduce the waveguide, itself, of which bandpass filters consist, double plane step waveguide discontinuities are considered. They are analyzed using the mode matching method and the genralized scattering parameter of the cascaded structure is calculated. The characteristics of the bandpass filter which is designed using the reduced size waveguide is optimized by OSA90$^{TM}$. The bandpass filter has a good characteristics and smaller size than those made of the E-plane or H-plane step discontinuities.s.

Square spiral inductors are designed with EM program in accordance with the inner diameter and the metal thickness which is 0.2.mu.m and 20.mu.m respectively. We propose a parameter extraction method based on the S-parameter. Lumped equivalent circuits of spiral inductors are analyed with reflection coefficient S$_{11}$, of witch freqency rnage is 1~10GHz. When metal thickness is 0.2.mu.m, S$_{11}$ with EM simulation is not the same as S$_{11}$ that of SPICE simulation. So we suggests a new lumped equivalent circuits which compensate circuits. Te new lumped equivalent circuits are adequate for other inductor with small scale at high frequencies.ncies.

A novel high-Q vertical jinductor for MMICs is proposed and characterized in a wide range of frequencies (DC~10 GHz) using the numerical methods such as the PeEC(partial equivalent element circuit), the FDM (finite difference method) and the MoM (method of moments). Electrical superiority of the vertical inductor to the horizontal is observed in terms of the magnetic flux linkage and the ground screening effect. The veritcal bondwire inductor is designed in consideration of the wire bonding feasibility and the optimum electrical peformance. This structure is also analyzed using the equivalent circuit and compared with the conventional spiral inductors From the calculated results, high Q-factor, inductance, and cut-off frequency are observed to be inherent characteristics of the veritcal bondwire inductor.

Optical filters with graded index profiles are designed by applying the fourier transform to a riccati equation which governs the reflection and transmission characteristics of inhomogeneous refractive index distributions. The inhomogeneous refractive index profile of an optical filter with specified target spectrum is obtained through iterations. The spectra response of the inhomogeneous refractive index layers are analyzed by using runge-dutta numerical method to solve the differential euations of the amplitude and the phase of reflection coefficient derived from the riccati equation and the results are in good agreement with the resutls obtained by using matrix method.

The dependence of the electrical properties in planar MQW - APD & pn, and p - i - n photodiode implanted with oxygen on the annealing emperatures and ion dose has been investigated. The oxygen implantation was performed for inter-device isolation. The leakage current of as-impanted p-i-n photodiode obtained was less than 50 nA. An annelaing temperature dependence study shows an abrupt increase of leakge current at 600.deg.C for all devices under study. This indicates that donor complex centers introduced by the chemical activity of oxygen increase with increasing annelaing temperatures. Furthermore, leakage current was highly correlated with oxygen dose due to th eimplanted related defects.

In order to improve the electrical performance of ITO/InP solar cells, sulfur passivation technique was employed using (N $H_{4}$)$_{2}$ $S_{x}$ solution. Passivation effects were analyzed by measuring the short circuit current density ( $J_{sc}$ ) of solar cells and photoluminescence (PL) of ITO/InP interfaces. This paper firstly reports the sulfur passivation effects by investigating the correlation between the PL intensity and the short circuit current. Generally, PL intensity and the short circuit current of sulfur passivated sampels wer eincreased, and showed the same trend. Especially, samples prepared at 60.deg. C (N $H_{4}$)$_{2}$ $S_{x}$ solution exhibited the highest $J_{sc}$ and PL intensity. These results demonstrated that the short circuit currents was influenced by the ITO/InP interface states.

There generally exists a large variation in the thereshold voltages of the flash EEPROM cells after they are erased by using th fowler-nordheim tunneling, thereby getting some cells to be overeased. If the overerased cells are programmed with the conventional one-step programming scheme where an 12-13V pulse with the duration of 100.mu.S is applie don the control gate for the programming, they can suffer from the significant degradation of the reliability of the gate oxide. A two-step programming schem, where an 8/12 V pulse with a duration of 50.mu.S for each voltage is applied on the control gate for the programming, has been studied to solve the problem. The experimental results hav eshown that there is little difference in the programming characteristics between those two schemes, whereas the degradation of the gate oxide due to the programming can be significantly reduced with the two-step programming scheme compared to that with the one-step programming scheme. This is possibly because the positive charge stored in the floating gate of the overerased cells is compensate dwith the electrons injected into the floating gate while the 8V pulse is applied on the control gate, which leaves the overerased cells in the normally erased state after the duration of the 8V pulse.

A new etching technique of meltback was investigated for GaAs lensed optical devices with selective windows opending in the LPE (liquid phase epitaxy) system. In the meltback process, the etch depth and the etch shape were controlled by the degree of under-saturation, etch time and other parameters. A GaAs/AlGaAs DH layer was grown on the selectively etched hemispherical well for optical device application such as lensed surface emitting LED. The regrowth process were related with the coolin grate and the well to well spacing. A novel surface emitting LED with hemispherical AlGaAs lens was fabricated using the meltbakc and regrowth as the key process for AlaAs lens array. The light emitting efficiency of the LED was upto three times higher than the similar structure LED without lens. The meltback and regrowth technique was applicable to manufacture the optical device in LPE.

The dominant physical process repsonsible for the nonlinear gain is from spectral-hole burning with the time constant fo about 50fs and the contribution to the nonlinea rgain form hot carriers effect is determined to be about 15% of the contribution due to spectral-hole burning. To prove the above results we fit the data of hall and found that hot carriers have a profound effect on their experimental data despite the fact that the magnitude of hot carriers effect is only 15% of spectral-hole burning. We suggest that the experimenta with a pump pulse width of 180 fs is very sensitive in detecting the effect of hot carriers, but is not sensitive in detecting much faster process associated with spectral-hole burning.