The general objective of this presentation is to demonstrate the great potential of InP-membrane photonic devices, with a special emphasis on applications for Optical Communications. Various classes of devices will be presented, which are based on MOEMS (Micro-Opto-Electro-Mechanical) structures, or 2 dimensional (2D)photonic crystals (PC) or a combination of both, according to a '2.5-dimensional' approach, which should broaden considerably the combinations of functionality beyond those presently contemplated with the two first classes. For the MOEMS devices, the basic building block consists in a multi-air-gap/suspended-membrane structure, which can be micro-machined using multi-layered III-V semiconductor based heterostructures : tunable filters will be presented for illustration. For PC devices, the basic building block consists in an InP (and related material) membrane including a 2D PC formed by a lattice of holes : the membrane is either suspended in air or bonded onto low index material, e.g. silica on silicon substrate, in the prospect of heterogeneous integration with silicon based microelectronics. Examples of devices will be presented, specifically micro-lasers based on 2D PC micro-cavities as well as on 2D in plane Bloch modes (2D Distributed-Feed-Back micro-laser). For the '2.5-dimensional' photonic structures, it will be shown that the multi-layer membrane approach, where individual layers or combinations of layers may be structured across to form a 2D PC, is naturally suitable for this purpose. Examples of devices will be presented (2D PC surface emitting micro-sources, switching devices combining vertical MOEMS multi-layer membrane structures and 2D PC).

We built a 20 channel, 200 GHz, fully reconfigurable optical add-/drop multiplexer with integrated variable optical attenuators and power monitor diodes. A single planar lightwave circuit chip contains demultiplexer, switch array, attenuators and multiplexers. It also serves as an "optical motherboard" for a hybrid, flip-chip assembly containing four 10-channel photo detector arrays. A thermal management concept which considers both microscopic and macroscopic aspects of the device was developed. The final device exhibits an insertion loss of 9 dB from "in"- to "through"-port, a 1 dB bandwidth of >50 GHz and switch extinction ratios in excess of 40 dB.

Design and fabrication of thin film GaAs Grating-Assisted Resonant-Cavity LEDs (GA-RCLED) emitting at 980 nm are reported. The devices are fabricated by gluing the sample and a transfer substrate with benzocyclobutene (BCB) and consequently thinning the sample. The design is optimised for a high extraction efficiency. The efficiency of substrate emitting planar RCLEDs is limited by leaky DBR modes. This quasi continuum of modes does not fall in the DBR stop band and is internally reflected at the semiconductor-air interface. Consequently, the leaky modes are absorbed in the substrate and are lost. The use of a thin film device can avoid this loss term. The laterally propagating mode can be recuperated by the use of a grating. The diffractive properties of the periodic grating can redirect the resonant guided mode to the extraction cone. The 2-D grating is defined on the surface of the thin film RCLED by means of holographic exposure. As the devices are thinned they are more compatible with wafer scale integration with electronic devices.

We report on a new concept for InGaAsP-InP 1.55 μm lasers with integrated spot-size converters based on antiresonant reflecting optical waveguides (ARROW). The mode expanders consist of a laterally tapered active region on top of a fiber-matched passive slab waveguide. The large slab mode is laterally confined by an antiresonant configuration of a couple of lateral waveguides defined in the same fabrication process as the active ridge. This feature makes the presented spot-size transformer as simple to fabricate as a standard waveguide, only requiring a planar growth step and a single conventional etch process. The fabricated tapers exhibit a low transformation loss and reduce the coupling loss to standard single-mode fibers from 8 to 4 dB. We also analyze by simulation two variants of the concept proposed in this work, including a taper structure for a buried waveguide, which are expected to show better performance. Simulation results show fiber-coupling efficiencies as low as 2.4 and 1.1 dB for both variants.

During the last years, great progress has been made in all disciplines related to the realization of III-V based Infrared detectors. As a consequence, these detectors have become mature and they have the potential to cope with the competition of II-VI or IV-VI based IR detectors. Due to the wide variety of III-V material systems, one can cover a wavelengths range from the UV to the second thermal window. Due to the improved material quality, the requirements on read-out circuitry and on the packaging of IR sensor chips become more and more important; this necessitates the user to approach the design and verification of a sensor chip from a system perspective instead of concentrating on the device physics.

The high gains in performance predicted for optical immersion are difficult to achieve in practice due to total internal reflection at the lens/detector interface. By reducing the air gap at this interface optical tunneling becomes possible and the predicted gains can be realized in practical devices. Using this technique we have demonstrated large performance gains by optically immersing mid-infrared heterostructure InA1Sb LEDs and photodiodes using hypershperical germanium lenses. The development of an effective method of optical immersion that gives excellent optical coupling has produced a photodiode with a peak room temperature detectivity (D*) of 5.3 x 10⁹ cmHz^½W^-1 at λ_peak=5.4μm and a 40° field of view. A hyperspherically immersed LED showed a f-fold improvement in the external efficiency, and a 3-fold improvement in the directionality compared with a conventional planar LED for f/2 optical systems. The incorporation of these uncooled devices in a White cell produced a NO2 gas sensing system with 2 part-per-million sensitivity, with an LED drive current of <5mA. These results represent a significant advance in the use of solid state devices for portable gas sensing systems.

We present design and fabrication methods for surface normal monolithic amplitude modulators with an aperture up to 14 x 14 mm², a contrast ratio of 6:1 and for low driving voltages (≤8 V). The modulators consist of undoped GaAs/AlGaAs quantum wells embedded in a Fabry-Perot (FP) resonance cavity grown by MOVPE. To improve the device performance the FP cavity, the period and thickness of the quantum well and doping concentration were optimised. Also, the dimension of the modulator were varied from 0.5 x 0.5 to 14 x 14 mm². The results show that the yield of the modulators increases significantly when decreasing the size of the modulators. To remedy the low yield issue for wide aperture modulator, a pixelated approach was used to divide the mono pixel in a monolithic modulator into several pixels, for example from 4 to 48. The modulation speed of the modulators with different dimensions was characterised by electro-optic (EO) response measurements. The temporal optical response of the large modulators was satisfactory up to the order of MHz modulation frequency where the RC constant limited the performance. A few of the modulators with wide apertures are to be assembled into an optical link system for free-space communication.

In this paper we propose an electrically controlled filter for the selection of channels in monitoring DWDM optical network. The highly selective filtering operation needed in this kind of communication systems is obtained with an all silicon device by combining the eletrical injection of free carriers in a forward biased P-i-N diode with a Bragg grating realized over the top of a Silicon on Insulator RIB waveguide. The combination of this fast electron device and the interference principle present under the periodic corrugation of the waveguide allows very fast reconfiguration of the filter. In order to increase filter performaces in terms of selectivity and crosstalk between adjacent channels, apodization of the duty-cycle of the grating is investigated; in this way the performances of the filter in terms of spectral bandwidth for 100 GHz and 50 GHz DWDM spacing become attractive and comparable to ITU specifications.

The implementation of efficient Si optical functions has attracted a considerable interest in the last years since it would allow the use of Si technology for the realisation of integrated optoelectronic (OE) devices. We have fabricated and characterised a novel Si-based light modulator working at the standard communication wavelength of 1.54μm. It consists of a three terminal Bipolar Mode Field Effect Transistor (BMFET) integrated in a silicon rib waveguide realised on epitaxial (epi) Si wafers. The optical channel of the modulator is embodied within its vertical electrical channel. Light modulation is obtained through the formation of a plasma of carriers, inside the optical channel, that produces an increase of the absorption coefficient. Modulation is achieved by moving the plasma inside and outside the optical channel by properly changing the bias of the control electrode. The devices have been fabricated using clean room processes fully compatible with ULSI technology. Electrical characterisation shows a strong channel conductivity modulation. Optical measurements confirm the plasma formation in the channel. The distribution of the plasma under different bias conditions has been directly derived from Emission Microscopy analysis. The devices exhibit modulation depths ranging from 68% up to 83% depending on the bias conditions.

In this paper, we demonstrate a thermo-optically tunable periodic wavelength filter (interleaver) with a 50 GHz free spectral range (FSR). It has an almost rectangular wavelength response and consists of an asymmetric Mach-Zehnder Interferometer (MZI) consisting of two tunable 3dB couplers interconnected by two waveguide channels of unequal length, with a ring resonator coupled to one of the branches of the MZI. The filter is fabricated in silicon oxynitride (SiON) waveguide technology. The bar and cross transmission spectra and chromatic dispersion of the filter have been measured and passband flattening and stopband broadening was observed in good agreement with the simulation. The isolation was 15 dB and 12 dB for TM and TE polarized light respectively, which was lower than the designed 29 dB. The main cause of lower isolation is an inaccuracy of the realization of the power coupling coefficient to the ring (59% instead of the designed 82%). The measured dispersion of the filter varies from 0 ps/nm at the center to 1660 ps/nm at the edge of the passband.

The hybrid integration of polymer and silica in optical waveguides can yield devices that combine the excellent thermo-optic properties of polymers and the superior passive waveguiding properties of silica. The large difference and opposite sign of the thermo-optic coefficients of both classes of materials can be utilized to create athermal waveguide devices. In addition, it can be utilized in thermo-optic devices to induce local changes in the refractive index with boundaries that are sharply defined by the material interfaces and not by gradual thermal profiles. This can also yield devices with attractive thermo-optic behavior.

This paper describes the realization of polymer-based optical structures and the assembly and packaging strategy to connect optical fiber ribbons to the waveguides. For that a low cost fabrication process using the SU-8^TM thick photo-resist is presented. This process consists in the deposition of two photo-structurized resist layers filled up with epoxy glue realising the core waveguide. For the assembly, a new modular vacuum gripper was realised and installed on an automatic pick and place assembly robot to mount precisely and efficiently the optical fibers in the optical structures. First results have shown acceptable optical propagation loss for the complete test structure.

Access networks represent a bottle neck in the present communication networks. The introduction of optical single mode technology into the access networks (Fiber TO THE HOME, FTTH; Fiber To The Antenna, FTTA etc.), would be highly desirable. In order for this to occur a drastic reduction of the cost for key optoelectronic components such as transceivers is needed. We report on and discuss different key technologies crucial for the production of low cost optical single mode components. In particular a technology demonstrator in the form of an array transceiver module has been designed and fabricated, thereby demonstrating the process compatibility between a number of low-cost technologies.

Optical switches and modulators are essential components in integrated optics applications. In this paper, we propose a three terminal optical amplitude modulator embedded in a silicon-on-silicon rib waveguide and we show new experimental results performed on a quasi optimized structure. The optimization of the optoelectronic behavior of the device is possible, by using both the two-dimensional electrical (2-D) semiconductor simulation package MEDICI and optical simulation codes.

Recently an ever-increasing activity in the area of coupling between optical waveguide and ring or micro resonator has been developed. Devices based on this coupling held the promise of a new modality of light switching, amplifications and modulation. In this paper, the feasibility of an all optical switch based on the integration of the potentiality of microcavity resonator and organic materials having large nonlinearities, i.e. liquid crystals (LC's), is discussed. The device is based on silicon technology with hybrid integration of liquid crystals as a nonlinear material.

Optical communication systems operating at 10Gbit/s such as 10Gigabit Ethernet are becoming more and more important in Local Area Networks (LAN) and Metropolitan Area Networks (MAN). This market requires optical transceivers of low cost, size and power consumption. This drives a need for uncooled DFB lasers directly modulated at 10Gbit/s. This paper describes the development of a state of the art uncooled high speed DFB laser which is capable of being manufactured in high volume at the low cost demanded by the GbE market. A DFB laser was designed by developing technological building blocks within the 'conventional’ InGaAsP materials system, using existing well proven manufacturing processes modules wherever possible, limiting the design risk to a few key areas where innovation was required. The temperature and speed performance of the InGaAsP SMQW active layer system was carefully optimized and then coupled with a low parasitic lateral confinement system. Using concurrent engineering, new processes were demonstrated to have acceptable process capability within a manufacturing fabrication environment, proving their ability to support high volume manufacturing requirements. The DFB laser fabricated was shown to operate at 100C chip temperature with an open eye at 10Gbit/s operation (with an extinction ratio >5dB). Up to 90C operation this DFB shows threshold current as low as 29mA, optical power as high as 13mW and it meets the 10Gb scaled Ethernet mask with extinction ratio >6dB. It was found that the high temperature dynamic behavior of these lasers could not be fully predicted from static test data. A production test strategy was therefore followed where equipment was designed to fully test devices/subassemblies at 100C and up to 20Gbit/s at key points in the product build. This facilitated the rapid optimisation of product yields upon manufacturing ramp up and minimization of product costs. This state of the art laser is now transferred into volume manufacture.

The optimization of a 1300nm buried heterostructure(BH)InGaAsP/InP DFB laser for uncooled directly modulated 10Gbit/s operation is described. The development process as well as the key process parameters are discussed and results are presented on an optimized structure. Bandwidths in excess of 10GHz were measured at 90C chip base temperature. Clean open eye diagrams were recorded over the full temperature range, resulting in error free transmission over 40km. To our knowledge the results represent the current state of the art for uncooled BH DFB lasers operating at 1300nm.

A rate equation-based model of MQW semiconductor lasers, has been developed describing power and chirp dynamics. The equations are implemented using an equivalent circuit approach, exploiting the analogy between rate equations and a Kirchoff current balance equation at a capacitor node. Current and voltages across the circuit components are equivalent to the main elements of the rate equations. This solution offers different advantages like the possibility to study the parasitic effects and the opportunity to integrate an high-speed laser model with an IC driver model in the same simulation environment. The model can be easily implemented in any circuit simulator (SPICE, Cadence, Agilent EESoft ADS). All parameters have been derived from measurements on real DFB devices. The model was used to improve static and dynamic performances of InGaAsP MQW-DFB laser 10Gb/s operations, as well as to study the problem of interfacing laser and IC driver. This optimization gave contributions to the realization of uncooled (up to 95°C chip temperature) DFB lasers directly modulated at 10 Gb/s for optical transceivers in 10Gb-Ethernet networks.

Accelerated life tests (ALTs) are aimed at revealing and understanding the physics of the expected or occurred failures. Another objective of the ALTs is to accumulate representative failure statistics. Thus, ALTs are able to both detect the possible failure modes and mechanisms and to quantitatively evaluate the roles of the phenomena and processes that might lead to failures. Adequately designed, carefully conducted, and properly interpreted ALTs provide a consistent basis for obtaining the ultimate information of the reliability of a product - the probability of failure. ALTs can dramatically facilitate solutions to the problems of cost effectiveness and time-to-market. Because these tests can help a manufacturer to make his device a product, they should play an important role in the evaluation, prediction and assurance of the reliability of photonics devices and systems. In the majority of cases, ALTs should be conducted in addition to the qualification tests, which are required by the existing standards. There might be also situations, when ALTs can be used as an effective substitution for qualification tests. Whenever possible, ALTs should be used as a consistent basis for the improvement of the existing qualification specifications. In this discussion, we describe different types (categories) of accelerated tests, with the emphasis is on the role that ALTs should play in the development, design, qualification and manufacturing of photonics products. The case of a laser welded photonic package assembly is used to illustrate some of the concepts addressed.

In technology and nature, tailored scaling represents a principle of success which allows the effectiveness of physical effects to be enhanced. For our optical microsystems, we state that appropriate miniaturization increases the mechanical stability and the effectiveness of spectral tuning by electrostatic and thermal actuation since the relative significance of the fundamental physical forces involved considerably changes with scaling. These basic physical principles are rigorously applied in micromachined 1.55μm vertical-resonator-based filters, capable of wide, monotonic and kink-free tuning by a single control parameter. Tuning is achieved by mechanical actuation of one or several air-gaps which are part of a vertical resonator including two ultra-highly reflective DBR mirrors of strong refractive index contrast: (I) Δn=2.17 for InP/air-gap DBR's (3.5 periods) using GaInAs sacrificial layers and (II)Δn=0.5 for Si₃N₄/SiO₂ DBR’s (12 periods) with a polymer sacrificial layer to implement the air-cavity. In semiconductor multiple air-gap filters, a continuous tuning of >9% of the absolute wavelength is obtained. Varying the reverse voltage (U=0 .. 3.2V) between the membranes (electrostatic actuation), a tuning range up to 142nm was obtained. The correlation of the wavelength and the applied voltage is accurately reproducible without any hysteresis. The extremely wide tuning range and the very small voltage required are record values to the best of our knowledge. Principles of III/V semiconductor micromachining and the detailed technological fabrication process of our filters are focused.

Compact integrated microring resonators are promising candidates for various applications in optical signal processing and communication such as optical filters, wavelength (de)multiplexers, single mode sources, dispersion compensators and wavelength converters. Active GaInAsP/InP material is particularly suited for the fabrication of "lossless" filter devices as well as novel laser components with outstanding performance. Design, technology as well as transmission / emission mode characteristics for typical applications are sketched and summarized.

In contrast to Distributed Feedback DFB) lasers for wavelengths of about 1060nm for α-DFB lasers the fabrication of Bragg gratings using conventional wafer stepper lithography is possible, due to the necessary larger grating period. The use of a wafer stepper enables us to study especially the influence of slant angle of the fabricated gratings and stripe width on optical properties of the fabricated lasers on the same wafer independent from fluctuations of the vertical structure. Best beam quality was achieved for 15° and 13.5° tilted gratings with a small period of 594nm and 658nm respectively. For optimised structures a nearly diffraction limited beam with an output power of more than 1W, a lateral far field divergence angle of θ= 0.3°, a beam quality factor M2 = 1.1 and 3.2 at 2.1W respectively, and a line width of 5.8 pm with 28dB side mode suppression rate was achieved.

The relative intensity noise (RIN) of a semiconductor laser subject to optical feedback ahs been experimentally studied. At low bias current, a low RIN is observed with low feedback ration, the RIN increased in the coherence collapse regime (regime IV) and decreased in regime V. The RIN in regime V is lower than that of the solitary laser. The measurements are found to be in good qualitative and quantitative agreement with theoretical predictions. For higher bias current, a higher feedback ratio is needed for the semiconductor laser to transit from regime IV to V.

We describe design, fabrication and performance of novel, electrically pumped, vertical compound cavity InGaAs lasers emitting at 980 and 920 nm. The concept is scalable and has been demonstrated using monolithic low power (~10 mW) devices all the way to high power extended cavity devices which have demonstrated 1 W cw multi-mode and 0.5 W cw in a TEM₀₀ mode and a single frequency, with 90% coupling efficiency into a single-mode fiber. We also describe uncooled vertical compound cavity lasers in TO-56 can packages which produce 50-100 mW of fiber coupled power. Finally, recent developments in intracavity frequency doubling are summarized.

Semiconductor lasers with high beam quality and high optical output power are very attractive for a variety of applications such as optical pumping of solid-state lasers, fiber amplifiers and medical treatment. When easy and low-cost fabrication is a further requirement, devices based on tapered gain sections are the most promising candidates. Low modal gain, single quantum well InGaAs/A1GaAs devices emitting at 1040 nm were grown by molecular beam epitaxy. The lateral design consists of a tapered gain guided and a ridge-waveguide section having an overall length of mm. An output power of more than 11 W in qcw mode, lifetimes of more than 20,000 h and a record value for the beam quality factor M²of less than 1.5 up to a cw output power of 3.5 W are achieved resulting in an improved brightness of more than 255 MW/(cm²sr). In addition an external-cavity diode laser including a ridge-waveguide tapered amplifier structure is demonstrated to emit more than 2 W cw. The wavelength is tunable over a 60 nm range centered at 1020 nm. The beam quality parameter M² remains below 1.4 for output powers of 1 W over the whole range demonstrating the nearly diffraction limited behavior.

We describe the different mechanisms to generate waves in the transverse section of lasers. Our analysis, based on the Maxwell-Bloch equations, is compared to recent experimental results.

In order to achieve a thermally stable diode laser system based on high power diode laser bars, micro channel heat sinks are used to face the dissipated power with a density of 10⁶ W/m². Passively cooled diode lasers are either lower in power or facing higher junction temperatures. As a matter of principle the cooling with micro channel heat sinks requires a sealing between the heat sink itself and the system around. The leakage of this sealing, normally achieved by O-rings, can be reduced but never avoided. Sensible systems and extreme lifetime requirements, like in the telecom applications, already require passively cooled diode lasers with no water in the inner system boundaries. To achieve a minimized temperature shift in the junction, we developed a new copper based heat sink, spreading the dissipated heat in an optimised manner. Based on this, our further research shows that the higher temperature shift in a passive submount compared with active ones can be tolerated for a system, if the heat resistance to the external water heat exchanger is minimized. For applications either with or without the requirement of a thermo electric cooling element (TEC), we developed a technical solution for a heat exchanger, to keep water out of the inner system boundaries. The thermal resistance is low enough to run up to 12 passively cooled diode lasers on an regular ambient temperature and a minimum of junction temperature mismatch.

A laser diode structure has lately been reported that is capable of generating high-power picosecond optical pulses (~ 50 W / 20 ps) in the near-infrared range for laser radars and other applications. The physical idea consists of achieving fast gain control through the effect of a transverse electric field on the carrier distribution across the active region, which controls the local gain and local absorption at each instant. The mechanism of field-assisted gain control, which has so far been formulated only as a qualitative idea, is justified in this work by simulations of the carrier transport and laser response using the semiconductor device simulator "Atlas" (Silvaco Inc.). A simplified approach is adopted which replaces photon-assisted carrier transport with carrier penetration over the lowered potential barrier. This points to reasonably good agreement between the experimental and simulation results for picosecond pulse generation, provided that the carrier mobilities are assumed to be higher than those in the heavily doped semiconductor structure by a factor of ~ 4. One important conclusion is that comprehensive modelling of the operation of the experimental laser diode is not possible without considering photon-assisted carrier transport, which has not been studied so far at very high carrier densities (exceeding the transparency concentration).