We present an overview of our study on the scientific and technological issues and challenges concerning the miniaturization, interconnection and integration of microphotonic devices, circuits and systems in micron or submicron scale. Devices that we have used for our study include: mode conversion schemes, directional couplers, micro-ring resonator devices, micro-racetrack resonator devices, arrayed waveguide devices, multimode interference devices, microdisk lasers, vertical cavity surface emitting microlasers, and the arrays thereof. Materials used for devices include semiconductors, silica, sol-gel, and polymers. First, in miniaturization, the issues include the size effect, proximity effect, energy confinement effect, microcavitiy effect, optical and quantum interference effect, high field effect, nonlinear effect, noise effect, quantum optical effect, and chaotic effect. Secondly, in interconnection between micro/nano-scale photonic devices, issues to be addressed include optical alignment, minimizing the interconnection losses, and maintaining optical modes. Thirdly, in integration, the issues include the integration of different kinds of devices, active-active, active-passive, passive-active and passive-passive. Scaling is also an important issue both for miniaturization and integration. In the fabrication of devices, the optical quality of the walls, surfaces and interfaces of the devices is of critical importance for high performance function. We found the near-field scanning optical microsope useful in the analysis of the micro/nano-scale photonic devices.

The mid-infrared (3<λ<5 μm) and far-infrared (8<λ<14 μm) spectral regions are presented as a viable, yet underutilized spectral region that would be suitable for mulitple WDM communication applications. Historically limited by the laser source, the quantum cascade laser, thanks to recent advances in performance, is proposed as a realistic source for addressing this spectral region. The current status of the laser technology is reviewed, as well the future outlook for this technology for WDM communication.

Using the finite-difference time-domain calculations, we study whispering-gallery-like modes in photonic crystal air-bridge slab hexagonal defect cavities as good candidates for high quality-factor (Q) and small mode-volume (V) resonant modes. In the hexapole mode of a modified single-defect cavity, structural parameters are optimized to obtain very large Qs of even higher than 2 x 10⁶ with small effective V of the order of cubic wavelength in material, the record value of theoretical Q/V. In addition, the H2-cavity whispering-gallery mode (WGM) is investigated and the defect geometry is modified to increase the Q of the WGM. By symmetrically distributing 12 nearest neighbor holes around the defect and controlling size of holes, it is possible to drastically increase the Q of >10⁵ while preserving effective mode volume of the order of the cubic wavelength in material. We expect the WGMs in photonic crystal cavities are quite promising for low-loss photonic integrated circuit elements and high-efficiency quantum optical devices.

The market for data modulators at 10 Gb/s is currently dominated by Mach-Zehnder phase modulators fabricated in LiNbO₃ (LN). However they are relatively expensive to manufacture and large compared to semiconductor devices. InP based electroabsorption modulators (EAMs), are more compact; however they have a limited bandwidth (5-8 nm) over which chirp is in the correct range to allow >80 km reach. This paper reports the broadband electroabsorption modulator (BEAM) concept in which reach performance in line with LN modulators can be achieved using integrated InP components. The BEAM consists of a series of EAMs, each one tuned to give the correct chirp over a certain wavelength range. The bandwidth of the BEAM can be extended to cover the C-band (1535nm-1565nm). In addition, a semiconductor optical amplifier (SOA) is serially integrated in order to recover the total insertion loss. Details of the design, fabrication and testing of prototype BEAM chips operating at 10 Gb/s are reported. Quantum well intermixing technology is employed to realize the multiple bandgaps required for the prototype chips which are fabricated on semi-insulating InP substrates. Highlights of the operational characteristics of the BEAM chips include extinction ratios of up to 12 dB at 10 Gb/s and SOA gains of 20 dB.

In silicon based photonic circuits, optical modulation is usually performed via the plasma dispersion effect or via the thermo-optic effect, both of which are relatively slow processes. Until relatively recently, the majority of the work in Silicon-on-Insulator (SOI) was based upon waveguides with cross sectional dimensions of several microns. This limits the speed of devices based on the plasma dispersion effect due to the finite transit time of charge carriers, and on the thermo-optic effect due to the volume of the silicon device. Consequently moving to smaller dimensions will increase device speed, as well as providing other advantages of closer packing density, smaller bend radius, and cost effective fabrication. As a result, the trend in recent years has been a move to smaller waveguides, of the order of 1 micron in cross sectional dimensions. In this paper we discuss both the design of small waveguide modulators (of the order of ~1 micron) together with a presentation of preliminary experimental results. In particular two approaches to modulation are discussed, based on injection of free carriers via a p-i-n device, and via thermal modulation of a ring resonator.

A CMOS optical fibre transmitter front-end with an analogue predistortion technique is proposed for reducing laser nonlinearity distortion to achieve broadband linearisation of radio frequency (RF) optical fibre communication systems. The technique uses a nonlinearity having the inverse transfer characteristic of the directly modulated vertical cavity surface emitting laser (VCSEL). The analogue predistortion lineariser comprises a linear and a non-linear transconductance differential amplifier. The combined linear and non-linear transconductance differential amplifiers provide the transfer characteristic to counteract the characteristic of the VCSEL laser diode. The post-layout simulation of the linearity of RF optical fibre systems using the predistortion linearisation technique shows 12dBm improvement. The integrated CMOS optical fibre transmitter circuit with the predistortion lineariser is being implemented using the austriamicrosystems (AMS) 0.35μm CMOS technology.

It has been realized that the lack of enabling technology of beam forming and steering devices significantly slows down the process of implementing wideband phased array antenna systems. In this paper, we present our research in developing an integrated electro-optic switched true-time-delay module as a boradband beam forming device for wideband phased array antennas. The unique feature of our approach is that both the true-time-delay waveguide circuit and electro-optic switching elements are monolithically integrated in a single substrate. As a result, this integration significantly reduces the device size while eliminating the most difficult packaging problem associated with the delicate interfaces between optical fibers and optical switches. Such a monolithic approach offers greater precision for the RF phase control than the fiber-delay-lines thanks to the sub-micrometer accuracy of lithography-defined polymeric waveguides. More important, the proposed optical switched true-time-delay network requires very low electrical power consumption due to the low power soncumption of electrically-switchable waveguide gratings. Furthermore, the electrically-switchable waveguide gratings have a very fast switching speed (<50 μm) that is at least 100 times faster than any existing commercial optical switching matrix. Photonic phased array antenna based on optical true-time delay lines offers improved performance and reduced weight and power consumption over existing parabolic dish antenna presently used for communications.

Crystal ion slicing can fabricate microns-thin-films from bulk, single-crystal metal oxides, which are important materials in optical, microwave, and electrical applications. These thin-films maintain single-crystal properties, which are very difficult to achieve in other thin-film technologies such as epitaxial growth. In this paper, ion-slicing technique is reviewed briefly from a process, material, and device perspective. The demonstrated applications in integrated optics are listed, along with a complete reference to ion-slicing related publications.

III-V semiconductor devices typically use structures grown layer-by-layer and require passivation of sidewalls by vertical etching to reduce leakage current. The passivation is conventionally achieved by sealing the sidewalls using polymer and the polymer needs to be planarized by polymer etch-back method to device top for metal interconnection. It is very challenging to achieve perfect planarization needed for sidewalls of all the device layers including the top layer to be completely sealed. We introduce a novel hard-mask-assisted self-aligned planarization process that allows the polymer in 1-3 μm vicinity of the devices to be planarized perfectly to the top of devices. The hard-mask-assisted process also allows self-aligned via formation for metal interconnection to device top of μm size. The hard mask is removed to expose a very clean device top surface for depositing metals for low ohmic contact resistance metal interconnection. The process is robust because it is insensitive to device height difference, spin-on polymer thickness variation, and polymer etch non-uniformity. We have demonstrated high yield fabrication of monolithically integrated optical switch arrays with mesa diodes and waveguide electroabsorption modulators on InP substrate with yield > 90%, high breakdown voltage of > 15 Volts, and low ohmic contact resistance of 10-20 Ω.

We report on advances in the hybrid organic/inorganic integration of passive and active optical functions. The integration approaches include chip-to-chip attach, flip-chip mounting, and insertion of films in slots formed in planar lightwave circuits. The materials integrated include polymer, silica, silicon, silicon oxynitride, lithium niobate, indium phosphide, gallium arsenide, yttrium iron garnet, and neodymium iron boron. The functions enabled by the hybrid integration approaches span the range of building blocks needed in optical circuitry, while using the highest-performance material system for each function. We demonstrate high-functionality optoelectronic integrated circuits, including fully reconfigurable optical add/drop multiplexers and tunable optical transmitters.

Reconfigurable optical add/drop multiplexer (ROADM) is a next generation critical component that facilitates the network system evolution from a point-to-point transmission-oriented structure to an all-optical, wavelength-flexible, dynamic network. ROADM enables flexible removal and insertion of WDM channels at either a head-end or intermediate nodes-making it possible for true network provisioning and reconfiguration. We will review the pros and cons of each of the techniques for tunable OADM, from their operating principles to their practical implementations with special emphasis on two approaches: a TTF based three-port tunable filter as a basic one-channel tunable add/drop multipexer and a full-scale, MEMS mirror array based 80-channel reconfigurable OADM subsystem. Comparative laboratory experimental results with theoretical calculations are presented.

Planar technology and design have evolved significantly in the past decade, both in terms of performance and yield, reducing the cost/performance advantage of thin-film filters (TFF) over Array-Waveguide Grating (AWG) devices. This evolution is primarily due to two reasons. One of the reasons for this is the adoption of the latest in semi-conductor fabrication techniques with respect to wafer scale, process equipment automation, and yield engineering. The other reason is the many advancements made in the Planar Light Circuit (PLC) design front which have resulted in lower optical insertion loss, reduced crosstalk, increased channel bandwidth, decreased channel spacing, and minimal chromatic dispersion. We demonstrate here how such state-of-the-art fabrication technology in combination with advanced PLC designs can be effectively used to engineer the filter shape (ripple, bandwidth, and flatness) and chromatic dispersion of AWG's to match or exceed that of their thin-film counterparts. Low passband ripple is critical for cascading multiple nodes in ring network architecture whereas minimal chromatic dispersion (CD) is desired in high rate data systems to avoid signal distortion. The AWG device presented here has a 1dB bandwidth that exceeds 80% of the channel spacing awhile exhibiting a high flatness (25dB/1dB ratio < 1.7), both of which are at least a 50% improvement over generic flat-top AWG designs available in the market and are equivalent in performance to TFF devices. At 100 GHz spacing, AWG's have intrinsic low-dispersion, but narrowing the spacing to 50GHz leads to a four fold increase in the CD. Here, we have successfully overcome this limitation and have been able to design and fabricate a 50GHz wide-band AWG with less than 1ps/nm chromatic dispersion, which exceeds TFF performance.

We present a photonic random access memory (RAM) that can write and read high-speed asynchronous burst optical packets freely by specifying addresses. The photonic RAM consists of an optical clock-pulse generator, an all-optical serial-to-parallel converter, a photonic parallel-to-serial converter, all developed by us, and a CMOS RAM as a storage medium. Unlike conventional optical buffers, which merely function as optical delay lines, the photonic RAM provides various advantages, such as compactness, large capacity, long-term storage, and random access at an arbitrary timing for ultrafast asynchronous burst optical packets. We experimentally confirm its basic operation for 40-Gbit/s 16-bit optical packets.

Digital imaging detectors are growing larger in silicon area and pixel count, which increases fabrication time defects, reducing yield, hence increasing costs and limiting area. In harsh environments, like high radiation conditions, what used to work might fail with time. Fault tolerant Active Pixel Sensors have been created by splitting the photodiode and readout transistors into two parallel operating halves with only a small area cost. These offer standard operation normally, but produce a recoverable image of half illumination sensitivity for single defects. The single-defect case can be compensated by a multiplication of two, whereas the double-defect case is much less likely but can be corrected via software. This paper presents the experimental and simulation results obtained from the fault-tolerant APS' fabricated in CMOS 0.18-micron technology, disregarding the effects of interpolation. Test results suggest that after compensation, the percentage differences between the normally operating fault tolerant APS and the single-defect optically stuck-high and stuck-low cases are 0.5% and 1.5% respectively, which falls within experimental errors. Combining these fault tolerant APS' with a software interpolation technique results in a system where initial simulations show the production of almost defect free images under error conditions with hundreds of dead pixels.

A test setup for static and dynamic measurements of a centroid circuit is presented in this paper. Centroid computations are used in wavefront sensing, motion tracking and in fiber optic sensors. Pixel arrays were fabricated using CMOS photodetectors and analog circuitry was designed to compute the centroid of an image. Generally, for testing the functionality and performance characteristics of pixel arrays and readouts built on an integrated circuit (IC), high precision optical setup and test equipment are required. This cost presents a limitation in testing optoelectronic integrated circuits. This paper provides a novel approach to building a simple test setup using optical lenses, filters/attenuators, a He-Ne laser, an IC probe station and other low-cost equipment present in an IC testing facility. Using this test setup the centroid of an image incident on a pixel array was measured and the performance of the centroid computation circuitry was also verified. The setup includes simple schemes to change the spot size of the laser beam incident on the pixel array. This provides the flexibility to test pixel arrays of different sizes with the same setup. Another scheme to pulse the laser beam externally using a chopper is also described. This scheme can be used for measurements involving time varying inputs. The test setup was used to successfully characterize the static and dynamic characteristics of a fully integrated CMOS centroid computation circuit.

For digital versatile disk (DVD) applications, amplifiers with high bandwidth and high sensitivity in the red spectral range are required. The presented optoelectronic integrated circuit (OEIC) achieves a bandwidth of 265MHz and a transimpedance of 210kΩ due to an advanced feedback network. This is an improvement by a factor of 4 compared to the same amplifier with a simple feedback resistor.