This PDF file contains the front matter associated with SPIE Proceedings Volume 7719, including the Title Page, Copyright information, Table of Contents, and the Conference Committee listing.

Silicon optical modulators have generated an increasing interest in the recent years, as their performances are crucial to achieve high speed optical links. Among possibilities to achieve optical modulation in silicon-based materials, index variation by free carrier concentration variation has demonstrated good potentiality. High speed and low loss silicon modulators can be obtained by carrier depletion inside lateral PN or PIPIN diodes. When the diode is reverse biased, refractive index variations are obtained and then phase modulation of the guided wave is obtained. Mach-Zehnder interferometers are used to convert phase modulation into intensity modulation. Experimental results are presented for both PN and PIPIN diodes.

An intriguing optical property of silicon is that it exhibits a large third-order optical nonlinearity, with orders-ofmagnitude larger than that of silica glass in the telecommunication band. This allows efficient nonlinear optical interaction at relatively low power levels in a small footprint. Indeed, we have witnessed a stunning progress in harnessing the Raman and Kerr effects in silicon as the mechanisms for enabling chip-scale optical amplification, lasing, and wavelength conversion - functions that until recently were perceived to be beyond the reach of silicon. With all the continuous efforts developing novel techniques, nonlinear silicon photonics is expected to be able to reach even beyond the prior achievements. Instead of providing a comprehensive overview of this field, this manuscript highlights a number of new branches of nonlinear silicon photonics, which have not been fully recognized in the past. In particular, they are two-photon photovoltaic effect, mid-wave infrared (MWIR) silicon photonics, broadband Raman effects, inverse Raman scattering, and periodically-poled silicon (PePSi). These novel effects and techniques could create a new paradigm for silicon photonics and extend its utility beyond the traditionally anticipated applications.

Silicon photonics have generated an increasing interest in the recent year, mainly for optical telecommunications or for optical interconnects in microelectronic circuits. The rationale of silicon photonics is the reduction of the cost of photonic systems through the integration of photonic components and an IC on a common chip, or in the longer term, the enhancement of IC performance with the introduction of optics inside a high performance chip. In order to build a Opto-Electronic Integrated circuit (OEIC), a large European project HELIOS has been launched two years ago. The objective is to combine a photonic layer with a CMOS circuit by different innovative means, using microelectronics fabrication processes. High performance generic building blocks that can be used for a broad range of applications are developed such as WDM sources by III-V/Si heterogeneous integration, fast Si modulators and Ge or InGaAs detectors, Si passive circuits and specific packaging. Different scenari for integrating photonic with an electronic chip and the recent advances on the building blocks of the Helios project are presented.

HISTORIC aims to develop and test complex photonic integrated circuits containing a relatively large number of digital photonic elements for use in e.g. all-optical packet switching. These photonic digital units are alloptical flip-flops based on ultra compact laser diodes, such as microdisk lasers and photonic crystal lasers. These lasers are fabricated making use of the heterogeneous integration of InP membranes on top of silicon on insulator (SOI) passive optical circuits. The very small dimensions of the lasers are, at least for some approaches, possible because of the high index contrast of the InP membranes and by making use of the extreme accuracy of CMOS processing. All-optical flip-flops based on heterogeneously integrated microdisk lasers with diameter of 7.5μm have already been demonstrated. They operate with a CW power consumption of a few mW and can switch in 60ps with switching energies as low as 1.8 fJ. Their operation as all-optical gate has also been demonstrated. Work is also on-going to fabricate heterogeneously integrated photonic crystal lasers and all-optical flip-flops based on such lasers. A lot of attention is given to the electrical pumping of the membrane InP-based photonic crystal lasers and to the coupling to SOI wire waveguides. Optically pumped photonic crystal lasers coupled to SOI wires have been demonstrated already. The all-optical flip-flops and gates will be combined into more complex photonic integrated circuits, implementing all-optical shift registers, D flip-flops, and other all-optical switching building blocks. The possibility to integrate a large number of photonic digital units together, but also to integrate them with compact passive optical routers such as AWGs, opens new perspectives for the design of integrated optical processors or optical buffers. The project therefore also focuses on designing new architectures for such optical processing or buffer chips.

The European BOOM project aims at the realization of high-capacity photonic routers using the silicon material as the base for functional and cost-effective integration. Here we present the design, fabrication and testing of the first BOOMgeneration of hybrid integrated silicon photonic devices that implement key photonic routing functionalities. Ultra-fast all-optical wavelength converters and micro-ring resonator UDWDM label photodetectors are realized using either 4um SOI rib or SOI nanowire boards. For the realization of these devices, flip-chip compatible non-linear SOAs and evanescent PIN detectors have been designed and fabricated. These active components are integrated on the SOI boards using high precision flip-chip mounting and heterogeneous InP-to-silicon integration techniques. This type of scalable and cost-effective silicon-based component fabrication opens up the possibility for the realization of chip-scale, power efficient, Tb/s capacity photonic routers.

The project is a consortium based activity involving researchers from the UK institutions of the Universities of Surrey, St. Andrews, Leeds, Warwick, and Southampton, as well as the commercial research institution QinetiQ. The aims of the project are to progress the state of the art in Silicon Photonics, in the areas of waveguides, modulators, couplers, detectors, Raman processes, and integration with electronics. Thus the field is vast, and impossible to cover comprehensively in one project, nor indeed in one paper. The programme is run on a truly collaborative basis, with members from each institution running one or more work packages within the project, each co-ordinating work from their own plus other institutions. To date, the most well developed work has emerged from the activity on basic waveguides and their characteristics, the modulator activity, optical filters, and work on Raman Amplifiers. This work will be the main focus of this paper, but an attempt will be made to update the audience on the remaining activities within the project. By the nature of the project, much of the work is medium term, and hence some activities are not expected to yield viable results until at least next year, hence the concentration on some activities rather than all activities at this stage.

We review the use of planar integrated optical waveguide ring resonators for label free bio-sensing and present recent results from two European biosensor collaborations: SABIO and InTopSens. Planar waveguide ring resonators are attractive for label-free biosensing due to their small footprint, high Q-factors, and compatibility with on-chip optics and microfluidics. This enables integrated sensor arrays for compact labs-on-chip. One application of label-free sensor arrays is for point-of-care medical diagnostics. Bringing such powerful tools to the single medical practitioner is an important step towards personalized medicine, but requires addressing a number of issues: improving limit of detection, managing the influence of temperature, parallelization of the measurement for higher throughput and on-chip referencing, efficient light-coupling strategies to simplify alignment, and packaging of the optical chip and integration with microfluidics. From the SABIO project we report refractive index measurement and label-free biosensing in an 8-channel slotwaveguide ring resonator sensor array, within a compact cartridge with integrated microfluidics. The sensors show a volume sensing detection limit of 5 x 10^-6 RIU and a surface sensing detection limit of 0.9 pg/mm². From the InTopSens project we report early results on silicon-on-insulator racetrack resonators.

Mode-evolution-based polarization rotators in silicon waveguides were studied. The rotator's performance was studied under normal and abnormal launching conditions. The rotator with minimum length of 40μm was demonstrated to provide the polarization rotation with polarization extinction ratio of 15dB at abnormal launching condition. The insertion loss at the transition region was less than 1dB.

This paper introduces a compact 90º optical hybrid, built on small size SOI waveguide technology .This optical hybrid is a critical component of a potentially low-cost coherent optical receiver design developed within the frame of our Optical Coherent Transmission for Access Network Extensions (OCTANE) project. In previous recent work, 90º optical hybrids were realized in SOI rib waveguide technology with 4 μm top silicon and a rib height of approximately 2 μm. In this paper, we introduce a compact 90º optical hybrid, built on small size SOI waveguide technology (1.5 μm SOI -based rib waveguide, with 0.8μm rib height). The proposed device consists of multimode interferometers (MMIs) connected in such a way that four different vector additions of a reference signal (local oscillator) and the signal to be detected are obtained. At the outputs, the hybrid provides four linear combination of the signal with the reference which differs by a relative phase shift of the reference of 90º. The four output signals are detected by a pair of balanced receivers to provide in-phase and quadrature (I&Q) channels. The phase differences arise naturally from the self imaging property of a MMI. The key elements of the 90º optical hybrid, including a 2×2 MMI, a 4×4 MMI, and polarization diversity configuration have been designed and simulated, using the numerical mode solving tool FIMMPROB. The 2×2 and 4×4 MMI had overall lengths of 701μm and 3712.5μm lengths respectively. Tapers are used to couple adiabatically single mode waveguides to the entrance and exit ports of the MMI to assure correct operation by avoiding coupling to the higher order transverse modes allowed at the entrance and exit ports of the MMI. The simulation results at 1550nm show polarization independence and phase errors between the ports of less than 0.03 degrees. Currently the design is in fabrication at the Canadian Photonics Fabrication Center with the support of CMC Microsystems and experimental results will be subject to a further report.

In this paper, we demonstrate a compact 8x8 λ-router using multimode-interference (MMI) crossing based on the microring resonator. The 8x8 λ-router was designed and fabricated with a CMOS compatible silicon on insulator technology. MMI is used to reduce the cross talk and the crossing losses of the device. Microrings with a nominal radius of 2.5 μm and small variations of 10 nm of the nominal value allow respectively a free spectral range of 32 nm and spacing between channels of 4 nm. The experimental results are in good agreement with the modeling. The basic add drop filters of the devices exhibit losses of -2 dB and on/off contrast of the resonance of 20 dB. The total losses for one channel are about -4 dB and the imbalance between the 8 channels is lower than 2 dB.

Slab waveguides were fabricated in Er-doped tungsten-tellurite glass and CaF₂ crystal samples via ion implantation. Waveguides were fabricated by implantation of MeV energy N⁺ ions at the Van de Graaff accelerator of the Research Institute for Particle and Nuclear Physics, Budapest, Hungary. Part of the samples was annealed. Implantations were carried out at energies of 1.5 MeV (tungsten-tellurite glass) and 3.5 MeV (CaF₂). The implanted doses were between 5 x 10¹² and 8 x 10¹⁶ ions/cm². Refractive index profile of the waveguides was measured using SOPRA ES4G and Woollam M-2000DI spectroscopic ellipsometers at the Research Institute for Technical Physics and Materials Science, Budapest. Functionality of the waveguides was tested using a home-made instrument (COMPASSO), based on m-line spectroscopy and prism coupling technique, which was developed at the Materials and Photonics Devices Laboratory (MDF Lab.) of the Institute of Applied Physics in Sesto Fiorentino, Italy. Results of both types of measurements were compared to depth distributions of nuclear damage in the samples, calculated by SRIM 2007 code. Thicknesses of the guiding layer and of the implanted barrier obtained by spectroscopic ellipsometry correspond well to SRIM simulations. Irradiationinduced refractive index modulation saturated around a dose of 8 x 10¹⁶ ions/cm² in tungsten-tellurite glass. Annealing of the implanted waveguides resulted in a reduction of the propagation loss, but also reduced the number of supported guiding modes at the lower doses. We report on the first working waveguides fabricated in an alkali earth halide crystal implanted by MeV energy medium-mass ions.

Because of the its indirect bandgap structure, it is a huge challenge to establish an efficient Si light emitting diode (LED) compatible with complementary metal-oxide-semiconductor (CMOS) process. In this paper, we provide an alternative route to overcome this difficulty based on the unique property of photonic crystals (PhC). A vertical-current-injection LED based on three-dimensional-confined structures with triangular-lattice air-hole PhC patterns has been fabricated with enhanced light extraction from the active region (i.e., silicon-rich-oxide/SiO₂ multilayer stack). The intensity and profile of photoluminescence (PL) and electroluminescence (EL) has been found to be efficiently modulated by controlling the optical modes of the periodic arrays via varying their structural parameters. It provides a convenient way of redistributing the light energy in desired form and orientation. With optimized lattice constant/radius ratio, significant enhancement up to ~7 times in both PL and EL emissions can be obtained. The mechanisms for different enhancement features have also been theoretically analyzed based on coherent scattering and quantum electrodynamics effects, which is well consistent with the experiment observation.

Silicon-based light emitting device is the missing piece in the design of complete optoelectronic circuits on silicon. A complete electrical and optical characterization of electroluminescent silicon-nanocrystals based devices is presented. This characterization is the first step in the design of coupling structures with optimal injection of light into optical waveguides, which will allow the development of all-silicon photonic circuits. In this paper, a novel coupling structure based on rectangular surface grating has been studied with promising results, increasing the coupling efficiency to SU-8 waveguides up to 25 times. The SU-8 photoresist is fully compatible with silicon technology and can be used to define waveguides as well as microfluidics channels. These properties are interesting for the final application of the present study, which is to obtain a Lab-on-a-chip device, integrating all the optical elements, control electronics and microfluidics channels.

Electroluminescent properties of thin silicon-rich oxide (SRO) films deposited by low pressure chemical vapor deposition (LPCVD) were studied. The gas flow ratio Ro = N₂O/SiH₄ was changed to obtain different silicon concentrations within the SRO films. After deposition, SRO films were thermally annealed at 1100ºC for 3h in N₂ atmosphere in order to create silicon nanoparticles (Si-nps). Simple capacitive structures like Polysilicon/SRO/n-Si were used for the study. These light emitting capacitors (LECs) show intense blue (~466) and red EL (~685) at room temperature depending on the silicon excess within the SRO films. Electroluminescence in these LECs is obtained at direct current (DC) at both forward and reverse bias conditions. Nevertheless, a stronger whole area EL is obtained when devices are forwardly biased.

There is an intense interest on integration of III-V materials on silicon and silicon-on-insulator for realisation of optical interconnects, optical networking, imaging and disposable photonics for medical applications. Advances in photonic materials, structures and technologies are the main ingredients of this pursuit. We investigate nano epitaxial lateral overgrowth (NELOG) of InP material from the nano openings on a seed layer on the silicon wafer, by hydride vapour phase epitaxy (HVPE). The grown layers were analysed by cathodoluminescence (CL) in situ a scanning electron microscope, time-resolved photoluminescence (TR-PL), and atomic force microscope (AFM). The quality of the layers depends on the growth parameters such as the V/III ratio, growth temperature, and layer thickness. CL measurements reveal that the dislocation density can be as low as 2 - 3·10⁷ cm^-2 for a layer thickness of ~6 μm. For comparison, the seed layer had a dislocation density of ~1·10⁹ cm^-2. Since the dislocation density estimated on theoretical grounds from TRPL measurements is of the same order of magnitude both for NELOG InP on Si and on InP substrate, the dislocation generation appears to be process related or coalescence related. Pertinent issues for improving the quality of the grown InP on silicon are avoiding damage in the openings due to plasma etching, pattern design to facilitate coalescence with minimum defects and choice of mask material compatible with InP to reduce thermal mismatch.

In this paper the realization and the characterization of a resonant cavity enhanced photodetector (RCE), completely silicon compatible and working at 1.55 micron, is reported. The detector is a RCE structure incorporating a Schottky diode and its working principle is based on the internal photoemission effect. In order to obtain a fabrication process completely compatible with standard CMOS silicon technology, a photodetector having copper (Cu) as Schottky metal has been realized. Performances devices in terms of responsivity, free spectral range, finesse are reported.

In this paper we present the integration of an InP-based photodetector with silicon-on-insulator (SOI) waveguides using thermocompression bonding. A BCB prism integrated on top of the light-sensitive area of a planar detector (PD) chip deflects the light from a 4 μm thick SOI waveguide upward into the flip-chip bonded PD. A trench is etched in front of the SOI waveguide to accommodate prisms with apexes up to 7 μm. Using thermocompression bonding between thin gold pads (~500 nm thickness) deposited on both, SOI and photodetector chips an excellent vertical alignment accuracy of ±100 nm can be achieved, limited only by etching and Au-deposition tolerances. A commercial flip-chip bonder provides a lateral alignment accuracy also in the sub-micron range. Together with a previously developed process for integrating lasers and SOA chips using the same technology, fully functional PICs can now be realized on the SOI platform using thermocompression bonding.

In this paper, we report our design and fabrication approach towards realizing a monolithic integration of Ge photodetector and Si CMOS circuits on common SOI platform for integrated photonic applications. The approach, based on the Ge-on-SOI technology, enables the realization of high sensitivity and low noise photodetector that is capable of performing efficient optical-to-electrical encoding in the near-infrared wavelengths regime. When operated at a bias of -1.0V, a vertical PIN detector achieved a lower I_dark of ~0.57μA as compared to a lateral PIN detector, a value that is below the typical ~1μA upper limit acceptable for high speed receiver design. Very high responsivity of ~0.92A/W was obtained in both detector designs for a wavelength of 1550nm, which corresponds to a quantum efficiency of ~73%. Impulse response measurements showed that a vertical PIN photodetector gives rise to a smaller FWHM of ~24.4ps, which corresponds to a -3dB bandwidth of ~11.3GHz where RC time delay is known to be the dominant factor limiting the speed performance. Eye patterns (PRBS 2⁷-1) measurement further confirms the achievement of high speed and low noise photodetection at a bit-rate of 8.5Gb/s. In addition, we evaluate the DC characteristics of the monolithically fabricated Si CMOS inverter circuit. Excellent transfer and output characteristics were achieved by the integrated CMOS inverter circuits in addition to the well behaved logic functions. We also assess the impact of the additional thermal budget introduced by the Ge epitaxy growth on the threshold voltage variation of the short channel CMOS transistors and discuss the issues and potential for the seamless integration of electronic and photonic integrated circuits.

We report on monolithically integrated PIN photodiodes whose responsivity values could be significantly enhanced over the whole spectral range by the implementation of a Bottom Antireflective Coating (BARC) process module into austriamicrosystems 0.35μm CMOS as well as high-speed SiGe BiCMOS technologies. The resulting photodiodes achieve excellent responsivities together with low capacitances and high bandwidths. We processed finger-photodiodes with interdigitated n+ cathodes, which are especially sensitive at low wavelengths, and photodiodes with full area n+ cathodes on very lightly p-doped start material. We present a method of depositing an antireflective layer directly upon the Si surface of the photodiode by changing the standard process flow as little as possible. With just one additional mask alignment and a well controlled etch procedure we manage to remove the thick intermetal oxide and passivation nitride stack over the photodiodes completely without damaging the Si surface. The following deposition of a CVD Silicon Nitride BARC layer not only minimizes the reflected fraction of the optical power but also acts as passivation layer for the photodiodes. Another benefit of BARC processing is the fact that in-wafer and wafer-to-wafer quantum efficiency variations can be dramatically reduced. In our experiments we deposited BARC layers of different thicknesses that were optimised for violet, red and infrared light. Responsivity measurements resulted in values as high as R=0.27A/W at λ=410nm, R=0.53A/W at λ=670nm and R=0.5A/W at λ=840nm.

The conventional streak camera (CSC) is an optoelectronic instrument which captures the spatial distribution versus time of a ultra high-speed luminous phenomena with a picosecond temporal resolution and a typical spatial resolution of 60 μm. This paper presents two Integrated Streak Camera (ISC) architectures called MISC (M for Matrix) and VISC (V for Vector) which replicate the functionality of a streak camera on a single CMOS chip. The MISC structure consists of a pixel array, where the column depth together with the sampling rate determine the observation window. For proper operation, the image of the slit has to be spread uniformly over the rows of the imager. The VISC architecture is based on a single column of photosensors, where each element is coupled to a front-end and a multi-sampling and storage unit. The observation window is determined by the sampling rate and the depth of the memory frame. The measurement of a 6 ns FWHM 532 nm light pulse laser is reported for both ISCs. For the two architectures, the spatial resolution is linked to the size and the number of the photodetectors.

Nonlinear silicon photonics has been an immense research subject in the past several years with promising prospects of delivering chip scale signal modulation, shaping and characterization tools. In particular, broadband parametric process has been considered for applications ranging from wideband light amplifiers to signal characterization and signal shaping tools. Although underlying nonlinear effect, Kerr phenomena, in silicon has generated promising result of wavelength conversion, the success of these devices have been challenged by the presence of nonlinear losses such as two photon absorption and the two photon generated free carrier absorption. Experimental demonstrations were limited to conversion efficiencies below -10dB. Here, we present the prospect of ultra wide discrete band conversion schemes and the prospect of parametric process at mid-infrared wavelengths where nonlinear losses are not present. In particular, we explore the parametric wavelength conversion scheme at mid-wave infrared wavelength (2μm~6μm) by four-wavefixing process in silicon waveguides with new cladding materials, such as sapphire, that can provide transparency up to 6μm and facilitate phase matching condition for discrete wavelength bands as far as 60THz away from each other. Design criteria include the optimization of mode overlap integrals and dispersion engineering for an ultra-wide band signals. The particular results of wavelength conversion between 2μm bands and 5μm bands, and between 1.8μm bands and >4μm bands will be presented. Prospects of frequency band conversion in generation of new infrared signals and low noise, room temperature detection of mid-infrared signals will also be discussed.

Silicon-based semiconductors offer optically low-loss and high-thermal-conducting lattice for the broad-band terahertz active media that can be used in the range of 5-7 THz. We report on realization of the terahertz-range stimulated emission from monocrystalline natural and isotopically enriched silicon crystals doped by group-V donor centers due to nonlinear frequency conversion. Lasing in the frequency bands of 1.2 - 1.8 THz; 2.5 - 3.4 THz has been achieved from silicon crystals doped by phosphorus and in the frequency band of 4.6 - 6.4 THz from different donors under optical pumping by radiation of mid-infrared free electron laser at cryogenic temperatures. Analysis of the data shows that the emission in high-frequency band corresponds to electronic Stokes-shifted Raman-type lasing. The low-frequency bands indicate on high-order nonlinear frequency conversion processes similar to four-wave mixing accompanied by highenergy intervalley g-phonons and f-phonons of host lattice. These lasers supplement terahertz silicon lasers operating on transitions between donor states.

We propose a silicon ring Raman converter in which the spatial variation of the Raman gain along the ring for TE polarization is used to quasi-phase-match the CARS process. If in addition the pump, Stokes, and anti-Stokes waves involved in the CARS interaction are resonantly enhanced by the ring structure, the Stokes-to-anti-Stokes conversion efficiency can be increased by at least four orders of magnitude over that of one-dimensional perfectly phase-matched silicon Raman converters, and can reach values larger than unity with relatively low input pump intensities. These improvements in conversion performance could substantially expand the practical applicability of the CARS process for optical wavelength conversion.

Strained silicon is a versatile new type of material, which has found application in microelectronics and integrated optics. The applied strain alters the electronic and optical properties and gives rise to new properties previously not known to exist in silicon, like a bulk second order nonlinear susceptibility. Here, we determine experimentally the strain dependence of the second order nonlinear susceptibility on the applied strain. To this purpose, the strain induced second harmonic signal generated in the silicon was measured in a reflection geometry with azimuthal angle dependence. The extracted components of the second order nonlinear susceptibility were determined and compared to the unstrained case. Additionally the measurements were compared to results obtained with an analytical model, that takes into account the exponential strain decay at the sample surface. The predicted linear dependence between the surface strain and the second order nonlinear susceptibility agrees well with the results of our experimental work.

Racetrack resonators based on the silicon-on-insulator platform are proposed for electro-optical modulation. The resonators are functionalized by a cladding of a second order nonlinear optical polymer. Two different concepts for the racetrack design employing different waveguide geometries for quasi-TE and quasi-TM polarization operation are presented. In both resonator designs the electrical contact is established by fully etched segmented electrode sections to allow for an easy fabrication process. For quasi-TM polarization the width of the strip waveguide is optimized to 400 nm. The Q factor of 2000 is measured for a sample with segmented electrode. A loss of 0.4 dB per segmented waveguide is deducted. For the quasi-TE polarization the slot waveguide geometry is optimized to 470 nm total width including a vertical slot of 90 nm width. Only the straight parts of the racetrack are slotted, while the bends are built from strip waveguides. To convert the mode from strip to slot geometry stub like couplers of 100 nm length are employed. The measured Q factor is 550. The in device Pockels coefficient is measured to r₃₃ = 1 pm/V. This small value indicates a very low poling induced polar order which needs to be improved. This is a topic of current investigation.

A new optimization method is described and performed on high-speed silicon optical modulators based on carrier depletion in a p-i-n junction. Quantitative results on the geometry of the waveguide and doping concentrations of the pand n-doped regions are presented at the end of the optimization. General rules can thus be applied to design high performances optical modulators in term of modulation efficiency and insertion loss. Complete electro-optical simulations have been performed on optimal designs to evaluate the corresponding Figures of Merit and theoretical limits on performances have been exhibited. V_πL_π as low as 1.25 V.cm has been obtained at best for the p-n configuration, for a bias of 5 V and for a rib height of 400 nm. A strong dependence of the total optical loss with the geometry of the waveguide has also been demonstrated with an optimal value of 3 dB.

Tunable coupled resonator optical waveguides (CROWs) are powerful and versatile devices that can be used to dynamically control the delay of optical data streams on chip. In this contribution we show that CROW delay lines fabricated on a silicon on insulator (SOI) platform are suitable for applications in the emerging scenario of optical systems at 100 Gbit/s. Issues concerning technology, design, limits and applications of SOI CROWs are discussed. The performances of silicon CROW delay lines activated by thermal tuning are compared to those of glass CROW in terms of power consumption, thermal crosstalk and reconfiguration speed. The continuous delay of 10-ps long optical pulses by 8 bit length is demonstrated by using a silicon CROW with a bandwidth of 87 GHz and made of 12 RRs. At 100 Gbit/s this structure provides comparable figures of merit (fractional delay of 0.75 bit/RR and fractional loss of 0.7 dB per bit-delay) of state-of-the art glass CROW operating at 10 Gbit/s, yet the area of the latter being three order of magnitude larger. The compatibility of silicon CROW with the emerging 100 Gbit/s systems is demonstrated by showing error-free phase-preserving propagation of a 100 Gbit/s return-to-zero (RZ) polarization-division-multiplexing (PolDM) differential quaternary phase shit keying (DQPSK) signal dynamically delayed by the CROW. It is also demonstrated that a silicon CROW can be used in a PolDM system to introduce a polarization selective delay in order to optimize the time interleaving of the two orthogonally polarized data streams.

A simple configuration for achieving a radio frequency transparent 90° hybrid, for broadband QAM wireless systems using silicon photonics is proposed. The device consists of a high Q ring resonator which induces an optical 90° phase shift between two adjacent resonant wavelengths. When these optical carriers are modulated by an RF carrier the resulting device behaves as an RF 90° hybrid. Numerical simulations of the phase shift were performed on a 40 GHz carrier, and to demonstrate the frequency transparency phase shift simulations was also performed at a carrier frequency of 60 GHz. One of the main applications of such a device is the generation of millimeter wave 10 Gb/s wireless based on quadrature amplitude modulation.

There is little doubt that the most important limiting factors of the performance of next-generation Chip Multiprocessors (CMPs) will be the power efficiency and the available communication speed between cores. Photonic Networks-on-Chip (NoCs) have been suggested as a viable route to relieve the off- and on-chip interconnection bottleneck. Low-loss integrated optical waveguides can transport very high-speed data signals over longer distances as compared to on-chip electrical signaling. In addition, with the development of silicon microrings, photonic switches can be integrated to route signals in a data-transparent way. Although several photonic NoC proposals exist, their use is often limited to the communication of large data messages due to a relatively long set-up time of the photonic channels. In this work, we evaluate a reconfigurable photonic NoC in which the topology is adapted automatically (on a microsecond scale) to the evolving traffic situation by use of silicon microrings. To evaluate this system's performance, the proposed architecture has been implemented in a detailed full-system cycle-accurate simulator which is capable of generating realistic workloads and traffic patterns. In addition, a model was developed to estimate the power consumption of the full interconnection network which was compared with other photonic and electrical NoC solutions. We find that our proposed network architecture significantly lowers the average memory access latency (35% reduction) while only generating a modest increase in power consumption (20%), compared to a conventional concentrated mesh electrical signaling approach. When comparing our solution to high-speed circuit-switched photonic NoCs, long photonic channel set-up times can be tolerated which makes our approach directly applicable to current shared-memory CMPs.

The conventional scheme of optical pick-up unit (OPU) should require two or three optoelectronic integrated circuits (OEICs) to cover triple-wavelength λ =780nm, 650nm and 405nm). In order to reduce cost and waste of resources, onechip solution of the OEIC is required. In this paper, the OEIC is designed which can cover triple-wavelength and three optical storage standards which are compact disk (CD), digital versatile disk (DVD) and Blue-Ray. The OEIC has dualarrays of photodiodes because focus of laser is varied depending on wavelength. One of arrays senses the laser of λ =780nm and another senses the lasers of λ =650nm and λ =405nm. For low power consumption and small die area, one wideband transimpedance amplifier (TIA) is used for two photodiodes which are for CD and DVD or Blue-Ray, respectively. And two small size switches are included to select photodiodes. The PIN fingerdiode with N+ fingercathode is integrated to guarantee high performances for λ =405nm and 650nm. And the isolation area between adjacent photodiodes is made by floated P+ implant for reducing power-loss. The measured cutoff bandwidth of the OEIC is 210MHz for λ =405nm. The OEIC is fabricated in a 0.6- μm BiCMOS technology and dissipates 150mW for a single supply voltage of 5V. The active area is 1.4x1.2mm².

In this paper, we presented a high performance monolithic Si DWDM receiver comprising a 1×32 Si-based AWG filter and a high speed waveguided Ge-on-Si photodetectors array on silicon-on-insulator platform. The Si-based AWG has 200 GHz channel spacing and its optical adjacent crosstalk performance is more than 18 dB. Each Ge-on-Si photodetector has 10 GHz bandwidth; and can transmit at least 10 Gbps data rate. So, the aggregated data rate of the DWDM receiver is at least 320 GHz. At a BER of 1 × 10^-11, the DWDM receiver showed an optical input sensitivity between -16 dBm and -19 dBm for all 32 channels in Lband. This first demonstration indicates the feasibility and potential of manufacturing low cost silicon DWDM receivers for terabit data communications. The size of entire receiver is 1.5×1.0 mm².

When the cross-section of an optical waveguide is much smaller than the operating wavelength, unique materials and structural dependent properties can be observed. In this regard silicon has been particularly attractive as the low-cost and mature CMOS fabrication technology widely used in the electronics industry can be exploited. The high index contrast of silicon allows light confinement in submicron size waveguides, along with the creation of very compact bends, to allow increased functionality of photonic integrated circuits. A rigorously H-field based vectorial modal analysis has been carried out, which can more accurately characterize the abrupt dielectric discontinuity of a high index contrast optical waveguide. As a result, the full-vectorial H and E-field and the Poynting vector profiles are shown in detail. The work done and reported reveals that the mode profile of a circular silicon nanowire is not circular and also has a strong axial field component. Arising from the results of the analysis, the characteristics of single mode operation, the vector field profiles, the modal ellipticity and the group velocity dispersion of this silicon nanowire both circular and planar are presented. The modal hybridness and birefringence of rectangular silicon nanowires and slot-type waveguides are also presented.

We report on the formation of spatially localized crystals in SiO₂-SnO₂ thin films fabricated by the sol-gel technique. This material presents an intense absorption band (α≈10³cm^-1) in the UV region. A continuous wave UV laser operating at 266nm focused through a microscope objective is used as an effective tool to modify locally the matrix containing photorefractive SnO₂. The UV micro-Raman spectrometer is used to study the evolution of SnO₂ crystals in the thin film. The appearance of the Raman scattering peak at 621cm^-1, assigned to the A_1g mode of rutile SnO₂, confirms the formation of nanocrystals in the focalised UV irradiated zone.

In this paper we present the performances of a modulator, realized by proton exchange, achieving a moderate rejection ratio in the K-band (2.2μm). The device consists on a simple Mach-Zehnder beam combiner, developed by our partners from Photline Technologies®, pushing forward their proton exchange technique in order to achieve single mode optical guiding above 1.9μm in Lithium Niobate X-cut substrates. Applying low modulation voltages (V_π=3.4V), and by a Fast Fourier Transform obtain the spectrum of the source with a moderate resolution, due to the reduced length of the active part (32mm). A white light interferometer is also shown, using a band-pass filter, from 1.8 to 2.6μm.

We present an integrated optical narrowband electrically tunable filter based on the whispering gallery modes of sapphire microspheres and double ion-exchanged channel BK7 glass waveguides. Tuning is provided by a liquid crystal infiltrated between the spheres and the glass substrate. By suitably choosing the radii of the spheres and of the circular apertures, upon which the spheres are positioned, arrays of different filters can be realized on the same substrate with a low cost industrial process. We evaluate the performance in terms of quality factor, mode spacing, and tuning range by comparing the numerical results obtained by the numerical finite element modeling approach and with the analytical approach of the Generalized Lorenz-Mie Theory for various design parameters. By reorienting the LC in an external electrical field, we demonstrate the tuning of the spectral response of the sapphire microsphere based filter. We find that the value of the mode spacing remains nearly unchanged for the different values of the applied electric field. An increase of the applied electric field strength, changes the refractive index of the liquid crystal, so that for a fixed geometry the mode spacing remains unchanged.

We report the first Silicon/III-V evanescent laser based on adiabatic mode transformers. The hybrid structure is formed by two vertically superimposed waveguides separated by a 100nm-thick SiO₂ layer. The top waveguide, fabricated in an InP/InGaAsP-based heterostructure, serves to provide optical gain, and the bottom Si-waveguides system, which supports all optical functions, is constituted by two tapered rib-waveguides (mode transformers), two distributed Bragg reflectors (DBR), and a surface-grating coupler. The supermode of this hybrid structure is controlled by an appropriate design of the tapers located at the edges of the gain region. In the middle part of the devices, almost all the field resides in the III-V waveguide so that the optical mode experiences maximal gain, while in regions near the III-V facets, mode transformers ensure an efficient transfer of the power flow towards Si-waveguides. The investigated device operates under quasi-continuous wave regime. The room temperature threshold current is 100 mA, the side mode suppression ratio is as high as 20dB, and the fiber-coupled output power is ~7mW.

In the framework of optical telecommunication systems, many functions are integrated on the same substrate. Nevertheless, one of the most important, such as isolation, is achieved using discrete components. It is based on magnetic materials which are always difficult to integrate with classical technologies. This is due to the annealing temperature of magnetic materials. In this paper we present another way for the realisation of such components. We use a dip coating process to report a magnetic nanoparticles doped silica layer on ion-exchanged glass waveguide. The advantages of this method is discussed and we demonstrate its compatibility with ion-exchanged technology. By varying the refractive index of the layer, we can adjust the interaction between the waveguide and the magneto-optical layer.

We propose a new passive optical network (PON) configuration and a novel silicon photonic transceiver architecture for optical network unit (ONU), eliminating the need for an internal laser source in ONU. We adopt dual fiber network configuration. The internal light source in each of the ONUs is eliminated. Instead, an extra seed laser source in the optical line termination (OLT) operates in continuous wave mode to serve the ONUs in the PON as a shared and centralized laser source. λ1 from OLT Tx and λ2 from the seed laser are combined by using a WDM combiner and connected to serve the multiple ONUs through the downstream fibers. The ONUs receive the data in λ1. Meanwhile, the ONUs encode and transmit data in λ2, which are sent back to OLT. The monolithic ONU transceiver contains a wavelength-division-multiplexing (WDM) filter component, a silicon modulator and a Ge photo-detector. The WDM in ONU selectively guides λ1 to the Ge-PD where the data in λ1 are detected and converted to electrical signals, and λ2 to the transmitter where the light is modulated by upstream data. The modulated optical signals in λ2 from ONUs are connected back to OLT through upstream fibers. The monolithic ONU transceiver chip size is only 2mm by 4mm. The crosstalk between the Tx and Rx is measured to be less than -20dB. The transceiver chip is integrated on a SFP+ transceiver board. Both Tx and Rx demonstrated data rate capabilities of up to 10Gbps. By implementing this scheme, the ONU transceiver size can be significantly reduced and the assembly processes will be greatly simplified. The results demonstrate the feasibility of mass manufacturing monolithic silicon ONU transceivers via low cost

Alkaline etching of silicon surfaces was studied to make anisotropic microstructures. An aqueous solution of potassium hydroxide was used as an etchant. The etching rate of silicon was heavily dependent on crystal orientation and temperature; i.e., the etching rate for the (100) surface was four times larger than that for the (111) surface, and they both increased by ten times as temperature rose from 25 to 60 °C. A laser beam was irradiated to a silicon surface to create a temperature distribution that realized selective etching. A pulsed green laser (532 nm) of 5 ns duration was used as a light source to enhance temperature difference between irradiated and nonirradiated portions. By passing through a photomask and an imaging lens system, the laser beam created an optical power distribution on a silicon plate dipped in an etchant. Depending upon the mask pattern, a groove array or a two-dimensional pit array was created on the silicon surface. These pits took a rectangular shape on the silicon (100) plate, while they took a triangular or hexagonal shape on the (111) plate.

In this paper a novel microwave photonics beamformer device concept, for single side band (SSB) 40 GHz modulated signals, is presented. The proposed device comprises tunable lasers, flat-top arrayed waveguide gratings (AWG), a Mach Zender Modulator (MZM), an all-pass ring resonator and photodetectors. The device can be produced as a photonic integrated circuit. The signals from the lasers (one for each beamformer radiant element) are multiplexed by the first AWG, modulated, and passed through the all-pass ring resonator. The AWG channel spacing and the ring resonator Free Spectral Range (FSR) are both set to be equal to 100 GHz. The signal is demultiplexed by a second AWG and finally photodetected. By tuning each laser within its corresponding AWG passband, the phase difference between the optical carrier and the 40 GHz microwave modulated signal for each beamformer element can be controlled. The difference is determined by the phase response of the all-pass ring resonator. A critical part of the design is the alignment between the resonances of the ring resonator and both AWGs, but this can be alleviated by using a single AWG in fold-back configuration. The power provided to each beamformer element is different due to the intrinsic non-uniform losses of the AWGs and the ring resonators, but this can also be solved either by properly setting the lasers power, or by means of additional optical amplifiers. The presented analysis is independent of the integration technology. In Silicon photonics, the AWGs and ring resonator can be produced, while the (hybrid) integration of lasers, modulator, photodetectors (and eventually amplifiers) is a challenge. The device can be monolithically integrated on semi-insulating InP technology.

The influence on photonic crystal waveguide properties of the fabrication-induced disorder was numerically studied. By comparing the transmission spectra obtained using 3D-FDTD for four kinds of fabrication disorders, it was shown that disorder modifies the waveguide mode properties, especially in the slow light regime. Emphasis was put on the influence of the disorder localization. Results have shown the major role played by technological fluctuations of the size, shape, and position of the two first rows of holes along PhC waveguide axis. Results have revealed that bandgap properties remain almost unaffected even for huge disorder levels provided that the two first rows of holes remain unchanged. Interestingly, 3D-simulation have also shown that sharp transmission spectrum cutoffs that characteristize slow wave modes in the two-dimensional PhC bandgap are then not suppressed by the introduction of disorder but are only blue-shifted. This point constitutes an interesting result for optical integrated devices relying on low group velocity phenomena.

The diffraction properties of a silica-based etched diffraction grating are investigated by numerical simulation in threedimension. The field distribution of the electromagnetic filed in the near and far away groove area, the diffraction efficiency and spectral response of 8 channels are analyzed respectively. The numerical results show that the diffraction angle is not satisfied the famous grating formula, the deviation of the diffraction angle is increasing with the diffraction order. And when the groove depth is in the low and medium modulation region, the diffraction efficiency reaches over 90%, and the spectral response of an etched grating is uniform within the fiber communication spectrum range.

We propose to use digital holographic microscopy (DHM) with an illumination in the near infrared spectrum bandwidth, where the silicon is known to have small absorption. With such an illumination condition, it is possible to observe a wider range of specimens than in the visible spectrum, providing a new metrology technique for 3D silicon micro-systems characterization. Suitability of DHM with near infrared illumination for micro-optical elements and wafer inspection is demonstrated. The intrinsic robustness and speed of the method place DHM as a valuable candidate for real-time quality check inside production chains, opening a wide field of applications in quality control.

The presented paper describes a 10 Gbps optical receiver. The transimpedance amplifier (TIA) is realized in standard 0.35 μm SiGe BiCMOS technology. The main novelty of the presented design - investigated in the European Community project HELIOS - is the hybrid connection of the optical detector. The used Germanium photodetector will be directly mounted onto the receiver. A model of the relevant parasitics of the photodetector itself and the novel connection elements (micropads, metal vias and metal lines) is described. Based on this photodetector model an optical receiver circuit was optimized for maximum sensitivity at data rates in the range of 10 Gbps. The design combines a TIA and two limiting amplifier stages followed by a 50 Ω CML-style logic-level output driver. To minimize power supply noise and substrate noise, a fully differential design is used. A dummy TIA provides a symmetrical input signal reference and a control loop is used to compensate the offset levels. The TIA is built around a common-emitter stage and features a feedback resistor of 4.2 Ω. The total transimpedance of the complete receiver chain is in the range of 275 kΩ. The value of the active feedback resistor can be reduced via an external control voltage to adapt the design to different overall gain requirements. The two limiting amplifier stages are realized as differential amplifiers with voltage followers. The output buffer is implemented with cascode differential amplifiers. The output buffer is capable of driving a differential 50Ω output with a calculated output swing of 800mVp-p. Simulations show an overall bandwidth of 7.2 GHz. The lower cutoff frequency is below 60 kHz. The equivalent input noise current is 408 nA. With an estimated total photodiode responsivity of 0.5 A/W this allows a sensitivity of around - 23.1 dBm (BER = 10-9). The device operates from a single 3.3 V power supply and the TIAs and the limiting amplifier consume 32 mA.

An efficient low-voltage lateral current-injection CMOS-compatible light emitting diode (LED) based on Si/SiO2 multiple quantum wells (MQW) is reported. This is the first time that a lateral current-injection LED is demonstrated with Si/SiO2 MQW structures. Strong electroluminescence (EL) in the wavelength ranging from 450 to 850 nm can be observed when the device is reverse-biased at the voltage of as low as ~6 V with the current of ~1 mA. With the lateral current injection structure, the working voltage of the LED is significantly reduced because the voltage is fully applied across the active region instead of dielectrics which cannot be avoided in vertical current-injection Si/SiO2 quantum well LEDs that have received intensive research attention during the last decade. The external quantum efficiency is ~20 times higher than that of the conventional vertical current-injection LEDs based on Si/SiO2 MQW. The light emission would probably originate from the impact ionization due to the hot carriers generated in ultra-thin Si film when the device is reverse-biased. The lateral configuration provides a versatile technology platform, since many light-extraction and mono-chromaticity enhancement techniques can be directly applied onto the top emission window.

By tapping on the volume manufacturing capability of the Si CMOS platform, Si photonics can potentially offer costeffective yet high performance optical interface solutions, and will be especially important in short reach applications. Numerous challenges lie ahead for monolithically integrated Si photonics. There are process integration and thermal budget constraints when monolithically integrating individual Si photonic components such as modulators and photodetectors, and also CMOS on the same chip. In this work, the CMOS-compatible monolithic integration of Si modulators and Ge-on-Si photodetectors on the same wafer is demonstrated and the details of performance optimization are also discussed. Besides process compatibility, the modulators and photodetectors should possess high efficiency and the ability to operate at low power supply voltages. The methods to achieve this are also described. The carrier depletion type Si modulators achieved high modulation efficiency and speed (Vπ.Lπ = 2.6 V.cm, 10 Gbps). At 10 Gbps, an extinction ratio of 6 dB was measured in a modulator with 2-mm-long phase-shifters using single-ended drive (VRF = 5 V_pp). Low voltage operation at 3.125 Gbps was also demonstrated using differential drive, which allowed the drive voltage to be reduced to only 1 V (VRF = 1 Vpp). Ge-on-Si photodetectors were integrated by using a selective epitaxial Ge growth process. The performance of such photodetectors was evaluated in terms of speed, responsivity and dark current for different temperatures and operating voltages. It is shown that introducing a low thermal budget post-epitaxy anneal improves the performance of the Ge photodetectors, resulting in significantly improved dark current. The responsivity and speed in the low voltage regime are also enhanced, which enhances low voltage or even short-circuit (V_Bias = 0 V) operation.

We present theoretical design process for plasmon-enhanced photodetectors with nanometer-scale germanium area. The nontraditional plasmonic metal aluminum is employed as the material of surface plasmon antenna instead of noble metals owing to its integration compatibility with existing silicon complementary metal-oxide-semiconductor technology. The electrode/antenna is patterned with shallow concentric grating surrounding a subwavelength aperture (bull's eye structure) for concentrating and guiding strong optical intensity into an ultra-small active area. The physical modeling and geometric parameters optimization are performed based on the finite-difference time-domain method. Due to the excitation of fundamental or 2nd-order Bloch surface plasmon polaritons, high absorption can be obtained at nearinfrared wavelengths of 1310 and 800 nm.

We present one theoretical analysis about the two resonators system. The model shows good agreement with other theoretical approaches. The properties of the system have been studied using the model. We also provide the analysis about the electromagnetically induced transparency (EIT) in the two resonators system,which shows, properly designed, the system exhibits a narrow high-quality-factor(Q) EIT-like resonant mode.

In this paper, a novel technique to set the coupling constant between cells of a coupled resonator optical waveguide (CROW) device, in order to tailor the filter response, is presented. It is known that using the same K value for all the couplers produces filtering responses with significant side-lobes for the side-coupled integrated spaced sequence of resonators (SCISSOR) or significant ripples in the pass-band for the direct coupled microrings (CROW). It is also known that the side-lobes/ripples can be reduced, and the pass/reject bands can be made wider, by apodizing the K value of each individual coupler in the structure, starting from a nominal K value (either increasing or decreasing it). This technique consists on changing the effective length of the coupling section by applying a longitudinal offset between the resonators. On the contrary, the conventional techniques are based in the transversal change of the distance between the ring resonators, in steps that are commonly below the current fabrication resolution step (nm scale), leading to strong restrictions in the designs. The technique has been experimentally demonstrated employing a racetrack ring resonator geometry. The proposed longitudinal offset technique allows a more precise control of the coupling and presents an increased robustness against the fabrication limitations, since the needed resolution step is two orders of magnitude higher. Both techniques are compared in terms of the transmission response of CROW devices, under finite fabrication resolution steps. The offset technique presented is sufficient by itself for apodization, and optimized CROW's can be produced with a fixed distance between the rings, solely by changing the offsets.

Compact integrated optical directional couplers with symmetrically- and asymmetrically etched S-bend waveguides on SOI platform have been designed, fabricated and characterized. We have found that the directional couplers with asymmetrically etched waveguide structures can increase the device compactness to about 4 to 5 times that of the conventional symmetrically etched bend waveguide structures without compromising much on the optical loss budgets. The over all waveguide loss and loss per S-bend have been measured to be ~ 0.5 dB/cm and ~ 1 dB, respectively. The directional couplers are found to be polarization dependent (up to ~ 2 dB) and nearly wavelength independent (1510 nm < λ < 1600 nm).

An electro-absorption (EA) modulator holds distinct advantages over the silicon Mach-Zehnder interferometer (MZI) modulator by having lower energy consumption, a smaller footprint on-chip, and a potentially higher modulation speed. These are crucial for efficient encoding of optical signals in silicon photonics circuits. Furthermore, the development of a Group IV-based (i.e. silicon- or germanium-based) EA modulator allows compatibility with standard complementary metal-oxide-semiconductor (CMOS) processing. In this work, we demonstrate a novel evanescent germanium (Ge) EA modulator structure. A lateral electric field is employed in the Ge rib to enhance absorption via the Frank-Keldysh effect. This shifts the absorption edge significantly with applied bias for wavelengths beyond 1600 nm. A peak extinction ratio of ~15 dB at 1600 nm could be achieved for a <3 V dynamic voltage swing from a 20 μm modulator. The impact of device dimensions and design structure on optical modulation and insertion loss are also investigated. In addition, monolithic integration of waveguided Ge-based modulator and photodetector can be simplified with our proposed EA modulator structure. The results from this work can make a low power and high speed Ge-based EA modulator viable for future silicon photonics applications.

Optical frequency shifters are essential components of many systems. In this paper, a compact integrated optical frequency shifter is designed making use of the combination of surface acoustic waves and Mach-Zehnder interferometers. It has a very simple operation setup and can be fabricated in standard semiconductor materials. The performance of the device is analyzed in detail, and by using multi-branch interferometers, the sensitivity of the device to fabrication tolerances can be drastically reduced.

Gradual evolution of silicon surface topology from one-dimensional to two-dimensional nanogratings and then to isotropic sets of nanospikes was observed by increasing IR and UV femtosecond laser irradiation dose (the variable number of incident laser pulses at the constant laser fluence). The fundamental mechanisms of these topological transformations are discussed.

A full-vector mode solver for optical dielectric waveguide bends by using an improved finite difference method in terms of transverse-electric field components is developed in a local cylindrical coordinate system. A six-point finite difference scheme is constructed to approximate the cross-coupling terms for improving the convergent behavior, and the perfectly matched layer absorbing boundary conditions via the complex coordinate stretching technique are used for effectively demonstrating the leaky nature of the waveguide bends. The fundamental leaky modes for a typical bending rib waveguide are computed, which shows the validity and utility of the established method.

Arrayed waveguide gratings (AWG) play a key role in dense wavelength division multiplexing (DWDM) systems. While the standard channel count (up to 40) and standard channel spacing (100 GHz or 50 GHz) AWGs feature very good transmission characteristics, increasing the channel counts and narrowing the channel spacings leads to a rapid increase in the AWG size and this, in turn; causes the deterioration in optical performance like higher insertion loss and, in particular, higher channel crosstalk. Channel crosstalk is a result of amplitude errors of the far field profile at the end of the input coupler and phase errors appearing in the phased array as a result of possible effective index and geometrical irregularities of the arrayed waveguides. I our work we show that keeping the length of arrayed waveguides short and using specially-shaped couplers, both phase and amplitude errors can be minimized and, as such, the channel crosstalk strongly improved. To demonstrate this effect, we designed 256-channel, 25 GHz AWG with both, standard and also with specially-shaped couplers. The far field distribution at the end of standard input coupler features strong amplitude errors causing the high channel crosstalk and particularly very high background crosstalk. Applying the specially-shaped couplers led to elimination of amplitude errors in the far field distribution and this had a positive effect on the transmission characteristics. The adjacent crosstalk was improved by ~ 4 dB, non-adjacent crosstalk was improved by ~ 5 dB and background crosstalk by about 10 dB.