Microsystems are chip-scale devices that can sense, think, act, and communicate. Configured as micro-optic devices in a single package, they are hybrid assemblies of microelectronic chips and micro-electro-mechanical systems (MEMS). The devices hold promise for optical networking components and compact analytical instruments for identifying chemical and biological agents.
Biomedical MEMS (see oemagazine, January 2001, page 34) offer a compact approach to analyzing biological and chemical agentsthe so-called laboratory-on-a-chip approachwhich the military hopes to use to detect battlefield and terrorist agents. Such micro-analytical instruments also could find use in the rapidly growing medical and home health-care markets. Industrial control applications also abound for compact, economical, and rugged optical microsystems capable of rapid response.
Polysilicon surface micromachining (SMM) for free-space optics involves the construction of micrometer-scale mechanical elements using polysilicon as the structural material. Interleaved with the structural layers are layers of sacrificial silicon dioxide, which are selectively removed at the end of the process to release the freestanding mechanical structures. To build complex microsystems, designers use multiple mechanical polysilicon layers; thus one metric for the capability of a surface micromachining technology is the number of mechanical polysilicon layers that are available in a given process. Several SMM-MEMS foundries can produce prototype devices, including Robert Bosch GmbH (Stuttgart, Germany), which offers a single (thick) layer technology, and JDS Uniphase (Research Triangle Park, NC), which offers a three-level technology. At Sandia, we have developed Sandia ultra-planar multilevel MEMS technology (SUMMiT), a four-level foundry process, and recently released SUMMiT-V, a new five-layer process. stopping stiction
To make MEMS viable, we must better understand issues of fundamental surface-mechanical properties, optical surface performance, reliability, and packaging.Sandia is investing heavily in the area of reliability science for SMM technology. At the microscopic scale, problems such as fatigue or creep are not as important as they are in the macroscopic scale. However, surface forces and residual polysilicon stress play a very important role in the behavior of microscale devices.
A critical step in any surface micromachining technology is the final release process, which influences the surface chemistry and physics of the device, and therefore its performance and reliability. After removing the sacrificial oxide films with hydrofluoric acid, we rinse the wafers in de- ionized water. Exposing the polysilicon structures to water, as well as to air, leads to the formation of a thin native oxide, which is hydrophilic. The capillary forces caused by residual water absorption can lead to a permanent sticking phenomena in these devices, often referred to as "stiction." Stiction is a major impediment to the commercialization of MEMS components and is under intense investigation at many research laboratories around the world. Residual stress in the mechanical polysilicon layers compounds this problem by bending long, compliant structures toward underlying mechanical elements.
At Sandia, we are pursuing three main approaches to combat stiction. In the first approach, we coat the released MEMS parts with a thin hydrophobic film to inhibit water absorption. Several films are under investigation, one of the most promising being the class of octadecyltrichlorosilane (OTS) molecules. These self-assembled monolayer coatings are effective at rendering the surface hydrophobic; however, their long-term stability is still uncertain.
In the second approach, we modify the drying process to eliminate the water-air interface responsible for stiction. We are investigating supercritical CO2 drying and freeze sublimation, both of which eliminate the liquid phase from the drying process. Although these techniques overcome stiction of components during fabrication, environmental control subsequent to manufacture must still be addressed. The third approach in dealing with stiction is to design microsystem components to minimize their susceptibility to stiction-induced failure.
Early free-space optical mirror designs relied on hinged structures to provide out-of-plane motion. Such mirrors had fixed hinged parts and were raised by applying a lateral force through the rack-and-pinion mechanism. This force caused the mirror to move upward, rotating about the hinges. With the large rubbing surface area, these designs were susceptible to stiction-induced failure.
A robust alternative design developed at Sandia incorporates compliant structures with stiffening elements to counteract capillary forces, such that they operate with few or no rubbing surfaces (see figure 1). The design increases mirror mobility and standoff distance from the substrate, which yields mirror deflection angles of approximately 20°. Measured residual tensile stress values in the SUMMiT mechanical polysilicon typically are less than 10 MPa. The polysilicon residual stress level may be modified in a number of ways to compensate for additional deposited film-stress characteristics.
Figure 1. A 1.5-mm-wide polysilicon mirror minimizes rubbing surfaces.
As with most practical micro-scale devices, MEMS devices must be packaged. Research facilities recently have started to address such packaging issues as component precision- placement, alignment, and micro-assembly. Sandia is developing an IC-packaging technology called mini ball grid array, in which the packaged IC is the same size as the IC itself.1 The advanced substrate technology, designed to replace traditional wiring boards, uses alternating polymeric dielectric benzocyclobutene and copper conducting layers in a build-up photolithographic process to create high-density embedded passive electronic components. practical magic
Two devices developed at Sandia National Laboratories, the µChemLab and the Polychromator, illustrate the importance of integration in microsystems that use optical devices. At the heart of these optical microsystems are two key technologies: MEMS and vertical-cavity surface-emitting lasers (VCSELs).
The optical part of the µChemLab delivers pump light from a VCSEL source to generate and detect broadband laser-induced fluorescence in microfluidic chemical separation systems using electrochromatography.2,3 The goal is to maintain the sensitivity attained with larger tabletop machines while decreasing package size and increasing throughput by decreasing the required chemical volume. Light from the VCSEL is relayed with four-level reflective diffractive optical elements (DOEs) and delivered to the chemical volume through substrate-mode propagation (see figure 2). A high-numerical aperture DOE collects and collimates the indirect fluorescence from dye-quenched chemical species. A filter blocks the excitation wavelength, which allows the detection of the resulting signal as the chemical separation proceeds. In tests, the instrument indicated the presence of nine explosives and related degradation products in under a minute (at a concentration of 50 ppm for each explosive).
Figure 2. Lab-on-a-chip features a VCSEL, optics, and a detector on a 6-mm chip.
In a second analytical example, MEMS devices manipulate the spectral content of light. In collaboration with Honeywell Corp. (Minneapolis, MN) and the Massachusetts Institute of Technology (Cambridge, MA), our group has developed a MEMS-based programmable diffraction grating called the Polychromator chip. The device can diffract a polychromatic beam and deflect the respective diffracted orders for different wavelengths to the same diffraction angles, which simplifies detection.4 The technology offers some interesting applications; for example, a grating can be fabricated to synthesize the optical absorption spectra of molecules or any other complex spectral feature.
We fabricate the chip using SMM. The device consists of an array of 1024 individually addressable beams made of reflective polysilicon. The vertical position of each 10 µm X 1 cm beam can be adjusted electrostatically over a range of approximately 2 µm. The diffractive behavior of the device is determined by the vertical displacement profile across the array of beams. Using the Polychromator, researchers can perform remote chemical sensing using a holographic correlation radiometric approach with the reference gas cell being replaced by the programmable grating (see figure 3).
Figure 3. The MEMS Polychromator adjusts reflective beams to deflect diffractive orders of a polychromatic light through a single angle.
As shown in the examples above, practical optical MEMS promise complex capabilities in tiny packages. For the technology to reach its full promise, however, engineers must take into account stiction, optical surface performance, reliability, and packaging. oe acknowledgments
The authors would like to thank Shanalyn Kemme, Stephen Montague, and Rajen Chanchani, all of Sandia National Laboratories, for their contributions. Sandia is a multiprogram laboratory operated by Sandia Corp., a Lockheed Martin company, for the United States Department of Energy under Contract DE-AC04-94AL85000. References
1. R. Chanchani, K. Treece, P.V. Dressendorfer, Proceedings of the 47th Electronic Components and Technology Conference, (1997).
2. M. Warren et al., Proc. SPIE Vol. 3878, p.185 ( 1999).
3. S.A. Kemme et. al., Proc. SPIE Vol. 3952, p. 375 (2000).
4. M.B. Sinclair et. al., Appl. Opt. 36, p. 3342 (1997)
David Williams, Thomas Picraux, Alton Romig Jr.
David Williams, Thomas Picraux, and Alton Romig Jr. are with Sandia National Laboratories.