No abstract

Circuits for CMOS two-dimensional (2-D) array data transfer are indispensable for applications such as space and nuclear fields. Issues include to be operated with higher speed, lower power, fewer size penalty and radiation hardness. To meet these requirements, two kinds of CMOS 2-D array data transfer circuits, such as a shift register type and a memory access type, are proposed and fabricated by the standard 0.18-µm CMOS process technology. In the both types, 16 µm pitch, 8×124 array data transfer operations were realized with data rate of more than 1 Gb/s. Furthermore, we conducted 60Co γ-ray irradiation experiments on those circuits. The current consumption ratio of the shift register type to the memory access type ranges from 150 to 200% as the dosage increases. The result indicate that the memory access type has better radiation hardness at 1 Gb/s than that of the shift register type.

A Z-gate layout MOSFET (ZLT) is expected to reduce the total ionizing dose (TID) effects which occur in the shallow trench isolation (STI) layer made of oxide. To verify the ZLT’s merit, the ZLT was designed and fabricated with the 0.18-µm standard CMOS technology. The ZLT was irradiated with Co60 γ-ray up to 10 Mrad, and its I-V characteristic fluctuation was compared with that of a standard straight gate layout MOSFET (SLT). As a result, it was confirmed that the Z-gate can suppress off-current fluctuations by 1/5, on-current fluctuations by 1/2 and threshold voltage fluctuations by 1/6 compared with the SLT. Since the size penalty of the ZLT is small, it is possible to improve the radiation hardness of CMOS LSIs circuit by just replacing SLTs to ZLTs.

The response of N-type MOSFET characteristics to TID (Total Ionizing Dose) effects caused by a γ-ray irradiation was modeled as a sum of two exponential functions which expresses the charge accumulation of the oxide traps and the interface traps respectively. This model was applied to the circuit simulation of a ring oscillator operation. It was shown that the simulation result reproduces the variation in the measured oscillation frequency experimentally. The simulation also showed that there is an optimum size ratio between N-type MOSFET and P-type MOSFET which can minimize the variation in oscillation frequency due to TID effect. These obtained results verified that the proposed method can be applied to the prediction of response in a dynamic operation circuit to TID effects.

We demonstrate the optical gating function of VCSEL-based active Bragg reflector waveguide with slowing light. The insertion loss decreases with injection current, functioning as an optical gate. An on-off ratio of over 10dB was obtained for 30µm long device nearby the cutoff wavelength. waveguide, we formed additional 10 pair-Ta 2 O 5 /SiO 2 DBR on 50 µm square mesas. The length of the slow light waveguide is 30 µm. We are able to inject current from a top contact into GaInAs/GaAs QWs included in one-λ core, making the waveguide active. Figure 2 shows the calculated slow-down factor, which is defined as the ratio of the group velocity of slow light versus that in conventional semiconductor waveguides. The slow-down factor is over 10 nearby the cutoff condition. An important issue is how to couple with slow light in a Bragg waveguide. We already demonstrated a simple method of a tilt-coupling scheme as shown in Fig. 3. We assumed that an input beam is 40 degrees tilted from the vertical axis. The calculated intensity distribution in the crosssection is shown in Fig. 3. The coupling loss is less than 5 dB for TE and TM modes with a 4 µm-spot-size Gaussian beam input. Optical Gating with Injection CurrentWe carried out the coupling experiment of slow light using a tunable laser as shown in Fig. 4. The tilt angle of input light through a lensed fiber is 40 degrees as suggested in the modeling. We observed the near-field pattern of reflected light and the output of slow light guided in the Bragg reflector waveguide. The excitation of slow light is dependent on the position of input light as shown in Figs. 5 (a) and (b). With precise optical alignment of the input lensed fiber, we could see the output from the opposite side of a 30 µm long slow light waveguide as shown in Fig. 5(b). We obtained the insertion loss from the measured near-field intensity of the output. The result is shown as a function of wavelength for TM mode in Fig. 6. We could achieve a minimum insertion loss of 4 dB, indicating the low coupling loss of the tilt coupling scheme. The increase in the insertion loss at the longer wavelength side is due to the increase in the slow-down factor. The cut-off wavelength is around 980 nm for the measured device. We injected current from a top contact into triple InGaAs QWs contained in the Bragg reflector waveguide. The injection current is 10mA in Fig. 6, which is corresponding to current density of 400A/cm 2 . With injection current, we could see noticeable changes in insertion loss of the slow light waveguide, which are getting larger closed to the cut-off wavelength. While we could not get a "net" gain, the result shows a possibility of loss compensation of slow light waveguides, which is a common difficulty in slow light waveguides. The on-off ration with and without injection current is over 10 dB nearby the cut-off wavelength. This is larger than the extinction ratio for our previous slow light modulators [6]. ConclusionWe demonstrated a slow-light optical gate with slowing light in a Bragg reflector w...