Two of the greatest challenges in MEMS are those of packaging and integration with CMOS technology. Development of unreleased MEMS resonators at the transistor level of the CMOS stack will enable direct integration into front-end-of-line (FEOL) processing and minimal or no packaging, making these devices an attractive choice for on-chip signal generation.

Toward this goal, the authors have previously demonstrated the first fully unreleased MEMS resonator operating at 39 GHz with a quality factor (Q) of 129 ^[1] . The Si bulk acoustic resonator, surrounded on all sides by SiO₂, demonstrates the feasibility of unreleased resonators, providing a Q that is only 4x lower than its released counterpart ^[2] . At mm-wave frequency in the Landau-Rumer regime, resonator Q is limited primarily by anchor loss ^[3] . In the case of fully-clad resonators, the quality factor can be significantly improved by localization of acoustic energy using acoustic Bragg reflectors ^[4] .

The HybridMEMS lab has performed a study of fully unreleased resonator surrounded by lithographically defined ABRs, embedded in a homogeneous cladding layer (Figure 1). This one-mask design enables resonator banks of various frequencies on the same chip, providing multiple degrees of freedom in ABR design. With the goal of direct integration into FEOL CMOS processing, resonator performance is investigated for materials commonly found in the CMOS stack. The characteristics of these unreleased structures are compared with freely suspended resonators, released resonators isolated with lithographically defined ABRs ^[5] , and phononic crystal ^[6] based unreleased resonators (Figure 2).

1. W. Wang, L. C. Popa, R. Marathe, and D. Weinstein, “An unreleased mm-wave resonant body transistor,” IEEE MEMS Conference, 2011, pp. 1341-1344.
2. D. Weinstein and S. A. Bhave, “Acoustic resonance in an independent-gate FinFET,” Hilton Head Workshop, 2010, pp. 459-462.
3. R. Tabrizian, M. Rais-Zadeh, and F. Ayazi, “Effect of phonon interactions on limiting the f.Q product of micromechanical resonators,” IEEE Transducers Conference, 2009, pp. 2131-2134.
4. K. M. Lakin, “Thin film resonators and filters,” IEEE Ultrasonics Symposium, 1999, pp. 895-906.
5. R. H. Olsson, J. G. Fleming, and M. R. Tuck, “Contour mode resonators with acoustic reflectors,” US Patent 7385334 B1, 2008.
6. S. Mohammadi, A. A. Eftekhar, W. D. Hunt, and A. Adibi, “High-Q micromechanical resonators in a two-dimensional phononic crystal slab,” Applied Physics Letters, vol. 94, pp. 051906:1-3, Feb. 2009.

Figure 1: Scanning electron micrographs with uniform array of micropillars. The micropillars have the same height of 17 µm and pitch of 25 µm, while the diameters of pillars are (a) 6 µm, (b) 11 µm, and (c) 16 µm, respectively.

The mechanism of critical heat flux (CHF) is commonly attributed to two limits during boiling behavior: 1) the hydrodynamic limit due to Helmholtz instability and 2) the capillary limit determined by surface wettability ^[1] .  In recent years, a significant amount of research has been focused on CHF enhancement by utilizing micro/nanostructured surfaces to improve wettability ^{[2] [3] [4]} , with CHF of ~200 W/cm² being demonstrated ^[4] .  While most works are focused on making small structure sizes to improve surface wettability, the effect of this roughness-augmented wettability on CHF is poorly understood.  The limit of CHF enhancement with roughness-augmented wettability, where hydrodynamic instability becomes the dominant mechanism for CHF, has not been investigated.  In addition, boiling on nanostructured surfaces suffers from the requirement of high superheat due to bubble geometries closer to the homogeneous nucleation limit.  As a result, the heat transfer coefficient (HTC) on nanostructured surfaces is sacrificed ^[3] , which impairs the heat removal capability especially for applications demanding small temperature difference.

In this study, micro/nanopillar arrays are fabricated with a series of pitch and diameter size, as shown in Figure 1.  The sizes of pillar are designed to ensure that bubbles in the Cassie state, where vapor bubbles are suspended on the pillars, are energetically favorable such that bubble detachment is enhanced.  The series of sizes of the structured arrays generate various capillary forces, which allow the study on the mechanism for CHF and the limit of CHF enhancement with roughness-augmented wettability.  Furthermore, the investigation on surface roughness, where hydrodynamic instability dominants, gives the optimal size of structures for CHF enhancement and explores the feasibility of heterogeneous bubble nucleation on surfaces with proper structure geometry to reduce superheat.

1. S. G. Liter and M. Kaviany, “Pool-boiling CHF enhancement by modulated porous-layer coating: theory and experiment,” Int. J. Heat Mass Transfer, vol. 44, pp. 4287–4311, 2001.
2. C. Li and G. P. Peterson, “Parametric study of pool boiling on horizontal highly conductive microporous coated surfaces,” J. Heat Transfer, vol. 129, pp. 1465-1475, 2007.
3. S. Kim, H. D. Kim, H. Kim, H. S. Ahn, H. Jo, J. Kim, and M. H. Kim, “Effects of nano-fluid and surfaces with nano structure on the increase of CHF,” Exp. Therm Fluid Sci., vol. 34, pp. 487–495, 2010.
4. R. Chen, M.-C. Lu, V. Srinivasan, Z. Wang, H. H. Cho, and A. Majumdar, “Nanowires for enhanced boiling heat transfer,” Nano Lett., vol. 9, no. 2, pp. 548-553, 2009.

Significant interest exists in creating insect-based Micro-Air-Vehicles (MAVs) ^{[1] [2] [3]} that would combine advantageous features of insects—small size, effective energy storage, navigation ability—with the benefits of MEMS and electronics—sensing, actuation and information processing. The key part of the insect-based MAVs is the multi-site electrode, which interfaces with the nervous system of the insect to bias the insect’s flight path by controlling insect’s abdominal motions.

In this work, we have developed a flexible multi-site electrode (FME) for insect flight control that directly interfaces with the animal’s central nervous system as shown in Figure 1b. The FMEs are made of two layers of polyimide with gold sandwiched in-between in a split-ring geometry using standard MEMS processing ^[3] . The FMEs have a novel split-ring design that incorporates the anatomical bi-cylinder structure of the nerve cord of the Moth Manduca Sexta (Figure 1b) and allows for an efficient surgical process for implantation (Figure 1d). Additionally, we have integrated carbon nanotube (CNT)-Au nanocomposites into the FMEs to enhance the charge injection capability of the electrode.

To quantify the performance of the FME, we have developed a custom stimulation and measurement system that allows computer-controlled stimulation and automated image analysis of the resulting abdominal motion (Figure 2a). We measured the magnitude (r) and direction (q) of the abdominal movement by the position of the red dot on the abdomen tip of the moths (Figure 2 b) versus the stimulations signal delivered at various magnitude and across various site pair. Moreover, we measured the voltage and current across pairs of stimulation sites during stimulation signal delivery (Figure 2c); hence, we could estimate the power consumption and injection charge density for the FME stimulations. Finally, we have integrated the FMEs into a wireless system (Figure 1a). In the flight control experiment, we can force a freely flying animal to perform turning motions via the FME stimulations.

1. A. Bozkurt, F. Gilmour, and A. Lal, “Balloon-assisted flight of radio-controlled insect biobots,” IEEE Transactions on Biomedical Engineering, vol. 56, pp. 2304-2307, Sept. 2009.
2. H. Sato, C. W, Berry, Y. Peeri, E. Baghoomian, B. E. Casey, G. Lavella, J. M. VandenBrooks, J. F. Harrison, and M. M. Maharbiz, “Remote radio control of insect flight,” Frontiers in Integrative Neuroscience, vol 3, pp. 1-11, Oct. 2009.
3. W. M. Tsang, A. Stone, Z. Aldworth, D. Otten, T. Akinwande, T. Daniel, J. G. Hildebrand, R. Levine, and J. Voldman, “Remote control of a cyborg moth using carbon nanotube-enhanced flexible neuroprosthetic probe,” Proc. IEEE MEMS2010, pp. 39-42.

Microfabricated/microfluidic approaches to cell sorting, include purely dielectrophoretic (DEP) trap arrays ^[1] , passive hydrodynamic trap arrays with active DEP-based cell release ^[2] , and passive microwell arrays with optical cell release to permit sorting of non-adhered cells ^[3] .  As these proceeding technologies were best suited to operate with non-adherent cells, we are developing a solution for adherent cells.  Our approach to sorting adherent cells based on the morphological features uses a photolithography-inspired method, illustrated in Figure 1 (a). We first plated adherent cells into a dish and imaged cells using a microscope. Machine learning algorithm-based software CellProfiler ^[4] was used to quantitatively characterize the morphological features of the imaged cells, covering the cell area, shape, fluorescent intensity, and texture. As shown in Figure 1 (b), four classes were defined according to the fluorescent intensity difference in cell cellular compartments. A set of judging rules was generated by iteratively training the classifier based on hundreds of quantitative cell feature measurements to cluster the cells of similar phenotypes into a particular class. Desired cells were identified according to the classification. An alignment mark image was generated, with black features corresponding to locations of desired cells.  Aligning the transparency mask to the back of the cell culture dish showed that opaque mask features resided beneath desired cells. We then mixed a prepolymer solutizon consisting of cell culture media, a UV-photoinitiator, and poly(ethylene glycol) diacrylate (PEGDA) monomer. We added the prepolymer to the cell culture dish and shined ultraviolet (UV) light from a standard fluorescence microscope fluorescence source through the transparency and into the dish. The prepolymer then crosslinked into a hydrogel in all unmasked locations, encapsulating undesired cells. The desired cells, which were not encapsulated, can be enzymatically released from the substrate and recovered, as shown in Figure 2. The overall technique requires standard equipment found in biological labs and inexpensive reagents (<$10 per experiment), encouraging widespread adoption.

1. A. Rosenthal and J. Voldman, “Dielectrophoretic traps for single-particle patterning,” Biophys. J., vol.88, pp. 2193-2205, 2005.
2. B. M. Taff, S. P. Desai, and J. Voldman, “Electroactive hydrodynamic weirs for micro-particle manipulation and patterning,” Applied Physics Letters, vol. 94, no. 8, p. 084102, 2009.
3. J. R. Kovac and J. Voldman, “ Intuitive, image-based cell sorting using opto-fluidic cell sorting,” Analytical Chemistry, vol. 79, pp. 9321-9330, 2007.
4. T. R. Jones, A. E. Carpenter, M. R. Lamprech, J. Moffat, S. J. Silver, J. K. Grenier, A. B. Castoreno, U. S. Eggert, D. E. Root, P. Golland, and D. M. Sabatini, “Scoring diverse cellular morphologies in image-based screens with iterative feedback and machine learning,” PNAS, vol. 106, pp.1826-1831, 2009.

The development of new techniques to separate and characterize cells with high throughput has been essential to many of the advances in biology and biotechnology over the past few decades.  Continuing or improving upon this trend – for example, by developing new avenues for performing genetic and phenotypic screens – requires continued advancements in cell sorting technologies.  Towards this end, we are developing a novel method for the simultaneous separation and characterization of cells based upon their electrical properties.  This method, iso-dielectric separation (IDS), uses dielectrophoresis (the force on a polarizable object ^[1] ) and a medium with spatially varying conductivity to sort electrically distinct cells while measuring their effective conductivity (Figure 1).  It is similar to iso-electric focusing, except that it uses DEP instead of electrophoresis to concentrate cells and particles to the region in a conductivity gradient where their polarization charge vanishes ^{[2] [3] [4]} .

Previously, we have demonstrated the ability to perform continuous-flow, label-free, non-binary separations using IDS on a wide variety of cells and particles, while simultaneously extracting quantitative information from these samples as they are sorted ^[4] . In order to make IDS discovery more unknown cell types, dynamically changing the conductivity gradient is crucial for increasing the efficiency of finding the optimal separation condition. Therefore, we are developing a tri-syringe pump system to dynamically control conductivity gradients. We have verified the stability of the tri-syringe pump system via quantitative fluorescence imaging.  Combining this gradient control system with a computer-controlled function generator and automated microscope, we plan to fully automate IDS separation and electrical profile screening (Figure 2).

1. H. A. Pohl and J. S. Crane, “Dielectrophoresis of cells,” Biophysical Journal, vol. 11, pp. 711-727, 1971.
2. M. D. Vahey and J. Voldman, “An equilibrium method for continuous-flow cell sorting using dielectrophoresis,” Analytical Chemistry, vol. 80, no. 9, pp. 3135-3143, 2008.
3. M. D. Vahey and J. Voldman,  “Iso-dielectric separation: A new method for the continuous-flow screening of cells,” Micro Total Analysis Systems ’06, vol. 2, pp. 1058-1060, 2006.
4. M. D. Vahey and J. Voldman, “High-throughput cell and particle characterization using isodielectric separation,” Analytical Chemistry, vol. 81, no. 7, pp. 2446-2455, 2009.

Autocrine signaling is a mode of chemical signaling that occurs when cells can capture self-secreted diffusive factors.  Apart from its major role in sustaining cancer growth, autocrine signaling is also involved in the positive-feedback regulation of various physiological processes.  Due to the closed-loop nature and complex interplay of this signaling with other signaling cues, it is difficult to validate the presence and function of autocrine loops.  Studying these loops typically requires the use of specific inhibitors to perturb the underlying ligand/receptor pairing, limiting investigation of poorly characterized autocrine loops.

To promote the examination of autocrine signaling in broader biological systems, we have developed a general method for modulating autocrine activity using cell patterning.  In addition to capturing self-secreted ligands, cells with autocrine loops also acquire ligands from their neighbors.  By modulating the relative positioning between cells, we are able to modulate capture of autocrine ligands without needing specific inhibitors.  In particular, we use stencil cell patterning to organize cells as square-latticed arrays of circular patches of varying array spacing (Figures 1A & B).  We found that the cell-patterning platform can maintain uniform local cell density at all array spacings, in contrast to randomly plated cells, which exhibit increasing local cell density (Figure 1C).  By reducing the influence of these other environmental cues, we are able to more explicitly study the effect of autocrine signaling on cell phenotype.

In addition to studying the direct role of autocrine signaling, the cell-patterning platform can also be used to investigate the interplay of autocrine signaling with other signaling cues and to evaluate its contribution towards cell-to-cell variability.  To determine the concurrent role of cell-cell contacts, we can compare cell responses between single patches and multiple patches where the cell number of both designs is equal (Figures 2A & B). To evaluate the contribution of autocrine loops in causing cell-to-cell variability, we can determine how the inclusion of a large cell patch will perturb the response of an array of small patches (Figure 2C).  These innovative cell-patterning designs provide us novel tools for characterizing the impact of autocrine signaling without prior knowledge of the underlying ligand/receptor interactions.

Stem cell phenotype and function are influenced by microenvironmental cues that include cell-cell, cell-extracellular matrix (ECM), and cell-media interactions, as well as mechanical forces. Our research focuses on developing microscale systems for controlling the cellular microenvironment of mouse embryonic stem cells (mESCs), in particular mechanical forces (i.e., shear stress) and cell-media interactions (i.e., diffusible signaling).

Many emerging technologies used for ESC expansion or differentiation require perfusion culture, an example being pluripotent stem cell expansion in bioreactors for clinical applications ^[1] . We employ a multiplex microfluidic perfusion array to study the effects of shear stress on mESCs across a wide range of flow rates in a graded, quantitative manner. Using this device, we are able to show that perfusion elicits phenotypic changes and that the specific shear-responsive phenotype is due to mechanosensing by heparan sulfate proteoglycans (HSPGs, Figure 1A-C). This is the first study describing the ESC machinery capable of responding to shear stress, thus providing a foundation for further shear mechanotransduction studies ^[2] .

Cells are constantly secreting and responding to soluble signals, the removal of which can be mediated by modulating flow properties at the microscale. To assess the contribution of cell-secreted factors to mESC differentiation and self-renewal, we utilized a two-layer microfluidic perfusion device allowing for parallel comparison of different cell culture conditions (Figure 2A) ^[3] . Our results demonstrate that mESCs do not grow in differentiation conditions with minimal autocrine signaling, even with supplementation by Fgf4, a signal that has been shown to be a crucial factor in differentiation toward a neuronal stem cell fate (Figure 2B). Conversely, under self-renewal conditions, mESCs proliferate but lose self-renewal markers and upregulate differentiation markers (Figure 2C). These results, together with signaling and downstream differentiation assays, indicate that a differentiation towards an epiblast-like early differentiation state under conditions that had previously been shown as sufficient for self-renewal. Together, these results indicate the importance of cell-secreted signals for mESC growth and self-renewal.

1. D. Steiner, H. Khaner, M. Cohen, S. Even-Ram, Y. Gil, P. Itsykson, T. Turetsky, M. Idelson, E. Aizenman, R. Ram, Y. Berman-Zaken, and B. Reubinoff, “Derivation, propagation and controlled differentiation of human embryonic stem cells in suspension,” Nature Biotechnology, vol. 28, pp. 361-364, Mar. 2010.
2. Y.-C. Toh and J. Voldman, “Fluid shear stress primes mouse embryonic stem cells for differentiation in a self-renewing environment via heparan sulfate proteoglycans transduction,” The FASEB Journal, vol. 25, pp. 1208-1217,  2011.
3. K. Blagović, L.Y. Kim, A. M. Skelley, and J. Voldman, “Microfluidic control of stem cell diffusible signaling,” in Proc. Twelfth International Conference on Miniaturized Systems for Chemistry and Life Sciences, San Diego, USA, pp. 677-679, Oct. 2008.

The use of microsystems to manipulate and study cells in microenvironments is continually increasing.  However, along with such increase in usage is also a growing concern regarding the impact of these microsystems on cell physiology.  In this project, we are developing a set of cell-based fluorescent sensors to measure the impact of common stresses experienced in microsystems on cell physiology.  We are including stress agents commonly found in microsystems (e.g., UV exposure, heat shock, fluid flow, etc.).  Each sensor is designed to respond to one particular stress agent but can also be combined for multiplexed analysis of multiple stresses at once, as might be experienced in a typical microsystem. Designed to ease multiplexed analysis, each sensor will use different colors to both indicate the type of sensor and the strength of the signal.

One sensor in the system will be a heat shock sensor that responds to activation of the heat shock pathway, which is a generalized stress pathway in cells.  We are adapting a version of this sensor that we previously reported ^{[1] [2]} , which coupled fluorescent protein expression to activation of heat shock factor 1, from green fluorescent protein (EGFP) to a red fluorescent protein (RFP) and from red (DsRed) to yellow (YPet) for the constitutive color.  Figure 1 shows the heat shock sensor response to 15 min heating at 42 ºC.  Alongside this effort, we are using a multi-flow microfluidic device that can simultaneously apply different flows to cells across a 1000× range to understand the behavior of cells in flow ^[3] .  Figure 2 is an image of the multi-flow device used to test NIH3T3 mouse fibroblast cells.  Cells are seeded in 6 chambers concurrently and exposed to flow for 24 hrs, after which we can extract PCR from each chamber to study the cell response.

1. S. P. Desai and J. Voldman, “Cell-based sensors for quantifying the physiological impact of microsystems,” Integrative Biology, vol. 3, pp. 48-56, 2011.
2. S. P. Desai and J. Voldman, “Measuring the impact of dielectrophoresis on cell physiology using a high-content screening platform,” Micro Total Analysis Systems ’08, San Diego, CA, USA, 2008, pp. 1308-10.
3. Y.-C. Toh and J. Voldman, “Fluid shear stress primes mouse embryonic stem cells for differentiation in a self-renewing environment via heparan sulfate proteoglycans transduction,” Faseb Journal, vol. 25, pp. 1208-17, 2011.

Many immune responses are mediated by cell-cell interac­tions. In particular, cytotoxic T cells form conjugates with pathogenic and cancer cells in order to fight disease. More­over, T cell maturation and activation is governed by direct cell interactions with antigen-presenting cells (APCs). Er­rors in these processes can lead to the progression of severe diseases, such as multiple sclerosis (MS) and type 1 diabe­tes. The study of these intricate cell-cell interactions at the molecular scale is therefore crucial for understand­ing the dynamics and specificity of the immune response. One important feature of these interactions is the variability of response across populations. Cell-to-cell variability in pre­sumably homogeneous populations exposed to the same environmental conditions is ubiquitous, yet has long been neglected in immunology due to the limitations of conven­tional assay methods ^{[1] [2]} . Traditional methods to study cell-cell interactions, such as bulk measurements ^[3] or im­mobilization of cell pairs on a dish ^{[4] [5]} , suffer from both the inability to control cell-pairing at the single cell lev­el and the inability to study dynamic cell-cell interaction processes with high spatial and temporal resolution. We have overcome these limitations by developing a platform that can control cell pairing across thousands of individual immune cell pairs simultaneously while allowing visual­ization of the resulting responses. This approach enables us to quantify and understand variations in cell-cell inter­actions within large cell populations at the resolution of individual cell pairs. Previously, we developed a microflu­idic device with the capability to create thousands of such single cell pairs for the study of stem cell reprogramming (Figure 1, ^[6] ). To adapt the approach to work with smaller primary immune cells, we performed hydrodynamic mod­eling to guide redesign of the trap geometry (Figure 2). The modeling was used to determine how to adjust the trap ge­ometry to maximize flow through the center of the cups, which is crucial to the loading process. We determined that altering the cup-to-cup spacing transverse to the flow had the greatest impact on flow through the cups. We fabricated redesigned traps and are in the process of test­ing their pairing efficiency with primary immune cells.

1. S. L. Spencer, et al., “Non-genetic origins of cell-to-cell variability in TRAIL-induced apoptosis,” Nature, vol. 459, no. 7245, p. 428, 2009.
2. A. Colman-Lerner, et al., “Regulated cell-to-cell variation in a cell-fate decision system,” Nature, vol. 437,  no. 7059, p. 699, 2005.
3. S. Burgdorf, et al., “Distinct pathways of antigen uptake and intracellular routing in CD4 and CD8 T cell activation,” Science, vol. 316, no. 5824, p. 612, 2007.
4. M. S. Fassett, et al., “Signaling at the inhibitory natural killer cell immune synapse regulates lipid raft polarization but not class I MHC clustering,” Proc. National Academy of Sciences, vol. 98, no. 25, p. 14547, 2001.
5. X. Chen, et al., “Many NK cell receptors activate ERK2 and JNK1 to trigger microtubule organizing center and granule polarization and cytotoxicity, “Proc. National Academy of Sciences, vol. 104, no. 15, p. 6329, 2007.
6. A. M. Skelley, et al., “Microfluidic control of cell pairing and fusion,” Nature Methods, vol. 6, no. 2, p. 147, 2009.

A near-ultraviolet (UV) sensor based on zinc oxide (ZnO) nanowires (NWs) that is sensitive to photo excitation at or below 400-nm wavelength has been fabricated and characterized. The device uses a single optical lithography step, and the NWs are grown at a low temperature from solution. ZnO is a wide direct band gap (3.37 eV) semiconductor whose absorption edge is in the near-UV range, making it an ideal near-UV photodetector. This is the first reported ZnO NW near-UV sensor that is insensitive to visible light (visible blind) and fabricated using a low temperature solution process ^[1] . At a voltage bias of 1V across the device, a 29-fold increase in current is observed in comparison to dark current when the NWs are photo excited by 400-nm light-emitting diode (LED), 8.91 µA (photo excitation current) vs. 311 nA (dark current).

The fabrication of the near-UV sensor device is based on a single optical lithography step with no processing steps that exceed 100°C. The devices are compressed of a thin ZnO film with a metal cap. The sidewall of the ZnO film within the material stack acts as a seed for lateral growth of ZnO NWs. The metal cap restricts vertical growth of the NWs and doubles as the device electrodes. The symmetric devices have multiple electrode shapes and gaps between the electrodes ranging from 1-20 µm. The horizontally grown ZnO NWs bridge the gap between the two electrodes. The wires vary in length from 0.8 to 8.4 µm and diameter from 80 to 300 nm, depending on growth time. The result is a self-aligned ZnO NW ‘visible blind’ near-UV sensor that utilizes a low temperature process and a simple one-mask optical lithography step that can be integrated on a flexible substrate.

1. M. E. Swanwick, S. M.-L. Pfaendler, A. I. Akinwande, and A. J. Flewitt, “Near-ultraviolet sensor based on horizontal low temperature solution grown zinc oxide nanowires,” presented at 2010 MRS Fall Meeting, Boston, MA, Nov. 2010.