Search

Zusammenfassung Mikroelektromechanische Systeme (MEMS) werden heute in einer Vielzahl von Anwendungen eingesetzt. Zur Herstellung von MEMS werden üblicherweise fotolithografische Verfahren eingesetzt. Insbesondere bei sicherheitsrelevanten MEMS, wie z. B. Airbag- oder ESP-Sensoren, ist eine 100 % Inspektion erforderlich. Nach einer optischen Inspektion der MEMS-Strukturen werden diese durch das Bonden eines Silizium-Kappenwafers geschützt. Dieses Bonding oder das anschließende Packaging kann zu einer zusätzlichen Spannung im Waferstapel führen, die die MEMS-Funktion negativ beeinflussen kann. Im Falle eines fehlgeschlagenen elektronischen Tests kann das Problem nicht lokalisiert werden, da der Kappenwafer für die gängigen optischen Oberflächensensoren undurchsichtig ist. In dieser Publikation wird die Herausforderung der optischen Topografiemessung von verkapselten MEMS-Strukturen diskutiert. Dabei werden Defekte wie Verbiegung oder Festklemmen einzelner MEMS-Finger betrachtet. Für diese Messaufgabe wird ein kurzkohärentes Interferenzmikroskop implementiert. Die Wellenlänge liegt im Short-Wave Infrarot (SWIR), da Silizium im Infraroten transparent wird. Die interferometrische Oberflächenmessung der MEMS-Strukturen wird durch den Kappenwafer stark beeinflusst. Hierfür wurden Simulationen durchgeführt, die die systematischen Messabweichungen aufgrund der Kappe zeigen. Eine Möglichkeit zur Korrektur und Reduzierung der systematischen Abweichungen wird ebenfalls beschrieben.

Micro-electro-mechanical systems (MEMS) are used today in a multitude of applications. The production of MEMS is a fully automated wafer process consisting of deposition, patterning and etching. Especially for safetyrelevant MEMS devices such as airbag or ESP sensors, a 100% inspection is required. After an optical inspection of the MEMS structures, they are sealed and protected by bonding a silicon cap wafer on top. This bonding can lead to an additional stress in the wafer stack that interferes with the MEMS function, which is tested by an electrical test. This stress can cause individual elements to stick together or to be bent. In the case of a failed test, the problem cannot be located on the MEMS structures because they are opaque to visible light. The defects are to be detected by a topography measurement. This publication discusses the challenge of topography measurement using an low-coherence interference microscope, which is necessary to achieve a high lateral and axial resolution. It turns out that dispersion does not have such a large influence on the measurement signal as the uneven thickness of the cap. The influence of thickness variation could be reduced by measuring both cap surfaces and subtracting them from the MEMS topography. A comparative measurement of the MEMS without a cap shows that the residual deformation is of a similar magnitude. Removing the cap, however, will cause any stress that may be present to dissolve.

Systems for imaging require to employ high quality optical components in order to dispose of optical aberrations and thus reach sufficient resolution. However, well-known methods to get rid of optical aberrations, such as aspherical profiles or diffractive corrections are not easy to apply to micro-optics. In particular, some of these methods rely on polymers which cannot be associated when such lenses are to be used in integrated devices requiring high temperature process for their further assembly and separation. Among the different approaches, the most common is the lens splitting that consists in dividing the focusing power between two or more optical components. In here, we propose to take advantage of a wafer-level technique, devoted to the generation of glass lenses, which involves thermal reflow in silicon cavities to generate lens doublets. After the convex lens sides are generated, grinding and polishing of both stack sides allow, on the first hand, to form the planar lens backside and, on the other hand, to open the silicon cavity. Nevertheless, silicon frames are then kept and thinned down to form well-controlled and auto-aligned spacers between the lenses. Subsequent accurate vertical assembly of the glass lens arrays is performed by anodic bonding. The latter ensures a high level of alignment both laterally and axially since no additional material is required. Thanks to polishing, the generated lens doublets are then as thin as several hundreds of microns and compatible with micro-opto-electro-systems (MOEMS) technologies since they are only made of glass and silicon. The generated optical module is then robust and provide improved optical performances. Indeed, theoretically, two stacked lenses with similar features and spherical profiles can be almost diffraction limited whereas a single lens characterized by the same numerical aperture than the doublet presents five times higher wavefront error. To demonstrate such assumption, we fabricated glass lens doublets and compared them to single lenses of equivalent focusing power. For similar illumination, the optical aberrations are significantly reduced.

Some of the critical limitations for widespread use in medical applications of optical devices, such as confocal or optical coherence tomography (OCT) systems, are related to their cost and large size. Indeed, although quite efficient systems are available on the market, e.g. in dermatology, they equip only a few hospitals and hence, are far from being used as an early detection tool, for instance in screening of patients for early detection of cancers. In this framework, the VIAMOS project aims at proposing a concept of miniaturized, batch-fabricated and lower-cost, OCT system dedicated to non-invasive skin inspection. In order to image a large skin area, the system is based on a full-field approach. Moreover, since it relies on micro-fabricated devices whose fields of view are limited, 16 small interferometers are arranged in a dense array to perform multi-channel simultaneous imaging. Gaps between each channel are then filled by scanning of the system followed by stitching. This approach allows imaging a large area without the need of large optics. It also avoids the use of very fast and often expensive laser sources, since instead of a single point detector, almost 250 thousands pixels are used simultaneously. The architecture is then based on an array of Mirau interferometers which are interesting for their vertical arrangement compatible with vertical assembly at the wafer-level. Each array is consequently a local part of a stack of seven wafers. This stack includes a glass lens doublet, an out-of-plane actuated micro-mirror for phase shifting, a spacer and a planar beam-splitter. Consequently, different materials, such as silicon and glass, are bonded together and well-aligned thanks to lithographic-based fabrication processes.

Low-coherence-interferometry (LCI) is a well-known optical metrology technique which offers high resolution and a low measurement uncertainty. For the inspection of dynamic and fast processes, single-shot point sensors can show their full capability. Therefore, a new approach of an interferometer is presented, allowing an single-shot acquisition of a LCI signal. The presented concept is based on tilted wavefronts to generate a spatial modulated interferogram. Using a three faceted mirror (TFM) in the reference path, a virtual field point is generated leading to a tilted planar reference wavefront in the pupil plane of the telecentric objective lens. Three mirror surfaces, which are diamond milled out of a monolithic copper cube, are applied to guarantee a stable TFM geometry, resulting in an invariant reference field point position.

International audience; This Letter reports on the fabrication of glass lens doublets arranged in arrays and realized at wafer level by means of micro-fabrication. The technique is based on the accurate vertical assembly of separately fabricated glass lens arrays. Since each one of these arrays is obtained by glass melting in silicon cavities, silicon is employed as a spacer in order to build a well-aligned and robust optical module. It is shown that optical performance achieved by the lens doublet is better than for a single lens of equivalent numerical aperture, thanks to lower optical aberrations. The technique has good potential to match the optical requirements of miniature imaging systems.