The heterogeneous integration of optoelectronic, electronic, and micro-mechanical components from different origins and substrates makes possible many advanced systems in diverse applications. Besides the monolithic integration approach, which is the basis for the success of today's silicon industry, various hybrid integration technologies have been explored. These include flip-chip bonding, micro-robotic placement, epitaxial lift-off and direct bonding, substrate removal and bonding, and several self-assembly methods. In this paper, we describe the results of our monolithic integration effort involving a 2 by 2 optoelectronic switching circuit and an 8 by 8 active-pixel sensor array on GaAs substrates, and a 16 by 16 spatial light modulator array produced by flip-chip bonding of III-V multi-quantum-well (MQW) modulators and silicon driver circuits. We also present our preliminary experimental results on the self-assembly of small inorganic devices coated with DNA polymers with self- recognition properties.

Progress towards integrated photonic devices by three-step selective-area MOCVD is reviewed. Using the selective growth process, the quantum well thickness and, hence, the emission wavelength of buried heterostructure devices can be defined anywhere on a wafer surface by using an appropriate mask geometry for the active region regrowth. This in-plane bandgap energy control allows the designer to fabricate devices with different wavelengths on the same wafer for integrated photonic applications. Since no growth occurs on the oxide mask, the spacing between stripes defines the width of the lateral waveguide, and the width of the stripes defines the amount of growth rate and composition enhancement in the quantum well. As a result, a very wide range of emission wavelengths (960 - 1060 nm) can be obtained over the wafer surface in a single growth. This paper reviews a few of the high performance photonic devices fabricated using this method.

Realization of optoelectronic integrated circuits on silicon substrates has many difficulties, one of which is depositing high quality light-emitting material on the silicon surface. A desirable depositing method from a manufacturing point of view is chemical vapor deposition. Because there are currently no light emitting semiconductor alloys lattice-matched to silicon, epitaxial growth of III-V compound devices on Si has required a lattice constant engineering step such as wafer bonding or thick buffer layer growth. Growth of GaInP/GaP strain-induced quantum dots offers an opportunity to grow single crystal light-emitting devices monolithically on silicon substrates without lattice constant engineering steps, since single crystal GaP can be grown on silicon. In this presentation, progress on MOCVD growth of GaInP/GaP quantum dots and its device applications are reviewed.

In recent years photodetector structures have become more complex in response to more stringent performance demands such as higher bandwidth and lower noise. There are, however, fundamental performance limitations such as the responsivity/bandwidth tradeoff that have not been adequately addressed. At present, it appears that significance increases in performance will be the result of major technological breakthroughs rather than incremental improvements in conventional structures. Our productive approach is the implementation of resonant-cavity structures.

Resonant cavity enhanced (RCE) photodiodes are promising candidates for applications in optical communications and interconnects where ultrafast high-efficiency detection is very desirable. In RCE structures, the electrical function of the photodiode is largely unchanged, but optically it is subject to the effects of the cavity, mainly wavelength selectivity and a large enhancement of the resonant optical field. The increased optical field allows photodetectors to be made thinner and therefor faster in the transit-time limited operation, while simultaneously maintaining a high quantum efficiency at the resonant wavelengths. The combination of RCE detection scheme with Schottky photodiodes allows for fabrication of high-performance photodetectors with relatively simple material structure and fabrication process. In RCE Schottky photodiodes, a semi-transparent metalization can be used simultaneously as the electrical contact and the top reflector for the resonant cavity. Device performance is optimized by varying the thickness of the Schottky metalization and utilizing a dielectric matching layer. We present theoretical and experimental results on spectral and high-speed properties. We have demonstrated RCE Schottky photodiodes in (Al, In)GaAs/GaAs material system with temporal response of 10 ps full-width-at-half-maximum. These results were measurement setup limited and a conservative estimation of the bandwidth corresponds to more than 100 GHz. The photodiodes were designed and fabricated for 900 nm and 840 nm resonant wavelengths. The best measured quantum efficiency is around 50% which is slightly less than the theoretical prediction for these devices.

We describe a new method of sensing the linear polarization of light in a single mesa device structure by vertically integrating two photodetectors. The monolithic architecture eliminates the need for several discrete components, such as polarization filters and beam splitters, thus reducing critical alignment requirements and cost for various optical systems. Applications include the simplification of reading heads in magneto-optical (MO) data storage devices and constructing imaging arrays for polarization vision. In imaging, polarization sensing can extract additional information from a scene otherwise not noticeable to the human eye, facilitating remote sensing, material classification, and biological imaging. The operation principle of our vertical cavity polarization detector (VCPD) is based on a resonant cavity enhanced (RCE) photodetector, being vertically integrated with a conventional photodetector. The RCE detector is constructed by integrating a thin absorption region into an asymmetric Fabry-Perot cavity. The top reflector is formed by the semiconductor air interface, while the bottom mirror is a distributed Bragg reflector (DBR). For off-normal incidence of light, the reflectivity of the semiconductor-air interface and DBR are significantly different for TE (s) and TM (p) polarizations. Thus the RCE detector provides resonance enhancement for TE, capturing the TE polarized light in the top detector. For TM polarized light, both reflectivities are small, therefore, light is transmitted to and absorbed in the bottom detector. A large contrast in TE/TM response of the top and bottom detectors is achieved and the linear polarization can be computed from their relative responses. Experimental results displaying good agreement with simulation results have been recently achieved and are presented.

Silicon photodiodes with active area dimensions of 1-mm and 5- mm lengths and approximately 30-micrometers width fabricated by a standard, 2-micrometer complimentary metal-oxide-silicon process have been used in a coherent optical processing system with ac coupling. The effective bandwidth of the detectors as measured in the coherent optical test arrangement was greater than 1.25 GHz, and it is estimated that the actual bandwidth of the detectors as used in such a system may be 5 GHz, where ac coupling removes the effects of the full widths at half maximum and of the fall times of the detectors on the test arrangement response. Photodetector rise times in response to impulse optical excitation at 810-nm wavelength were of the order of 100 ps.

This paper reviews novel techniques which are used to modify the properties of the optical beam generated by vertical cavity surface emitting laser diodes. The main emphasis within the paper concerns the use of etches applied to the device surface either adjacent to or across its optical aperture. Such etches can be used to modify facet reflectivities sufficiently to pin the optical mode precisely and reduce the onset of higher order modes. In addition, single polarization operation is achieved by using deeper etches placed outside the light emitting region. Using this technique, previously multimode components may be transformed to single mode components with precisely fixed polarization states. The effect of these etches on other lasing properties such as spectral performance and power efficiency is also discussed.

Birefringence, polarization, and wavelength filtering properties of subwavelength transmission gratings (SWTGs) and their applications in vertical cavity surface emitting lasers (VCSELs) are presented. Particularly, large birefringence (over two orders of magnitude larger than natural birefringence crystals) and polarization effects have been achieved in thin (240 nm thick or less) amorphous silicon subwavelength gratings (SWTGs). The SWTG polarizers were used to control the polarization of VCSELs (i.e., fixing, enhancing and switching of the polarization), making the maximum polarization of a VCSEL creased to 300:1 from 20:1. The SWTG's waveplates were used to build a polarization-switching VCSEL oscillator that has a can tunable frequency up to terahertz.

We report the fabrication and testing of long wavelength triangular ring lasers operating at 1.26 micrometer. The triangular ring laser, fabricated in InGaAsP/InP, employs a ridge waveguide structure with the ridge and facets dry etched using high temperature CAIBE. Pulsed L-I, spectra and output characteristics are analyzed. Preliminary testing is done for ring lasers monolithically integrated with tapered amplifier. Temperature characteristics are studied by cooling the ring lasers with a TE cooler. Further work on noise characteristics require the use of optical isolator to reduce feedback from the optical fiber used in the measurement setup.

We have constructed a novel VCSEL structure in which a voltage biased quantum well absorber is embedded in the mirror stack. Several applications for these devices are demonstrated. The devices exhibit regimes of negative differential resistance in the absorber under varying laser bias conditions. By choosing the absorber bias conditions such that the operating load line intersects the absorber IV trace three times, the laser can be made to exhibit optical bistability. The magnitude of the bistability and consequent hysteresis loop can be controlled by adjusting the absorber bias. If the bias conditions are adjusted so that the load line is tangential to the portion of the absorber IV exhibiting negative differential resistance, then the device can be made to self-pulsate. Self-pulsations at frequencies as high as 2 GHz with 700 kHz rf linewidth FWHM (full-width at half-maximum) have been obtained using these devices. Furthermore, the self-pulsation frequency can be tuned over 700 MHz by adjusting the bias conditions, representing a substantial advance over existing self- pulsating VCSELs. We have also demonstrated a novel modulation scheme using these devices, in which the drive signal is applied to the bias voltage across the absorber. Theoretical analysis of the chirping mechanism leads us to expect that this technique will minimize the chirp at high modulation frequencies, while still providing substantial modulation depth and high speeds. We achieved a minus 3 dB bandwidth of 9 GHz; the bandwidth increases with the laser bias current at a rate of 7 GHz/(root)mA.

New enabling technologies are needed for optical communication systems to accommodate rapidly growing traffic demands. Wavelength conversion and high-speed optical packet switching/routing will be key technology components for realizing more flexible and efficient optical networks. Lasers capable of wide-band, high-speed wavelength tuning will be essential to support these advanced functions. Also, many applications will require high launch powers in order to access an increasing number of users, nodes, or base stations. Hence, laser transmitters capable of suppressing stimulated Brillouin scattering (SBS) would be highly desirable. We have developed an ultrafast, broadband tunable laser, based on an electroabsorption modulator laser (EML), which exhibits wavelength switching speeds as fast as 56 ps. Here, we report system performance results on wavelength conversion high-speed optical packet switching, and SBS suppression using this device. We have tested multiple wavelength conversion sequences and demonstrated penalty-free transmission through two cascaded wavelength conversion stages including 200 km of standard non-DS fiber. When used to perform packet switching at 2.5 Gb/s, the tunable laser allows switching between optical packets on 4 wavelength channels in less than 1 bit period, thereby requiring no significant guardband. The modulated data packets have been transmitted through 200 km of non-DSF and yield open eye diagrams. The tunable laser has also been used to perform SBS suppression. We have measured SBS thresholds of approximately 25 dBm on 4 separate WDM channels. The required modulation signal is very small, 95 mV_pp, and the residual AM is only approximately 1%.

Lateral current injection lasers are ideal sources for monolithic OEICs. They are grown on semi-insulating substrate for low parasitic conduction and delays; consist predominantly of undoped layers for ease of inter-device isolation and low- free carrier induced absorption and chirping; have both electrodes on the surface and can be readily integrated with other devices using a planar process with a minimum number of steps. Vertical injection lasers, though their performance as discrete devices is greatly perfected and presently enjoying the monopoly position with no competition in sight, are largely laterally invariant and do not generally possess these characteristics. On the other hand, lateral current injection lasers remain to be studied systematically and greatly improved to become a serious contender in the field. This article reviews the attempts, of ours and others to date, and outlines the important issues concerning LCI's unique internal device operation mechanisms and device design. It also aims to draw attention to the fact that ease of integration is only one feature of the great potential of the lateral injection lasers. This potential comes from the release of an additional degree of freedom, and offer many exciting and unique opportunities for developing new devices and new functionalities.

Monolithic, multiple-wavelength VCSEL arrays have been obtained by using the surface-controlled enhancement and reduction of the MOCVD epitaxial growth rate to produce a periodic and repeatable grading of the resonance wavelength over a span of greater than 30 nm. Room temperature, electrically-injected, cw lasing has also been achieved with a wavelength span of greater than 20 nm. We show here both the enhancement and the reduction of the growth rate of the entire VCSEL structure and demonstrate the controlled variation of the VCSEL lasing wavelength over a widened spectral range by exploiting both of these effects. Using the same growth techniques, wavelength-selective, resonance-enhanced photodetector arrays with closely-matched resonance wavelengths can be monolithically integrated on the same epilayer structure. We demonstrate the repeatability of this technique using different arrays from the same growth run.

The inherent structural incompatibility between lasers and transistors is an obstacle in the monolithic integration of these two devices for optoelectronic integrated circuits (OEICs). One attractive solution could be in devising a common device layer structure which would support both functionalities. This would enable easy integration without selective area regrowth or reduce wire interconnects to link the laser and transistor. Among others, recently presented at Photonics West '97 was two independent attempts to exploit the idea of using a common epitaxial layer structure to fabricate both a laser and a transistor. These have been mostly empirical in nature. The common layer structure strategy imposes more constraints in the design and also on the experimental conditions required to achieve optimal performances in both laser and transistor functions than that for an individual function. However, in making heterojunction bipolar transistor (HBT) compatible laser structures, design issues reach beyond the study of layer compatibility to include issues involving contact configuration and doping profile re-engineering. A self-consistent theoretical investigation is called for to examine these new related issues and to further expand our comprehension of the device characteristics and performance possibilities. In this paper, we report on our findings of an in-depth theoretical investigation of the unique operational features of a new class of laser structures that are HBT compatible. They are, in fact, 'doping-converted' lasers fabricated from modified HBT structures. We highlight some of the new, largely unexpected, effects arising from the necessity of doping- conversion and of HBT compatible contact configuration. We describe quantitative results using the two example structures recently reported at Photonics West '97. These two examples, with similar QW insertion locations in the HBT structure, exhibit dramatically different lasing threshold and differential efficiency, and thus serves as a good demonstration of the underlying physics involved with the effects arising from contact configuration and doping profile re-engineering.

External modulation of cw laser radiation by multiple quantum well electroabsorption modulators will potentially play an important role in rf photonic links, especially at high microwave frequencies and millimeter waves. InAsP/GaInP MQW on InP and GaInAs/InAlAs MQW on GaAs modulators have been grown by MBE and fabricated into p-i-n modulators. Performance with -26 dB link efficiency without amplification, 5 dB insertion loss, 15 mW of optical power and 17 GHz bandwidth has been experimentally demonstrated. Extension to 100 GHz bandwidth with -39 dB link efficiency (without amplification) can be expected. Traveling wave modulators and on-chip impedance matching of p-i-n modulators have been designed, fabricated and evaluated. Traveling wave modulators with flat frequency response over 40 GHz have been experimentally demonstrated.

We demonstrate the performance of InGaAs/InP multiple quantum well modulators for fiber access applications. On/off contrast ratios in reflectivity are 22:1 with a 10:1 level extending over a 26 nm bandwidth at 1.55 micrometers. Arrays of high quality pin InGaAs with low dark and zero bias operation are realized and further applied to avalanche photodiodes.

Ridge waveguide, edge-emitting single quantum well GaAs lasers with an integrated gating electrode have been fabricated. These devices integrate a MESFET structure with the laser PN junction so that the SBD (Schottky barrier diode) depletion layer can be used for transverse current confinement in the laser. Device fabrication was very simple requiring only an anisotropic etch for waveguide definition followed by a single self-aligned contact deposition step. The Schottky barrier depletion layers on either side of the ridge waveguide act to confine free carriers. This structure allows for separation of the optical and electrical confinement in the transverse direction without requiring complex fabrication. The device demonstrated modulation of the pulsed lasing threshold with gate control voltage on a 30 micron wide ridge. Above threshold, increasing power output with increasing gate voltage was demonstrated with negligible gate current. The multimode lasing spectrum showed that the increased power output occurred for all modes with no shift in the mode wavelengths to within the resolution of the measurement system.

A new cascaded multiple quantum well (MQW) vertical light modulator with low output phase shift characteristics is proposed. In the proposed structure, one of the cascaded MQW modulator is used as a light switch and the other MQW modulator is used as a phase compensator to offset output phase shift. We modeled the performance of the proposed modulator in which GaAs/AlGaAs MQW waveguide segments were positioned outside the main waveguide composed of AlGaAs material. Calculations based on beam propagation method (BPM) show that the proposed modulator has an output phase shift as low as 0.35 radian for 15 dB attenuation with a total device length as short as 300 micrometers for TE single-mode operation at wavelength of 800 nm.

LiNbO₃ waveguides with Si overlays are emerging as a basic building block for a variety of integrated-optic components. However, the development and optimization of these devices are, in large part, hindered by the lack of understanding of the specifics of the Si-on-LiNbO₃ structure which appear to differ dramatically from those of the Si and LiNbO₃ waveguides, considered separately. In this work, we provide a specific insight into the waveguiding properties of vertically stacked Si-on-LiNbO₃ waveguides. In particular, we present a detailed theoretical analysis of the effect of the Si film on the modal characteristics (propagation constant and field distribution) of the structure. We show that for approximately 70% of all Si thicknesses, in the range from 0 to 1.6 micrometer, the highest-order normal mode of the entire structure has more than 99.9% of the total energy confined in the LiNbO₃ region, i.e., beneath the Si overlay. This fact is quite intriguing given the fact a planar Si layer of submicron thickness on bulk LiNbO₃ is already multi-moded. Furthermore, we show that the effective mode index of the structure is considerably modified compared to that of the LiNbO₃ waveguide while the propagation loss is, on the other hand, practically unaffected (approximately 0.3 dB/cm) even in the presence of the lossy Si film, as confirmed by our previous experimental results. Evidently, large modulation of the effective index and low-loss propagation provide an ideal combination of properties suitable for the fabrication of high-reflectance corrugated waveguide gratings, essential for a number of practical devices, in particular, WDM filters.

We report stress-induced channel waveguides formed in epitaxial BaTiO₃ films. BaTiO₃ epitaxial films (doped with and without erbium) were grown on MgO (001) single- crystal substrates using rf magnetron sputtering. In the channel waveguides developed, the lateral confinement of light is achieved via the photoelastic effect in BaTiO₃ induced by thin-film stress. As a stress-inducing film, a 0.5- micrometer-thick SiO₂ film was sputter-deposited on top of a 3.0-micrometer-thick BaTiO₃ film with a 7 to 10- micrometer-wide window opening. The fabricated structures were characterized in terms of their guided mode profiles at 1.3 and 1.55 micrometer wavelength. The measurement result clearly shows both the lateral and vertical confinement of light in the channel region. The stress distribution in the channel structure was calculated by solving the coupled equations that describe the elastomechanical and piezoelectric effects in the ferroelectric material. The refractive index changes were than calculated taking into account both the photoelastic and electro-optic effects of BaTiO₃. The simulation results show a good agreement with the measurement results. The waveguide structure developed in this work does not require etching of BaTiO₃, and is expected to be useful as a simple and economical method for forming channel waveguides with other ferroelectric films as well.

We theoretically and experimentally investigate a recently proposed integrated tunable laser which consists of a curved waveguide distributed-feedback (DFB) laser and a passive electroabsorption (EA) modulator-separated by an isolation section. The modulator provides wavelength selectivity in the integrated device. The facet at the laser side is anti- reflection (AR) coated to enhance the light output power, and the EA side facet is high-reflection (HR) coated to increase the coupling between the two sections. We have developed a transfer-matrix model based on the coupled-mode theory which takes into account the curved waveguide and above-threshold spatial hole burning effects. Using this model, we study the wavelength tuning behavior of the integrated device when a voltage bias is applied to the modulator section.

Opto-electronic devices offer a wide-band, low power, compact solution to a variety of functional requirements in both military and civilian systems. The applications include switching and modulation for telecommunications, microwave and mm wave radar and electronic warfare. The integration of opto-electronic devices on to waveguides promises the benefit of increased functionality and reduced cost systems. A key requirement of such integration is that devices be electrically isolated from each other but remain optically connected. This isolation may be achieved by etching a notch through the top conducting layer of a p-i-n structure, for example using reactive ion etching. This can be problematical owing to the need to control etch depths to high precision. We have developed a process for isolating hard walled AlGaAs/GaAs waveguides using proton implantation. The effect of the proton energy and the level and type of doping on the degree of isolation and transparency achievable have been investigated. Post implantation annealing has been found to have a significant effect. We have found that our process does not significantly alter the transparency of the waveguides and gives excellent electrical isolation. In contrast to the etched notch method our process also gives very good yield. Using this technique MachZehnder (MZ) modulators have successfully been fabricated and demonstrated. Keywords: Optoelectronics, proton isolation, waveguides, integration.

Optical amplification and electro-optic modulation have been observed simultaneously in one polymeric material photo-lime gel which has been used as a volume holographic material to produce dichromated gelatin (DCG) films. In this paper, the dual functions were achieved by doping neodymium chloride hexahydrate (NdCl₃(DOT)6H₂O) and chlorophenol red (C₁₉H₁₂Cl₂O₅S). The optimized doping concentrations of Nd⁺³ and chlorophenol red were 6.7 X 10¹⁹/cm³ and 23% respectively. We observed a gain of 3.8 dB at 1.04 micrometers and an electro-optic coefficient of 30 pm/V at 633 nm. The experimental results confirms that the co-doping process does not degrade the respective functions of Nd⁺³ for optical amplification and chlorophenol red for electro-optic modulation.

This paper presents an overview of the silicon-on-insulator photonic integrated circuit (SOIPIC) technology. Recent progress in material and device technologies including WDM filters are presented.

We report a strong and extremely stable electroluminescence (EL) from silicon based EL device. The active layer of the device utilizes a crystalline Si/O superlattice, grown by molecular beam epitaxy with in-situ oxygen exposure. Oxygen exposure is used to limit the growth of oxides to a monolayer in order to continue the silicon epitaxial growth and to create a highly localized interaction between the oxygen and silicon atoms. The visible EL is peaked at 1.8 - 2 eV, showing no degradation in a six month life-test under continuous operation, longer than other silicon based schemes in the literature. The efficacy of both photoluminescence and electroluminescence is similar to better than that from porous silicon. The robustness and stability of the c-Si/O superlattice opens the door for a silicon chip combining IC with integrated optics.

In recent WDM communication systems, a wavelength-selective filter or detector is required at receiver end to pick up desired channels from incoming data streams. In addition to the wavelength-selectivity, one also desires these devices to have wavelength tunability such that the entire system is reconfigurable. Filters and receivers based on arrayed waveguide grating (AWG) structures have been developed. Even though AWG devices demonstrated good side-lobe suppression ratios (SSRs), they are not tunable. Devices based on vertically stacked grating-assisted codirectional couplers (GACC) have therefore attracted great attention due to the wide and fast wavelength tunability. However, the response of a one-stage GACC device with a uniform coupling strength across the entire filter section is a sinc²-like function which limits its SSR to less than 9 dB. It is well known that there is roughly a Fourier transform relationship between the coupling coefficient distribution and the filter response. Therefore, apodizing the coupling strength along the filter can effectively suppress the side-lobes. In this paper, we report on the apodization of the coupling strength in the GACC filter by varying the grating duty ratio. Using this technique in an integrated wavelength-selective receiver, we demonstrate a device with a 40 nm tuning range and a 16 dB SSR. A schematic drawing of the integrated apodized GACC receiver is shown in Figure 1. The device includes a semiconductor optical amplifier (SOA), a two-stage GACC optical filter, and a waveguide photodetector on InP/InGaAsP based materials. The operating principle of the device is explained as the following: Input light is first coupled into the top waveguide and is amplified the SOA. It is then filtered by two apodized GACC filters in series. At the wavelength where phase matching condition is satisfied, light will be coupled down to the bottom waveguide and then back up to the top waveguide. Finally, it will be detected by the waveguide photodetector. The middle of the top waveguide is broken by a 30 degree tilted etched groove. The tilted groove deflects any uncoupled light laterally to avoid multiple reflections of unwanted signals to the detector.

We have investigated the performance characteristics of InP- based 1.55 micrometer single and multi-channel photoreceiver arrays consisting of a front-end p-i-n photodiode and a 3- stage HBT-based transimpedance amplifier and realized by single-step epitaxy. The single-channel photoreceiver circuits are characterized by an optical bandwidth of 20 GHz and transimpedance gain of 46 dB(Omega) . Sixteen-channel arrays, made with the same circuit, demonstrate individual channel bandwidth of 11 GHz. By using a novel monolithically integrated radiation shield, we have been able to reduce the crosstalk to minus 35 dB at 11 GHz. These parameters represent the best performance in multi-channel integrated photoreceiver arrays. We have performed an electromagnetic full-wave solution of the array, which shows that the measured crosstalk in arrays without the radiation shield could be dominated by radiation crosstalk. The magnitude of the electrical crosstalk in the arrays has also been determined. These results are presented and discussed.

We developed waveguide beamsplitters integrated with an Er- doped thin-film optical amplifier. A 1 to 2-cm-long Er-doped optical amplifier is monolithically cascaded to a beamsplitter in order to compensate for the losses of the beamsplitter. Both devices use a highly Er-doped silicate glass film as a guiding layer. Two different types of structures were investigated for a beamsplitter, i.e., a Y-branch type and a self-imaging multimode interference (MMI) type device. The beam propagation method (BPM) was used for the analysis and design of the beamsplitter structures. The beamsplitter part was designed to be much shorter than the amplifier (i.e., 1 - 2 mm versus 1 - 2 cm) in order to minimize the absorption of a signal beam by Er ions in the potentially underpumped splitter section. Both the amplifier and the beamsplitter parts have a ridge waveguide structure. A novel process technique was developed and used in forming the ridges. The process does not require etching of an Er-doped film in defining the lateral dimension of a waveguide, but involves a lift-off process with a collimated magnetron sputtering. A 1.7-cm-long waveguide thus fabricated shows a 1.55-micrometer signal enhancement of 15.4 dB with a 980 nm pump power of 40 mW. This enhancement fully compensates for both Er absorption and waveguide losses, and results in a gain of 7.2 dB. This demonstrates that 1 - 2-cm-long waveguide amplifiers can provide an optical gain sufficient to compensate for the 1 by 2 or 1 by 4 splitter losses.

A newly fabricated monolithic InGaAs active pixel image sensor is presented, and its readout characteristics are described. The sensor is fabricated from InGaAs epitaxially deposited on an InP substrate. It consists of an InGaAs photodiode connected to InP depletion-mode junction field effect transistors (JFETs) for signal buffering, selection and reset. The monolithic sensor eliminates the need for hybridization with a silicon multiplexer, and in addition, allows the sensor to be front illuminated, making it sensitive to visible as well as IR radiation. With further development, the sensor is ideal for dual band (visible/IR) applications, including optical communication. It is also well suited to applications requiring near room temperature, broad band response such as for atmospheric gas sensing and target identification. Two different types of small 4 by 1 test arrays have been fabricated. One is a source follower per detector architecture. Here the signal charge is integrated on the photodiode capacitance. The photodiode is connected to a gate of a JFET configured as a source-follower, which buffers the photodiode voltage. The other test circuit uses a capacitive transimpedance amplifier. This circuit contains an invertor using an input JFET with a passive JFET load. The photodiode is connected to the JFET gate. A feedback capacitor causes the circuit to act as an integrator, while keeping the diode input bias relatively constant. Both circuits also contain JFET switches for reset and selection. Selection connects the output of the chosen cell onto a common output bus. In this exploratory development effort, the effectiveness of these two different readout circuits will be discussed in terms of leakage, operating frequency, and temperature. These results then will guide for the second phase demonstration of integrated two dimensional monolithic active pixel sensor arrays for application in transportable shipboard surveillance, night vision and emission spectroscopy.

Optical fibers offer the wide bandwidth, low losses and low interference required in broadband network applications. Currently, routing the signals to their destination is done by converting the incoming optical signals to an electrical form, carrying out the switching function using electronic circuitry then reconverting to light for the next transmission stage. Recently, we have reported a 3 by 3 optoelectronic switch which combines the functions of conversion and switching. This matrix monolithically integrates metal-semiconductor-metal (MSM) detectors with amplifiers. Very good isolation and crosstalk characterize this switch matrix, but the packaging requires the alignment of nine fibers, the square of the number of inputs, to the various detector crosspoints. In this presentation, we report the fabrication and evaluation of 4 by 4 optoelectronic switching matrices integrating MSM detectors with polyimide waveguides which perform the optical signal distribution on the wafer. These waveguides were fabricated on top of the semiconductor using a photolithographic process. The detector electrodes were formed using a transparent ITO film to maximize the responsivity. The incoming light is distributed using the 'tap' approach, which is more compact than the Y-branching configuration. Two 2 by 4 monolithic arrays were assembled on an alumina circuit using microwave hybrid circuit technology. The bandwidth of the assembled switch exceeds 1 GHz and with improved circuit design, should approach the 5 - 10 GHz bandwidth of the individual MSMs. A similar switch is based on a 4 by 4 monolithic array. The isolation is typically better than 35 dB. These characteristics are compared to the performance of the 3 by 3 OEIC switch and another 4 by 4 switch array assembled using four GaAs MESFET SP4T switches.

Photonic true-time-delay (TTD) lines offer many advantages over their electronic counterparts and attract more and more research efforts. In this paper, we report the demonstration of a 32 TTD lines (5-bit) based on substrate guided wave propagation combined with slanted photopolymer volume phase gratings on a quartz substrate. The system design, device fabrication, optimization of fanout intensity uniformity, and device performance evaluation are addressed as well. The fanout beam intensity uniformity is within plus or minus 10%. The packing density is 2.5 delay lines/cm². The device has a measured bandwidth of up to 2.4 THz and a measured fanout delay step of 50 ps. A 50 GHz optically heterodyned microwave signal is generated, sent through the device, and then detected at the output end.

The description of our designed molecular implementation of two variable logic functions: OR, NOR, AND, NAND and basic elements of molecular cellular automata and neuromolecular networks is performed. The design of the basic elements of the molecular computers is based on the semiempirical quantum chemical calculations of organic photoactive electron donor, electron acceptor and insulator molecules, supermolecule, supramolecules and substituted fullerene molecules.

Silica-based planar lightwave circuit (PLC) devices are starting to be introduced into commercial optical communication systems. PLC devices such as optical splitters, wavelength-insensitive coupler (WINC) arrays, and hybrid- integrated wavelength-division-multiplexing (WDM) transceivers are used to construct cost effective optical access systems. In trunk lines, on the other hand, arrayed-waveguide gratings (AWG) are employed for dense WDM systems to increase the transmission capacity. This paper reviews the current status and recent progress on these practical PLC devices.

Photonic integrated circuits are devices that integrate active photonic elements such as lasers or semiconductor optical amplifiers (SOAs) with passive waveguides made on a single III-V semiconductor substrate. Photonic integrated circuits (PICs) have been used for making several experimental elements which may be useful for WDM systems, such as WDM source lasers as well as optical switches, amplifier arrays and interferometric wavelength converters. Several schemes have been suggested for fabrication of photonic integrated circuits on InP substrates. These fabrication techniques must produce high quality active-passive transitions, and low loss passive waveguides which can generate various passive optical elements such as y-branches, directional couplers, multimode interference (MMI) couplers, and free space splitters- combiners. In this talk we describe processes such as PPro-3 that have been developed for this purpose. This process is based on a new laser configuration that is suitable for photonic integration. The PPro-3 laser has a buried rib waveguide and does not require the deep mesa etch (2 - 3.5 micron) which is typically used with the conventional SIPBH laser configuration. The new process is especially suitable for laser or amplifier arrays. Although the high power-high temperature performance of the SIPBH type lasers are superior, these are not the crucial requirements for temperature controlled laser array sources. There are, however, several advantages that stem from the simplification, better transition, better wavelength control and better planarity of the new PIC process, leading to improved yields and uniformity in array devices.

In this paper some recent advancements in functional guided- wave devices and circuits having high potential impact in the space applications are discussed. Particular attention is addressed to the fields of earth observation, telecommunications, radar surveillance, and remote sensing, since they represent specific fields in the space sector where guided-wave devices can produce a significant improvement.

An optical communication networks can be divided in two levels: communication level, which defines the protocols, the control and the management of the networks and physical level formed by photonic and electronic components in order to transmit and receive the data between different nodes of the network. Traditionally, these two levels are considered separately in the optical communication network design process. This can lead to an erroneous or non-ideal networks implementation, due to the fact that the communication and physical levels are not independent. For example, in WDM communication network the maximum achievable data rate is limited not only by the networks protocol, but depends also on the implementation of the physical level: tuning delay of the optical multiplexers. Also the lack of the possibilities for co-verification of the communication and the physical levels together could lead to misinterpretations between the designers of the different levels and thus induce design faults. Since the prototyping is extremely expensive and time consuming, an integrated simulation of both communication and physical levels is necessary, at least in some extend. In this paper, a behavioral modeling approach that allows a co- simulation of the communication and the physical levels is presented. It is based on the use of a VHDL-AMS-like hardware description language, dedicated to electronic system modeling, but also suitable for modeling and simulation of non- electronic and mixed-domain systems. The behavioral models for photonic and electronic components, as well as the software are integrated in a unique simulator in order to co-simulate the communication (control) and the physical level (data path) of a WDM optical communication network.

A new 2-D true time delay (TTD) generation system architecture for phased array antennas is described. The method uses fiber chirp gratings and acousto-optic beam deflectors. By combining free-space optics and guided optics, the device complexity in conventional TTD systems has been significantly reduced. A proof-of-concept experimental results are demonstrated.