The ultra-fast emission and nonlinear optical properties are reported in novel transition metal dendrimer nanocomposites. Time-resolved results show an ultra-fast decay for the gold (Au) and silver (Ag) dendrimer nanocomposites with a decay time of approximately 70 fs and a complex anisotropy decay result for the silver dendrimer nanocomposite. A weak decay result was obtained from the pristine PAMAM dendrimer and the emission spectrum is also reported. The wavelength dependence of the ultra-fast decay was also investigated for all systems. Nonlinear optical measurements were carried out with both gold and silver dendrimer metal nanocomposites. A strong nonlinear absorption effect was observed for the external configuration of the silver and gold dendrimer metal nanocomposites. The mechanism of the nonlinear absorption in this novel architecture is also investigated. These results show the potential of emission and nonlinear optical effects in a new architecture.

Recently, people have become interested in fine metal particles such as Au,Ag,Pt and Pd because of their importance in applications such as optical devices, electronic devices and heterogeneous catalysis. (1 to approximately 3) In particular, anisotropic fine metal particles have been researched as one of the key technologies to light control. We propose a new technique for preparation of anisotropic Au fine particles. SiO2 film dispersed Au fine particles was prepared by sol-gel method on TiO2 film made by sputtering. Silicon alkoxides and Au salts were used as raw materials for SiO2 film dispersed Au fine particles. The gel film was given heat- treatment at 300 degrees Celsius, and then Au salts we decomposed into Au fine particles. It was identified by TEM (transmission electron microscope) observation that anisotropic Au fine particles were trapped at the interface between SiO2 film and TiO2 film. In addition, compositional distribution in depth direction was investigated. It was confirmed by X-ray photoelectron spectroscopy that Au was distributed on the interface between TiO2 film and SiO2 film and the SiO2 film surface. Furthermore, the polarized light spectrum of Au fine particles separated out on the interface between TiO2 film and SiO2 film were measured, and as a result a polarized absorption spectrum caused by a minor axis and a major axis of Au fine particles were observed.

A real-time recognition system of three-dimensional (3D) objects by use of microlens array is presented. The proposed system consists of two subsystems: one is a projection system of a 3D object into an array of two-dimensional (2D) images that have different perspectives of the 3D object and the second is an optical processor to perform the correlation. A 3D object is illuminated by incoherent light and then is projected into two-dimensional array of elemental images by use of microlens array. Use of microlens array allows us to make the system compact. Each elemental image corresponds to a different perspective of the 3D object. The set of elemental images contain information of the 3D object. After an optical incoherent-to-coherent conversion by an optically addressed spatial light modulator, an optical processor is used to perform the correlation between the input and the reference 3D objects. All the processes can be implemented optically for real-time recognition. Experimental and numerical results of the recognition of 3D objects are presented. These results show that the proposed system can recognize the 3D object even when an input object is very similar to the reference.

A compact image capturing system called TOMBO (thin observation module by bound optics) is developed with compound-eye imaging and post digital processing. To demonstrate effectiveness of the TOMBO architecture, several prototype systems have been constructed with a refractive microlens array and a CMOS (complementary metal oxide semiconductor) image sensor. As a new algorithm for image reconstruction, the pixel rearrange method has been developed. With several test targets, the characteristics of the prototype systems are evaluated.

Individual identification based on biological characteristics such as fingerprint, iris and countenance is regarded as a highly essential technique in security systems. As a simple and rapid recognition system satisfying required performance, we have proposed an opto-electronic system, which combines a parallel joint transform correlator (PJTC) with a personal computer. In this paper, the PJTC method using a new design multiple diffractive optical element as a Fourier transform lens was reviewed and proved to be one of the most practical optical computers for face recognition. Furthermore, based on these first trial results, we proposed the design and fabrication of a portable type compact PJTC (COPaC), of which the size is 23 X 15 X 16.3 cm³ and the weight is 4 kg. We obtained its high accuracy performance for one-to-one correlation using 300 front facial images in a database and proved its practicability. Additionally we performed experiments on ID-less discrimination of twins who look alike for human eyes. Successful recognition rate was obtained, indicating its excellent performance and feasibility.

Optical crossconnect switches with large port counts are the key components for the management of upcoming optical networks. Most technologies proposed for optical switches are essentially planar in geometry, which leads to switch dimensions scaling with the square of the port number. Planar technologies will not scale beyond 64 X 64 ports. To achieve larger (256 X 256, 1296 X 1296) switches, a three dimensional switch geometry is required. Micro electro-mechanical systems (MEMS) are the key technology to implement array of small two tilt axis beam steering mirrors. The presented systems consist of 2D arrays of MEMS mirrors and 2D fiber arrays each with a collimating microlens array. A cross connect path consist of light leaving one fiber and being collimated and projected onto a MEMS micro-mirror by a microlens. The first micro-mirror tilts so as to direct the beam onto a second micro-mirror, and the second micro-mirror tilts so as to direct the light towards a microlens where it is coupled into the output fiber. In this configuration the length of the switch scales linearly with number of ports, and the maximum port number is determined by used the micro mirror technology. Two switches are presented: the first with 256 port and a mean insertion loss of 7 dB and the second with 1296 ports and an insertion loss of 5.1 dB. Both switch show a crosstalk smaller than -50 dB. The optical performance has been verified with input optical signals ranging from 40 DWDM 40 Gb/s and 320 Gb/s TDM data. On the switch with 1296 ports a potential aggregate switch capacity of 2.08 Petabit/s has been demonstrated.

In this paper, we present a calculation of the linewidth enhancement factor for the electro absorption modulator. A model of the quantum confined stark effect (QCSE) is described. The absorption coefficient at different field is calculated. Through the well known Kramers-Kronig relationship, the induced refractive index change has been determined. The calculated linewidth enhancement factor ((alpha) -factor) varies with the applied electric field. This variation is in the 0.1 to 0.3 range.

All-optical actuators based on static or moving holographic gratings could have an advantage over current actuators because of their smaller size, less power and less RF interference. Instead of an ultrasonic wave produced by an electrically driven piezoelectric actuator as in ultrasonic motors, the wave resulted from mechanical deformation of the crystal caused by photo-generated electric charge distribution due to the converse piezoelectric effect. The charge distribution was periodical since it was produced by a holographic grating generated by two interfering coherent laser beams. Surface gratings associated with holographic volume gratings in photorefractive crystals of iron-doped lithium niobate have been studied using diffraction of a reflected probe beam and high-resolution phase-shifted interferometric profilometry. Both techniques show that the surface gratings do in fact exist in the form of periodical corrugations of the same period as that of the volume grating. The maximum amplitude of the surface grating measured by both techniques was close to 6.5 nm. We also demonstrated that the periodical electric forces on the surface were capable of assembling polystyrene microspheres along the fringes of the grating.

To support the rapid growth of communication traffic in information age, all optical node technology is vitally needed. The present review summarizes the design consideration, the fabrication procedures and the result of the test experiments. Major technical issues toward the realistic development of all optical switching modules are listed and the possible solutions are discussed.

Free-space parallel optical interconnects represent a possible solution to the off-chip communication bottleneck. We review developments in this field and compare the performance of some recently demonstrated systems with the projected future requirements of high speed digital systems. We also discuss progress in the area of design methodologies, tolerance analysis and packaging techniques. Manufacturing issues are considered and a comparison is made with other optical interconnect technologies.

We studied the feasibility of two different topologies for board-to-board free-space optical interconnects: a unidirectional ring and a beacon-type star. In each node of the ring bus, the incident VCSEL beam is split both to the next node and to the detector. The star consist of beacon-like nodes based on cones; in the transmitter it reflects the VCSEL beam into all directions on the PCB plane and in the receivers it reflects the incoming beams towards the detector. 3-D models of the optical systems were optimized considering realistic mechanical requirements and tolerances of a typical PCB system. The power budgets and optomechanical tolerances were analyzed by ray-trace simulations. With four nodes and total length of 80 cm, the simulated path loss of the ring bus was 27 dB. At the corresponding 40-cm range, the beacon link had 38 dB signal attenuation. At the same bit-rate, the ring bus provides longer separation between nodes whereas the beacon system allows more nodes. The ring bus also enables parallel interconnects, but the alignment requirements would be very tight. The beacon link was demonstrated. The resulted attenuation was 10 dB higher than in the simulations, mostly due to losses in the receiver optics.

We present an optical interconnection system that transfers 3- D optical data packets among network nodes using fiber image guides. The fiber image guides (FIGs) are potentially easier to align and package and more flexible in their application than alternative techniques. The testbed system includes a reconfigurable field programmable gate array (FPGA), a transimpedance amplifier (TIA) receiver array, and an optoelectronic very large-scale integrated (OE-VLSI) monolithic smart pixel device. The smart pixel device contains vertical-cavity surface-emitting lasers (VCSELs) and metal - semiconductor-metal (MSM) detectors, and performs optical I/Os for the electronic devices. The TIA receiver chip is used with the MSM detectors to convert photodetected current signals into 5V CMOS compatible signals. These components are mounted on a printed circuit board for testing and evaluation of integrated monolithic OEIC designs. We discuss techniques for butt-coupling the fiber image guides to the VCSEL array and optimizing the coupling distance for VCSEL output. The optical signals are transferred most efficiently through the core fibers in FIGs. When the position of the incident beam is near the center of the cladding the power loss is significant. The system demonstrates high data transmission rates over 16 channels with low crosstalk.

We describe an optical fiber splicing process based on the phenomenon of optical beam self-trapping in photocurable organic materials with light induced increase of the refractive index. The ends of two optical fiber cables to be spliced are immersed in a photocurable material with a gap maintained between them. The terminal ends of individual fibers in each cable are facing their counterparts. Two optical beams are sent from opposite fiber cores to meet each other. The beams increase the refractive index of the material. This results into formation of an optically induced bridge that traps and directs the like beams toward the opposite fiber core. This permanent optical connection can be simultaneously built for all the fibers within each of two cables. The splice connection is tolerant to some inevitable tilt or displacement. The advantages of such an approach are high productivity, maintainability, reliability, and cost efficiency. We report the results of theoretical simulations and experiments that demonstrate the feasibility of the approach.

The advances in the design and fabrication of microlaser arrays, photodetectors and free-space optical interconnection elements have driven the creation of ever more 'real world' demonstrator systems. In this paper we review the progress made to date on two separate demonstrator projects which have been assembled at Heriot-Watt University. We shall describe some of the enabling technologies used in the creation of these systems and outline the potential for scaling the architectures described up to sizes where the computational advantages of the optics-in-computing paradigm become highly attractive.

To increase the bandwidth of high-performance chip-to-chip interconnects optical on-board interconnects can be used. Since the design procedure of such optical interconnects has to be widely compatible with current computer aided board design processes, adequate simulation methods are required. In this paper an efficient and design process compatible method for simulating the transmission behavior of optical multimode chip-to-chip interconnects is presented. The approach is based on a time domain description where an optical multimode waveguide is represented by a multiport. The different transfer paths between the input- and output ports describe the transmission behavior of the entire waveguide. The transmission behavior of each individual path can be characterized by its step response, which can be computed by the aid of an extended ray tracing method. Due to some fundamental properties of these step responses, its piecewise approximation by simple exponential functions is possible. As a consequence the pulse responses of each transfer path can be determined analytically and they are also approximated by exponential functions. Finally this procedure enables the application of a semi-analytic recursive convolution method for the computation of the waveguide transmission behavior. The simulation procedure is illustrated and discussed by a set of examples.

A principal performance limitation of current computers is memory access latency. The random access time of DRAM can be as low as 20 ns but the overhead imposed by communication latency can increase the retrieval time to 150 ns in single processor systems or 1 ms in large multiprocessor systems. Optically interconnected VLSI offers the possibility of reductions in the communication component of memory latency of an order of magnitude. The improvement arises from the potential of direct high bandwidth low-latency links between any one chip and each one of a set of others. This potential principally arises from the ease of an optical implementation of fan-out and fan-in operations, together with the intrinsically high bandwidth of optical links. We have designed a scaleable system of processor-memory interconnections to explore this technology. Optical fan-out and fan-in modules will link a single processor to a bank of memory chips. The approach allows for multiple processors to be connected to multiple memory banks in an analogous fashion. The demonstrator will use 1-D VCSEL and photodiode arrays to provide optical i/o for the CPU and memory chips. The optical fan-out, fan-in and image relay can be implemented using an integrated planar optical system.

Centimeter-range high-density optical interconnect between chips is coming into reach with current optical interconnect technology. Many theoretical studies have identified several good reasons why to use such types of interconnect as a replacement of various layers of the traditional electronic interconnect hierarchy. However, the true feasibility and usefulness of optical interconnects can only be established by actually building and evaluating them in a real system setting. This contribution reports on our experience in using short-range high-density optical inter-chip interconnects. It is based on the design and construction of a fully functional optoelectronic demonstrator system. We discuss the rationale for building the demonstrator in the first place, the implications of using many low-level optical interconnections in electronic systems, and the degree to which our expectations have been fulfilled by the demonstrator. The detailed description of the architecture, design and implementation of the demonstrator is not presented here, but can be found elsewhere in this issue.

Architectural studies have identified field-programmable gate arrays (FPGA) as a class of general-purpose very large scale integration components that could benefit from the introduction at the logic level of state-of-the-art massively parallel optical inter-chip interconnections. In this paper, we present a small-scale optoelectronic multi-FPGA demonstrator in which three optoelectronic enhanced FPGAs are interconnected by 2D Plastic Optical Fiber (POF) ribbon arrays. The full-custom FPGA chips consisting of an 8 X 8 array of very simple programmable logic cells are equipped with two optical sources and two receivers per FPGA cell yielding a maximum of 256 optical links per chip. The optical links are designed for signaling rates of 80 to 100 Mbit/s (160 to 200 Mbaud using Manchester coded data) compatible with the maximum clock frequency of the, in 0.6 micrometers CMOS implemented, FPGA chips. The results of parallel link experiments between such modules with both VCSELs and LEDs as sources will be shown. A large scale parallel bit error rate experiment at 90 Mbit/s/channel between two half-populated VCSEL-based FPGA modules with 112 of their 128 channels operational at bit error rates below 10^-13 on all active channels (approximately equals 10 Gbit/s/chip) proves the feasibility of this approach. We first briefly discuss the general architecture and the realization of the optoelectronic FPGA demonstrator system. We then present measurement results on the available modules, followed by some conclusions on this work.

For optical interconnecting waveguides to fiber array coupling, we study a linearly tapered silicon-on-insulator (SOI) waveguide with analyzing the input spot-sizes and the launching positions. In this paper, three cases of the linearly tapered waveguide are compared for indicating optimal light coupling by beam propagation simulations. The best spot size is 4 micrometers with input position (0,4) micrometers for 9 micrometers height Si guiding layer. We also simulate the transverse and longitudinal offset effect for coupling loss analysis between a fiber and the linearly tapered waveguide. The coupling losses can be minimized with properly choosing the longitudinal offset.

Poling induced losses of split-ground plane, push-pull polymeric electro-optic modulators have been investigated. Two sources of loss are found: loss due to the presence of oxygen and loss due to deforming the waveguide structure by large poling fields. Deformation is the most severe at the edges of the electrodes, where the electric field amplitude is largest. Experiments were done by poling waveguides with different architectures and poling in air and in an inert atmosphere. There is an apparent rapid increase in poling induced loss (to the 4^th power) with poling voltage due to the presence of oxygen (up to 6.5 dB/cm for poling field of 170 V/micrometers ), whereas loss due to deformation increases linearly with poling voltage (up to 2.5 dB/cm). Oxygen-induced loss can be minimized by poling in inert atmosphere, while deformation induced loss can be minimized by optimizing device architecture.

Wavelength Division Multiplexing has become a leading technology for long haul transmission systems which operate at 1550 nm wavelength. One of the key components of such systems are tunable filters. Beside low insertion loss, polarization insensitivity and large tuning range there is a strong demand for cost effectiveness and reliability. Two-chip micromachined filters are very promising candidates to fulfill these demands. Two Bragg mirrors are deposited on distinct chips. One of them is engineered as actuable membrane. The Fabry- Perot cavity is created by proper adjustment of the two chips one against the other. Modifying the cavity length by thermal induced heating of the membrane mirror or by applying an electrostatic force provides tunability of the transmission function. Tuning stability and insertion loss can be considerably improved if a stable half symmetric cavity containing a bend membrane instead of a flat one is used. This also helps to overcome some severe fabrication problems. On the other hand the half symmetric cavity is more sensitive to mismatch between filter geometry and phase fronts of the existing Gaussian beam. This aspect and the tolerances which can be accepted are discussed in this paper in detail.

We analyze the potentialities of microlens-based free-space optical pathway blocks for on-chip interconnects. To assess the promises of these modules, researchers typically make use of simple analytic Gaussian beam propagation (GBP). Although this approach leads to a first order layout of a microlens system it does not include aberrations. Aberrations however and -- spherical aberrations in particular -- become important when lenses with a small focal number are implemented. This is especially true when surface emitting lasers with a relative high beam divergence such as e.g. VCSELs are used. In this paper we evaluate how these aberrations affect the performances of such optical interconnection systems and we verify the validity of the GBP method. We enter various GBP layouts in the photonics design software SOLSTIS, which traces real rays or propagates spatially coherent optical beams through the system. We model and compare the performances of different microlens-relay system configurations and we focus on optical efficiency and scalability issues of these micro-optical interconnection components. To conclude we relate optical pathway lengths to minimum microlens diameters and to maximum achievable channel densities.

Chip-to-chip interconnects on printed circuit boards within high-speed electronic systems act increasingly as a limiting bottleneck for the achievable system performance, since local processing speed often exceeds the bandwidth capabilities of conventional electrical interconnects. In addition, rising signal frequencies or clock rates also result in increased susceptibility to electromagnetic interference. The well known limitations and problems of electrical interconnects can be overcome with optical interconnects, which have made their way from long haul telecommunication networks to parallel fiber optical modules for board-to-board interconnects within systems. Extending the advantages of optical signal transmission for very short reach interconnect applications, i.e. board or module level interconnects, therefore is a consequent logical step. This paper presents the integration of optical waveguides into conventional printed circuit boards to achieve hybrid electrical-optical boards with high- bandwidth optical interconnects. The realization of such electrical-optical boards is demonstrated with boards containing 4-channel transmitter and receiver modules, utilizing lead-frame based array GaAs-VCSEL and Si-PIN-diode components. The waveguides are manufactured by hot embossing and laminated into the boards within a standard printed circuit board production process. To couple light into and out of the optical waveguides a butt-coupling technique is applied.

We report on the design, the fabrication, the characterization and the demonstration of scalable multi-channel free-space interconnection components with the potential for Tb/s.cm² aggregate bit rate capacity over inter-chip interconnection distances. The demonstrator components are fabricated in a high quality optical plastic, PMMA, using an ion-based rapid prototyping technology that we call deep proton lithography. With the presently achieved Gigabit/s data rates for each of the individual 16 channels with a BER smaller than 10(superscript -13 and with inter-channel cross-talk lower than -22dB the module aims at optically interconnecting 2-D opto-electronic VCSEL and receiver arrays, flip-chip mounted on CMOS circuitry. Furthermore, using ray-tracing software and radiometric simulation tools, we perform a sensitivity analysis for misalignment and fabrication errors on these plastic micro-optical modules and we study industrial fabrication and material issues related to the mass-replication of these components through injection-molding techniques. Finally we provide evidence that these components can be mass-fabricated in dedicated, highly-advanced optical plastics at low cost and with the required precision.

A coupling concept for a self-aligning and passive assembly of optical transmitter and receiver modules to board-integrated multimode waveguides is presented. The coupling mechanism is based on a 90 degree(s) beam deflection provided by micro-mirrors which are part of the board-integrated waveguides. The alignment is obtained by a mechanical high precision interface consisting of alignment pins well known from MT connectors at the module side and corresponding holes which are part of the optical layer. As the production of the alignment holes is part of the manufacturing of the optical layer, the required accuracy can be achieved without noteworthy difficulties.

An adaptive alignment technique is presented that provides precise control and active positioning of sub-millimeter-sized spherical lenses in two-dimensions through the application of electrophoretic forces in a microfluidic well. The device is comprised of a lithographically patterned microfluidic well and electrodes that can be addressed to position or align the spherical microlens to the corresponding beam source. The motion of the microlens is controlled using CMOS compatible voltages (3V - 1 (mu) A) that are applied to opposite electrodes in the microfluidic well, creating an electrical field in the solution. By applying voltages to opposite electrode pairs, we have demonstrated the movement of spherical microlenses with sizes ranging from 0.87 micrometers to 40 micrometers in directions parallel to the electrode surface. Under a bias of 3 volts, the microspheres had an experimentally measured electrophoretic velocities ranging from 13 to 16 micrometers /s. Optical alignment of the spherical or ball microlens can be accomplished using feedback from a photodetector to position the lens for maximum efficiency. Using this device, it is possible to actively align microlenses to optical fibers, VCSELs, LEDs, photodetectors, etc.

Planar optics is a promising approach to the integration of free-space optical systems. From a systems perspective we discuss the processing steps and tools necessary for the engineering and fabrication of planar optical systems. These contain layout and simulation of the integrated free-space optical systems, generation of lithographic masks, and systems fabrication. We present integrated planar-optical systems for applications in optical interconnects, optical security and signal processing systems, as well as optical sensors.

We studied the possibilities of LTCC (low temperature cofired ceramics) technology to fabricate transmitter arrays equipped with vertical-mounted multimode fiber pigtails. The developed LTCC module can be mounted vertically on a printed-circuit-board (PCB), thus providing small, essentially one-dimensional PCB footprint. The fiber is aligned and supported using a hole structure through the layers, and the surface-emitting source, such as a VCSEL, is flip-chipped on the other side of the substrate. Thus, this kind of module can be used as a detachable electrical interface between the fiber-optic and electronic media. To evaluate the feasibility of the system, a 5-channel transmitter based on a 4-layer LTCC substrate was designed and realized, and mounted vertically on a test board. Each transmitter, sized 5 mm X 5 mm², included VCSEL and laser-driver chips as well as discrete passives. 62.5/125-micron fibers were used with metallic tubes as strain relieves. Before implementation, the optical alignment tolerances were examined by measurements and simulations, and the tolerances of fiber-mounting holes were evaluated by preparing test structures. The fiber alignment accuracy seemed adequate even for the first transmitter prototypes. Nevertheless, stringent requirements for the LTCC process control are necessary to achieve the needed accuracy.

Hybrid glass materials are used in the photolithographic fabrication of optical and opto-mechanical structures. Two different methods are introduced. The first one is referred as photolithographic patterning and the other as direct photolithographic deforming of hybrid glass materials. Fabrication of isolated lenslets, lens arrays, gratings and other binary structures is presented. The hybrid glass material used in the photolithographic patterning features a maximum spectral extinction coefficient of 2.0 X 10^-4 micrometers ^-1 between 450 nm and 1,600 nm and a refractive index of 1.53 at 632.8 nm. The fabricated structures feature large convex lens sags (up to 100 microns) with rms surface roughness values ranging from 10 to 45 nm, when the photolithographic patterning is applied. The hybrid glass material used in the direct photolithographic deforming exhibits a maximum spectral extinction coefficient of 1.6 X 10^-3 micrometers ^-1 at wavelengths ranging from 450 nm to 2200 nm and a refractive index of 1.52 at 632.8 nm. The fabricated structures exhibit rms surface roughness between 1 and 5 nm, when direct photolithographic deforming is applied. These materials and methods are highly promising for micro- optics fabrication.

There are several properties of microlenses which have to be characterized: paraxial parameters like the focal length and more complex properties like the surface quality or the wave aberrations. For arrays of microlenses the homogeneity of the lenses and the accuracy of the positioning may also be very important. Additionally, the surface of the microlenses can have spherical, aspherical or cylindrical shape depending on the application. Several measurement methods for nearly all of these parameters will be presented. A Twyman-Green interferometer is used for the measurement of the surface deviations of a microlens from an ideal spherical shape and for the measurement of the radius of curvature of the surface. A Mach-Zehnder interferometer measures the wave aberrations of microlenses and the focal length. With the help of grazing incidence interferometry all surfaces with cylindrical symmetry can be measured. The cross-section of the surfaces can be circular or non-circular. Diffractive optical elements are applied as beam shaper and reference elements. The testing of complete arrays of microlenses can be done with several methods. A Smartt test can be employed to measure the wave aberrations of an array of microlenses and a shearing device might be suitable to measure the uniformity of the focal lengths of the microlenses.

In this paper we present our latest results on the fabrication and characterization of plastic microlenslet arrays using Deep Lithography with Protons (DLP) and highlight their geometrical dimensions, their surface profile and their uniformity. We also present quantitative information on their optical characteristics such as focal length and spherical aberration as measured with a Mach-Zehnder interferometer. Furthermore we demonstrate the flexibility of the DLP technology to fabricate arrays of microlenses that feature different pitches and different sags. Although the DLP technology is a valuable tool to rapidly prototype refractive micro-optical components, the approach is unpractical for mass-fabrication. We therefore introduce a replication technique, called vacuum casting, which is very appropriate when only a few tens of copies have to be made, and we bring forward the first quantitative characteristics of these microlens replicas.

The mask structured ion exchange in glass (MSI) is a powerful tool for realizing general planar phase distributions and in particular custom designed planar GRIN micro lenses with diffraction limited performance and high fill factor. For lens characterization the numerical aperture is a key parameter. However the classical geometrical definition of the N.A. disregards aberrations. Here we suggest an addition to this classical definition, which is based on diffraction limited performance. For a testing of micro lens arrays, global process parameters are assessed by interferometric measurements of a subset of the lenses. Local process variations typically result in small non-symmetric aberrations. These aberrations mainly lead to a lateral shift of the focus. Thus, for rapid quality control of micro lens arrays we analyze all focal positions in parallel. From the lateral deviations of the focal positions a quality criterion for each individual micro lens can be derived.