The Digital Micromirror Device™ (DMD™) is an array of individually addressable micromachined mirrors fabricated on a matching array of SRAM cells. The DMD and its associated electronics form the heart of myriad projection displays: portable multimedia projectors, high definition televisions, and Digital Cinema installations worldwide. First commercialized ten years ago, the DMD represented a signal achievement in the large-scale integration of micromechanical elements on a standard CMOS substrate. Here we present an overview of the materials and integration challenges that have been overcome to scale the technology to accommodate ever-shrinking feature sizes (17, 14, 10, … microns) and ever-increasing resolutions (SVGA, XGA, SXGA, SXGA+, 720p, 1080p, ….). Specifically, we will detail the evolution of the DMD architecture and concomitant tradeoffs necessary as the process and the material set become ever more interdependent at shrinking feature sizes.

OLED display technology has matured into viable products that have applications in the military and consumer market. To date, most OLED based display products are passive matrix devices, with the exception of a few low temperature polysilicon based active matrix devices. Single crystal silicon based microdisplays (OLED-on-silicon) have many advantages, both technologically and commercially. The technology is unique and is based on the top emitter architecture using small molecule organic materials. White light emitting OLED devices are fabricated using a host-dopant system. The devices are fabricated using high Tg materials with a dual emission layer structure wherein a blue-green light emission layer precedes an orange light emission layer. The devices are hermetically sealed with vacuum deposited thin film layers. Over the thin film seal, LCD-type color filters are patterned using standard photolithography methods to generate primary R, G, B colors. The OLED device architecture and fabrication process will be broadly described. Results of recent improvements in the OLED-on-silicon microdisplay technology, with emphasis on efficiencies, lifetimes, grey scale and CIE color coordinates, will be presented. Improved performance microdisplay will feature an SVGA resolution OLED-on-silicon full color microdisplay that is only 0.44 inches diagonal. eMagin’s 3-D video capable headset consisting of OLED microdisplays will be described and demonstrated.

Packages for MEMS, microsystems and photonic devices enable both electrical and non-electrical signals to pass between the external world and the die thus enabling their use at the system level. The need for this non-electrical interface combined with the die’s greater sensitivity to stress as well as the need to establish an environmentally controlled cavity around the die, place challenging requirements on the package technology. These factors in turn drive the selection of materials used in the package which ultimately impact time to market, cost, manufacturability, performance and reliability of the device. Optical, fluidic (liquid and gas), thermal and mechanical interfaces must be adapted to the unique requirements of the device and they often add significant cost. Optical devices require transparent interfaces that can include highly polished surfaces, anti-reflective coatings, and precise alignment of components. Fluidic devices require interfaces such as tubulations and seals that must withstand the pressures and chemical reactivity of the gasses and liquids that are used in the system. Mechanical interfaces must take into account externally applied stress, flexure and vibration while thermal interfaces must designed to either rapidly dissipate heat or to provide thermal isolation.Virtually all microsystems are subject to performance and reliability issues due to the stresses induced because of differences in the coefficient of thermal expansion of the materials. Stress insensitive designs, materials with similar CTEs and stress isolating materials are critical to addressing these issues; differential signal paths and symmetric layouts to minimize the effect that CTE mismatch. Materials such as Kovar and aluminum nitride are important when trying to minimize the CTE mismatch with silicon. Low modulus materials such as elastomers and certain composite materials are also helpful in reducing stresses imparted on the die during manufacture and operation in the field.Physical devices such as inertial sensors and electrical switches require primarily electrical interfaces to the system, but the devices often require precisely controlled gas environments due to the electrical and physical requirements such as mechanical damping, standoff voltage, etc. Obtaining the desired environment inside the hermetic package prior to sealing can present major challenges.This talk will conclude with a detailed description of the packages used in several devices that have successfully entered production after having addressed these material issues. The package for an ultra-high pixel count display, a MEMS switch array and an inertial sensor will be discussed in detail. The issues unique to these applications and the methods used to combat their respective issues will be highlighted. Opportunities for future improvements will then be discussed as well as strategies to address cost and timing issues so critical in today’s product development environment.

Optoelectronic structures combined with standing waves [1 – 6] allow various ways of making wavelength-sensitive devices, devices that can also perform spectral pre-processing on signals. Such approaches can also allow processing directly in transform domains that can give useful results without actually having to perform the inverse transform. Modern microstructures and nanostructures allow many ways of fabricating such devices, and micromechanical systems integrated with the optoelectronic structures allow yet further functionalities. Some technologies also can take advantage of integration with electronics for highly functional integrated sensing systems.In this talk we will summarize work in various spectral sensors exploiting standing waves. These include diode structures with quantum well absorbers [2], thin absorbers combined with micromechanical structures for complete spectrometers [3], adaptive sensing techniques that can exceed conventional spectroscopic resolution [4,5], spectral signature detection [4,5], and detectors tunable in nanosecond timescales [6]. Recent work has shown the possibility of integration directly with silicon integrated circuits [7], and other work has evaluated upper bound limits to the performance of nanophotonic spectral sensing devices [8].[1] 172.D. A. B. Miller "Laser Tuners and Wavelength-Sensitive Detectors Based on Absorbers in Standing Waves" IEEE J. Quantum Electron. 30, 732 749 (1994).[2] 170.L. Carraresi, E.A. DeSouza, D. A. B. Miller, W. Y. Jan, and J. E. Cunningham "Wavelength-selective detector based on a quantum well in a standing wave" Appl. Phys. Lett. 64, 134 136 (1994).[3] H. L. Kung, S. R. Bhalotra, J. D. Mansell, D. A. B. Miller, and J. S. Harris, Jr., “Standing-Wave Transform Spectrometer Based on Integrated MEMS Mirror and Thin-Film Photodetector,” IEEE J. Selected Topics Quantum Electron. 8, 98 – 105 (2002)[4] S. R. Bhalotra, H. L. Kung, Y. Jiao, D. A. B. Miller, “Adaptive time-domain filtering for real-time spectral discrimination in a Michelson interferometer,” Optics Lett. 27, 1147-1149 (2002)[5] 255.Y. Jiao, S. R. Bhalotra, H. L. Kung, and D. A. B. Miller, “Adaptive imaging spectrometer in a time-domain filtering architecture,” Optics Express 11, 1960-1965 (2003)[6] 289.R. Chen, D. A. B. Miller, K. Ma, and J. S. Harris, Jr., “Novel Electrically Controlled Rapidly Wavelength Selective Photodetection Using MSMs,” IEEE J. Sel. Top. Quantum Electronics, 11, 184-9 (Jan.-Feb. 2005)[7] Y.-H. Kuo, Y.-K Lee, Y. Ge, S. Ren, J. E. Roth, T. I. Kamins, D. A. B. Miller, and J. S. Harris, “Strong quantum-confined Stark effect in germanium quantum-well structures on silicon,” Nature 437, 1334-1336 (2005)[8] 331.D. A. B. Miller, “Fundamental Limit for Optical Components,” J. Opt. Soc. Am. B, published on-line at http://www.opticsinfobase.org/abstract.cfm?msid=76778, January 18, 2007

This paper will discuss how temperature sensors can be integrated on a silicon chip using standard integrated circuit technology. Compared to traditional temperature sensors such as thermistors or platinum resistors, integrated temperature sensors can be mass-produced at low-cost, while providing additional functionality such as local data storage/processing, digital output and bus interfaces. Integrated temperature sensors usually exploit the well understood temperature dependency of bipolar transistors. This dependency is affected by process spread, and so without trimming, such sensors typically achieve an inaccuracy of only a few degrees over the military temperature range (–55°C to 125°C). However, it will be shown that by careful circuit design, and by a single room-temperature trim, sensor inaccuracy can be reduced to less than 0.1°C. In the future, a new class of integrated temperature sensors may not even require trimming. Such sensors exploit the temperature-dependent thermal diffusivity of silicon. This is very insensitive to process spread, since it is a material constant of the high-purity silicon used in integrated circuits. Recently, it has been shown that such sensors can achieve untrimmed inaccuracies of less than 0.5°C over the military temperature range.

Tuesday, Nov 27New Presenter*N2.3 @ 11:00 AMSi Resonators as Transducers and Frequency References. Mark Brongersma

Depositing a wide range of materials as nanoscale features onto diverse surfaces with nanometer registration are challenging requirements for any nanoscale processing system. Dip Pen Nanolithography® (DPN), a high resolution, scanning probe-based direct-write technology has emerged as a promising solution for these requirements. Many different materials can be directly deposited using DPN, including alkane thiols, metal salts and nanoparticles, metal oxides, polymers, DNA, and proteins. Indirect deposition allows the creation of many interesting nanostructures: eg. using MHA to create arrays of antibodies which then bond specifically to antigens on the surface of viruses or cells to create cell or virus arrays. The DPN system is designed to allow registration to existing features on a writing substrate via optical alignment or nanoscale alignment using the core AFM platform. This allows the nanoscale deposition of sensor materials onto directly onto monolithic chips with both sensing and circuit features.To enable the DPN process, novel pen and ink delivery systems have been designed and fabricated using MEMS technology. These MEMS devices bridge the gap between the macro world (instrument) and the nano world (nanoscale patterns). We realized that the initial MEMS devices would need to be simple and robust both in design and fabrication to get the products into the marketplace quickly. The first MEMS device was a passive pen array based on silicon nitride AFM probe technology from Cal Quate’s group at Stanford. The next two devices were more complicated and took considerable effort to commercialize; an inkwell chip and a thermal bimorph active pen. Some of the difficulties in bringing brand new MEMS devices from the prototype stage into production will be shared. The subsequent MEMS products have become even more complicated both in design and fabrication but the development process has improved as well. For example, the 2D nano PrintArray has 55,000 pens in one square centimeter for high throughput writing over large areas. The 2D arrays enable templated self assembly of nanostructures giving researchers the ability to control the placement of self assembled features rather than allowing the self assembly to occur randomly.Applications of the DPN technology vary from deposition of DNA or proteins in nanoarrays for disease detection or drug discovery, deposition of Sol-gel metal oxides for gas sensors, and additive repair of advanced phase-shifting photomasks.

Optofluidics refers to a class of adaptive optical circuits that integrate optical and fluidic devices. Familiar examples include liquid crystals and dye lasers. The introduction of liquids in the optical structure enables flexible fine-tuning and reconfiguration of circuits so they can perform tasks optimally in a changing environment. I will discuss how the emergence of fluidic transport technologies at the micron and nanometer levels opens possibilities for novel adaptive optical devices.

The suitability of electrowetting-on-dielectric (EWD) microfluidics for true lab-on-a-chip applications is discussed. The key advantage of EWD microfl;uidics is that a wide diversity in biomedical applications can be parsed into manageable components and assembled into architecture that requires the advantages of being programmable, reconfigurable, and reusable. This capability opens the possibility of handling all of the protocols that a given laboratory application or a class of applications would require. And, it provides a path toward realizing the true lab-on-a-chip. EWD microfluidics (also known as digital microfluidics) offers an almost complete set of elemental fluidic components that support the required fluidic operations. One way of implementing multiple applications on a chip is through the use of a reconfigurable array. Of course, there are certain requirements for creating such a versatile architecture that is capable of accommodating multiple applications. First, the components must be integrated on a common substrate. That is, the components must be accessible by selective routing of reagents, voltage signals, and sensor input/output, depending on the need. There are various ways of doing routing, including programming before use (programmable hardware or modules) or reconfiguring devices on the fly (reconfigurable computing and electronic control).The use of a reconfigurable EWD microfluidic array requires high-voltage drive signals (up to 100V) and logic to control the clocking of the voltages. Methods for sample collection and analyte detection must also be integrated. Two avenues for integrating high-voltage electronics into the array are being pursued: multi chip and system on a chip. The multi chip approach uses standard CMOS logic for the controller and commercial high voltage MOS devices to drive the microfluidics. This multi chip approach is low risk, but adds complexity and cost to the integration. For the most compact design, the CMOS IC could also have the micro fluidics integrated onto the die with the circuits. The main issue with this approach is the size of the microfluidics. The current nanoliter microfluidic technology uses 100µm electrodes for manipulating regent droplets, and these dimensions lead to a die a few centimeters on a side in size. This may not yield low cost, unless very high volumes are produced and low cost CMOS processes, such as 15-year-old 1.5 um CMOS, are used. Alternatively, a more state–of-the-art CMOS could be used, but with 10-20 um pads for controlling the microfluidics. This would require development of new microfluidic control and design strategies at the picoliter scale. Approaches to picoliter chip integration are discussed.

In the field of clinical diagnostics and pharmacology or drug screening, the use of biological cells from living body becomes important for fast monitoring of specific or non-specific reactions. Especially, the dynamic monitoring of single cell in an independently controlled environment is important in order to exclude the influence by other cells such as a mixture of hormones, ions, and neurotransmitters released from neighboring cells. Conventionally, single cell has been manually loaded and isolated using a micromanipulator. This allows only a few cells to be analyzed at a time and makes it difficult to monitor many cells simultaneously in a short time. Commercialized flow cytometry such as fluorescence activated cell sorter (FACS) can provide single cell sorting but the requirement of fluorescence tagging, expensive price, and slow sorting time due to serial sorting still remain as an issue.In this presentation, we will introduce the basic microfluidic components that provide the required functions for single-cell assay: to load cells, introduce reagents for cell stimulation and monitor cell reaction in a fast, high-throughput and highly-parallel manner. Current microfabrication technologies can provide a good platform in an integrated form for proper functions such as micromixers, micropumps, microvalves, microchannels and microwells [1]. Also, the advantages of microfluidic devices for cell analysis include high functionality by integration of electrodes, various structures, detection parts, and electronics, which makes possible several different analyses simultaneously on a single chip. The significant reduction in analysis volume can improve the sensitivity and save the cost of expensive reagents. We will introduce a few prototype devices that we implemented in an array of microwells [2,3]. The implemented device was designed to provide three consecutive functions: autonomous capture of cells into multiple micro-wells in a two dimensional array; active physical sealing of micro-wells for cell isolation; and specific chemical injection into each micro-well. We used multiple polydimethylsiloxane (PDMS) layers to form microfluidic channels on the top of silicon substrate. A cell capture experiment has been performed using polystyrene beads as well as CHO DG44 live cells, and successful positioning of cells on micro-wells was achieved. Also, the selective drug injection into a specific target cell has been demonstrated. [1]Book Chapter: K.-S. Yun and Euisik Yoon, “Micropump Systems Techniques and Applications in MEMS,” in MEMS/NEMS Handbook: Techniques and applications, Springer 2006.[2]K.-S. Yun and E. Yoon, "Micro/Nanofluidic Device for Single-Cell-Based Assay," Biomedical Microdevices, Vol. 7, No. 1, pp. 35-40, 2005.[3]K.-S. Yun, D. Lee, H.-S. Kim and E. Yoon, “A Microfluidic Chip for Measurement of Bio Molecules Using Microbead-Based Quantum Dot Fluorescence Assay,” Meas. Sci. Technol. 17, pp. 3178-3183, 2006.

In the semiconductor industry socalled ASICs (Application Specific Integrated Circuits) have long become a very important segment of the market. ASICs are characterized bystandardized technological processes available in large waferfabs, like standard CMOS processes, the design of custom chips in smaller design houses that own specific intellectual property for certain applications which is plugged into design kits supplied by the waferfabs,and a rapid access to prototypes. It is anticipated that in the emerging field of labchips a similar market will develop. Therefore an urgent need exists for the exploration of technological platforms allowing the realization of a multitude of microfluidic, actoric andsensoric components in standardized processes. At the IMT / Cµ a technological platform has been developed which aims at these applications. It uses a backend process allowing the fabrication of a wide variety of microfluidic components on both, glass or plastic substrates as well as on processed CMOS chips. While today most labchips still use glass or plastic substrates, it is anticipated that, asthe dimensions of microfluidic devices will shrink and a need for on chip electronic detection and processing will be recognized, substrates may also be CMOS chips with all the electronic functions required underneath the microfluidic system. Moreover, it is not unlikely that many of these applications will be one time use applications, particularly, if the dimensions of the chip are reduced to square millimeters and the cost per chip to cents. Theconcept is termed ALM (Application specific Lab on Microchip). The photomicrograph shows a multifunctional test device fabricated at the IMT. The ALM technology platform also includes a-Si:H based devices for on chip optical detection. The process flow is outlined and experimental and theoretical results on a varietyof elementary modules are presented, including novel planar narrow channel electro-osmotic micropumps working at voltages of a few volts, a micro mass flow controller based on the thermal anemometric principle and an on chip microcapillary electrophoresis system equipped with highly sensitive a-Si:H pin diodes for optical detection. The obtained results demonstrate the feasibility of the proposed concept.

Direct electrical interfacing of semiconductor chips with neuronal tissue may lead to novel experimental approaches in brain research and also give rise to hybrid computational devices. For this purpose we developed a CMOS based multi-transistor-array with a spatial resolution of 7.8μm on an active area of 1mm$^2$. The high bandwidth of the CMOS chip allowed a temporal resolution of 0.5ms for full frame readout in a first generation of the chip. In a second generation of chips we were able to improve the rms noise from 230 μV to a level of 70 μV. The frame rate was increased to 6kHz. By choosing subsets of the active area higher temporal resolutions are possible. To avoid ohmic coupling between chip and bath the sensor electrodes are covered with an inert titanium dioxide surface.We tested the chip by measuring time resolved maps of the electrical signals of single cells and small networks of snail neurons. From these maps e.g.\ inference can be drawn about the distribution of ion channels in the membrane.In a next step we applied the first generation of chips to measurements of signal maps of hippocampal slices of the rat brain. Brain slices were cultured on the chip for 2-3 weeks by an adapted Gaehwiler method. Upon stimulation in the CA3 region with a tungsten electrode we observed fast propagating waves of negative field potentials which we assign to action potentials in the mossy fibers and slower transient field potentials of postsynaptic activity in CA3 and CA1. The transistor signals matched local micropipette recordings of electrical field potentials in amplitude and shape. Using the second generation of chips we were able to resolve more complex dynamics in the slices. Due to increased sampling rate and reduced noise we could detect weaker and faster signals. Propagation of action potentials along the mossy fibers and also along the Schaffer collaterals could unambiguoulsy be resolved. We were also able to measure complete maps of long-term potentiation induced by theta pulse stimulation.In a third system we applied the chip to the retina of rabbit. Retinal patches were prepared from the rabbit eye and placed on the chip with the ganglion cell layer facing the chip surface. Stimulation was performed by various light patterns generated on a small TFT screen which was projected onto the retina through a microscope objective. Due to the high spatial resolution on a large area we were able to measure signals from many ganglion cells simultaneously. Several types of ganglion cells (e.g.\ on-cells, off-cells, direction sensitive cells) could be identified.Direct interfacing of an MTA chip provides a complete observation of neuronal signaling in an extended area of neuronal tissue. This technique is suitable to elucidate the functionality of planar neuronal systems at a high resolution. A next generation of chips that will use bidirectional pixel elements which can be used for stimulation and detection is currently under development.

Spintronic biomolecular recognition platforms allow fast ( tens of minutes), sensitive ( down to few femptomole/cm2) detection of biomolecular recognition processes ( DNA-cDNA, antibody-antigen) occurring in various biomedical, environmental, and quality control assays 1. The portable platform built at INESC MN has three main modules: the detection module, the fluidics module , and the electronic control and communications module. In the fluidics module (PDMS+plastic), biological targets are labeled with magnetic nanoparticles ( the labels), and sent to the detection chamber. The detection module, is a scalable sensor matrix fabricated on glass or Si and containing 16x16 probes at present. Each matrix node consists of a thin film diode in series with a magnetic tunnel junction linear sensor 2. Current lines guide the biological targets to the probe sites. The linear magnetic sensors detect the fringe fields created by the immobilized magnetic labels after hybridization to complementary probes, and proper washing cycles. Signal is linear on label number and target surface concentration. Particular care was take on chip passivation and on the functionalization chemistry. The electronics control and communications module 3 take care of AC signal detection ( < 1 kHz), and wireless data communications to a PDA or laptop. The platform is being tested for SNIP and mutation detection in cystic fibrosis, and for Salmonella and e-Coli detection in water, using 250nm magnetic labels. 1-“Magnetoresistive DNA chips”,P.P.Freitas et al., in Magnetoelectronics, Chapter 7 (Ed. M.Johnson), Elsevier Inc. (2004)2-“Noise characteristics and particle detection limits in Diode+MTJ matrix elements for biochip applications”, F.A.Cardoso, R.Ferreira, S.Cardoso, J.P.Conde, V.Chu, P.P.Freitas, J.Germano, T.Almeida, L.Sousa, M.S.Piedade. IEEE Trans Magn., in press.3-“A new hand-held microsystem architecture for biological analysis”, M.Piedade, L.Sousa, T.M.Almeida, J.Germano, B.A.Costa, J.M.Lemos, IEEE Trans on Circuits and Systems, vol.53, pp.2384-2395, November 2006