What are MEMS

1. What are MEMS? The term MEMS first started being used in the 1980’s. It is used primarily in the United States and is applied to a broad set of technologies with the goal of miniaturizing systems through the integration of functions into small packages. The fabrication technologies used to create MEMS devices is very broad based. The three most used fabrication technologies include Bulk Micro Machining, Surface Micro Machining and LIGA. There are a wide variety of materials and processes which are part of the MEMS industry. The graphic shows some of the variety found in MEMs systems Micro-pump – used to pump small amounts of fluid (all the way down to pico-liters) Micro-gear – this is a SEM (Scanning Electron Micrograph) of a Sandia Gear, each tooth is about 8um or the size of a human red blood cell Micro-mirror – used in telecommunications and also displays, one example here. Fluid Channel – well, if you have a micro pump, you need a fluid channel Heads Up display – the reason this is a MEMs device is because it utilizes micro-mirrors This could be used to navigate within a complex machine (aircraft) or building (refinary?) – would pop up the proper schematics, blue prints, etc. while you are actually working on a system. (Did anyone read Michael Crichton’s “Airframe?”). The term MEMS first started being used in the 1980’s. It is used primarily in the United States and is applied to a broad set of technologies with the goal of miniaturizing systems through the integration of functions into small packages. The fabrication technologies used to create MEMS devices is very broad based. The three most used fabrication technologies include Bulk Micro Machining, Surface Micro Machining and LIGA. There are a wide variety of materials and processes which are part of the MEMS industry. The graphic shows some of the variety found in MEMs systems Micro-pump – used to pump small amounts of fluid (all the way down to pico-liters) Micro-gear – this is a SEM (Scanning Electron Micrograph) of a Sandia Gear, each tooth is about 8um or the size of a human red blood cell Micro-mirror – used in telecommunications and also displays, one example here. Fluid Channel – well, if you have a micro pump, you need a fluid channel Heads Up display – the reason this is a MEMs device is because it utilizes micro-mirrors This could be used to navigate within a complex machine (aircraft) or building (refinary?) – would pop up the proper schematics, blue prints, etc. while you are actually working on a system. (Did anyone read Michael Crichton’s “Airframe?”).

2. What are Microsystems (MST)? Tiny, integrated, self-aware, stand-alone products, (based on microfabricated components) that can … Some benefits of Microfabricated Components: Small size Light weight Rugged Low power Batch fabricated Potential for low cost production Potential for high level of integration Parallel analyses Some benefits of Microfabricated Components: Small size Light weight Rugged Low power Batch fabricated Potential for low cost production Potential for high level of integration Parallel analyses

3. Microsystems and Nanotechnology In Europe, Microsystems is the term of choice. Also, Nanotechnology is often used interchangeably with Microsystems and MEMS. Hence the confusion. So… by the end of this discussion, you should be less confused! NASA used the term in 1998 to describe a satellite which weighs less than 10kg. Pico Satellites weigh less than 1kg. Femto Satellites weigh less than 0.1kg Micro Satellite < 100kg…So… by the end of this discussion, you should be less confused! NASA used the term in 1998 to describe a satellite which weighs less than 10kg. Pico Satellites weigh less than 1kg. Femto Satellites weigh less than 0.1kg Micro Satellite < 100kg…

4. MEMS Vs. Integrated Circuits (IC’s) One way to look at it: IC’s move and sense electrons MEMS move and sense mass Another: IC’s use Semiconductor processing technologies MEMS can use a variety of processes including Semiconductor but also Bulk, LIGA, Surface Micromachining… Packaging IC packaging consists of electrical connections in and out of a sealed environment MEMS packaging not only includes input and output of electrical signals, but may also include optical connections, fluidic capillaries, gas channels and openings to the environment. A much greater challenge. MEMS act as transducers (sensors) converting a physical property into an electrical property. MEMS can also actuate mechanical devices (switches, mirrors, etc…) IC’s can sense electrons and move them about, amplify, attenuate etc. IC’s and MEMS can be integrated on one chip if the processes are compatible. What differentiates many MEMS devices and products from IC’s is that the processes used to fabricate can be radically different and non-compatible. MEMS act as transducers (sensors) converting a physical property into an electrical property. MEMS can also actuate mechanical devices (switches, mirrors, etc…) IC’s can sense electrons and move them about, amplify, attenuate etc. IC’s and MEMS can be integrated on one chip if the processes are compatible. What differentiates many MEMS devices and products from IC’s is that the processes used to fabricate can be radically different and non-compatible.

5. MEMS and IC’s IC’s IC’s are based on the transistor – a basic unit or building block of IC’s. Most IC’s are Silicon based, depositing a relatively small set of materials. Equipment tool sets and processes are very similar between different IC fabricators and applications – there is a dominant front end technology base. MEMS Does not have a basic building block – there is no MEMS equivalent of a transistor. Some MEMS are silicon based and use sacrificial surface micromachining (CMOS based) technology. Some MEMS are hybrids (different wafer materials bonded), some are plastic based or ceramic utilizing a variety of processes – Surface & bulk micromachining, LIGA, electrodeposition, hot plastic embossing, extrusion on the micro scale etc. There is no single dominant front end technology base but emerging and established MEMS applications have started to “self-select dominant front-end technology pathways” (MANCEF 2nd Roadmap). So… MEMS is far more complex than IC’s, and hence, these applications need to draw from a large variety of technologies to be successful. So… MEMS is far more complex than IC’s, and hence, these applications need to draw from a large variety of technologies to be successful.

6. More on “What are MEMS?” MEMS devices first took off in the sensor industry. Most MEMS devices have at least one transducer element. To sense To actuate Transducer is a device or system that converts one form of energy to another – force to voltage, voltage to force, … Transducer is a device or system that converts one form of energy to another – force to voltage, voltage to force, …Transducer is a device or system that converts one form of energy to another – force to voltage, voltage to force, …

7. MEMS Applications Accelerometers (Inertial Sensors – “Crash Bags”, Navigation, Safety) Ink Jet Print Heads Micro Fluidic Pumps Insulin Pump (drug delivery) Pressure Sensor Auto and Bio applications Spatial Light Modulators (SLM’s) MOEM – Micro Optical Electro Mechanical Systems DMD – Digital Mirror Device DM – Deformable Mirror Chem Lab on a Chip Homeland security RF (Radio Frequency) MEMS Low insertion loss switches (High Frequency) Mass Storage Devices The list goes on…The list goes on…

8. MEMS Pressure Sensors Pressure Sensors 1960’s technology Used primarily in Aerospace industry at the beginning. Companies: Kulite Honeywell The term micromachining started being used in the 1960’s and 1970’s. Note – pressure sensors where considered a niche market in those days. When the automotive industry found that these sensors could help improve engine performance including gas mileage, these systems become more and more useful. Really took off in the 70’s when fuel economy starting becoming more important. From the Kulite.com web site: Kulite was founded in 1959 as the first commercial source of bare silicon strain gages. The piezoresistive silicon sensor is the heart of the Kulite Pressure Transducer. It has evolved from the simple bar gage sensor of the past, to a very high-tech, dielectrically isolated, silicon on silicon sensor we are using today. We have recently released the next stage of evolution for the dielectrically isolated, silicon on silicon sensor, which is referred to as the Leadless Sensor. The Leadless Sensor is a revolutionary design that allows the sensor to be used in high vibration, high acceleration and high temperature environments not possible in the past. Kulite is currently working on the development of silicon carbide piezoresistive pressure sensors and diamond piezoresistive pressure sensors The term micromachining started being used in the 1960’s and 1970’s. Note – pressure sensors where considered a niche market in those days. When the automotive industry found that these sensors could help improve engine performance including gas mileage, these systems become more and more useful. Really took off in the 70’s when fuel economy starting becoming more important. From the Kulite.com web site: Kulite was founded in 1959 as the first commercial source of bare silicon strain gages.

9. Pressure Sensors TRW Commercial Gas Engine Sensor - 1985 Top view of the TRW (1985) pressure sensor, the metal components are on top of the silicon membrane and are stressed when there is a pressure differential.Top view of the TRW (1985) pressure sensor, the metal components are on top of the silicon membrane and are stressed when there is a pressure differential.

10. Ink Jet Ink jet printers are MEMS based – late 1970’s, IBM and HP Thermal InkJet Technology was developed by HP in 1979 Thermal InkJet Technology was developed by HP in 1979

11. The Accelerometer 1987 TRW NovaSensor Accelerometer First generation inertial sensor Poppy seed is on top to show scale. First generation inertial sensor – TRW Lucas Nova Sensor – 1987 Poppy seed is on top to show scale. Combined standard CMOS technology with MEMS fabrication MEMS-based systems answered the call of government regulated passive restraints in automobiles where these systems sensed rapid deceleration and in the event of a collision sent a signal to inflate rapidly an airbag.First generation inertial sensor – TRW Lucas Nova Sensor – 1987 Poppy seed is on top to show scale. Combined standard CMOS technology with MEMS fabrication MEMS-based systems answered the call of government regulated passive restraints in automobiles where these systems sensed rapid deceleration and in the event of a collision sent a signal to inflate rapidly an airbag.

12. Increasingly Sophisticated Inertial Sensors Are Being Developed Analog Devices and Bosch are leaders in automotive inertial sensors. Berkeley is a leader in microsystems research at the University level. Inertial sensors measure a change in velocity (acceleration). The first and most prevalent of these is the crash sensor. A more recent application is in IBM’s ThinkPad Laptop. Analog Devices and Bosch are leaders in automotive inertial sensors. Berkeley is a leader in microsystems research at the University level. Inertial sensors measure a change in velocity (acceleration). The first and most prevalent of these is the crash sensor. A more recent application is in IBM’s ThinkPad Laptop.

13. Hard Drive Magnetic Read/Write Heads By incorporating MEMS actuation, the head can be positioned more quickly and to finer tolerances, this results in higher density data capability. By incorporating MEMS actuation, the head can be position more quickly and to finer tolerances, this results in higher density data capability.By incorporating MEMS actuation, the head can be position more quickly and to finer tolerances, this results in higher density data capability.

14. Micro Machines Surface Micromachining takes off in the 1990’s. These photos are from Sandia National Laboratories Surface micromachining leverage standard CMOS fabrication process technology (CMOS – Complimentary Metal Oxide Semiconductor, a silicon based semiconductor “standard” process). This basically consists of alternating layers of structural materials (poly crystalline silicon) and sacrificial layers (Silicon Dioxide). The sacrificial layer is a scaffold and acts as a temporary support and spacing material. The last step of the process is the “release” step, where the sacrificial layer is removed freeing the structural layers so they can move. Surface micromachining started out in Berkeley in the late 80’s. DARPA supported the MUMPS program starting in 1992 (Multi User MEMS Projects) at MCNC (Microelectronics Center of North Carolina) Surface micromachining leverage standard CMOS fabrication process technology (CMOS – Complimentary Metal Oxide Semiconductor, a silicon based semiconductor “standard” process). This basically consists of alternating layers of structural materials (poly crystalline silicon) and sacrificial layers (Silicon Dioxide). The sacrificial layer is a scaffold and acts as a temporary support and spacing material. The last step of the process is the “release” step, where the sacrificial layer is removed freeing the structural layers so they can move. Surface micromachining started out in Berkeley in the late 80’s. DARPA supported the MUMPS program starting in 1992 (Multi User MEMS Projects) at MCNC (Microelectronics Center of North Carolina)

15. MEMS as Machines MEMS are often referred to as Micro Machines. Tiny devices that move things. look at the complexity of the gears, hinges etc. Now compare that to the mite legs. This very primitive creature is by far more complex than the micro machine it is standing on. Try to estimate the size of various parts in the SEM (Scanning Electron Microscope) image by using the fact that the gear tooth is about 8um (microns) wide.look at the complexity of the gears, hinges etc. Now compare that to the mite legs. This very primitive creature is by far more complex than the micro machine it is standing on. Try to estimate the size of various parts in the SEM (Scanning Electron Microscope) image by using the fact that the gear tooth is about 8um (microns) wide.

16. MOEMs Micro Optical Electro Mechanical Systems TI started the development of DMD’s in the mid 1980’s. In 1996 the first commercial product was released. It is estimated by some that TI’s investment was on the order of $1 Billion. Today they command over $400M in revenue/yr – projected to grow from $800M to $1.8B between 2005 and 2010. Other applications under MOEMS include: DM – Deformable Mirror technology (AgilOptics) GLV – Grating Light Valve (Silicon Light Machines and Sony) Switching systems for telecommunications applicationsTI started the development of DMD’s in the mid 1980’s. In 1996 the first commercial product was released. It is estimated by some that TI’s investment was on the order of $1 Billion. Today they command over $400M in revenue/yr – projected to grow from $800M to $1.8B between 2005 and 2010. Other applications under MOEMS include: DM – Deformable Mirror technology (AgilOptics) GLV – Grating Light Valve (Silicon Light Machines and Sony) Switching systems for telecommunications applications

17. How Small are these Mirrors? DMD mirrors – complete DLP units have over 2 million mirrors – all functioning!DMD mirrors – complete DLP units have over 2 million mirrors – all functioning!

18. 1996 Micro Optics Bench Berkeley – 1996 There are two mirrors, three Fresnel lenses and at the far right a semiconductor laser (placed there after the optic fabrication).Berkeley – 1996 There are two mirrors, three Fresnel lenses and at the far right a semiconductor laser (placed there after the optic fabrication).

19. Additional Applications of MOEMS AgilOptic (formally Intellite) is an Albuquerque based company producing DM’s (deformable mirrors). These are used for image enhancement (taking out wavefront distortion of images which occur due to atmospheric disturbances, also when image the retina of the eye and adjusting focus as a laser welding beam goes through a plum of hot gas). The telecon boom infused a lot of capital and R&D money into MOEMS, but the telecon bust resulted in giving MEMS a bad name, for a short period of time.AgilOptic (formally Intellite) is an Albuquerque based company producing DM’s (deformable mirrors). These are used for image enhancement (taking out wavefront distortion of images which occur due to atmospheric disturbances, also when image the retina of the eye and adjusting focus as a laser welding beam goes through a plum of hot gas). The telecon boom infused a lot of capital and R&D money into MOEMS, but the telecon bust resulted in giving MEMS a bad name, for a short period of time.

20. Micro Needles MEMS needle within the opening of a small hypodermic needle Smaller size reduces pain and tissue damage – now there are much smaller MEMS needle arrays. The plastic needle array is made through a standard MEMS fabrication process to make the molds, micro injection process is used to create the arrays. Smaller size reduces pain and tissue damage – now there are much smaller MEMS needle arrays. The plastic needle array is made through a standard MEMS fabrication process to make the molds, micro injection process is used to create the arrays. Smaller size reduces pain and tissue damage – now there are much smaller MEMS needle arrays. The plastic needle array is made through a standard MEMS fabrication process to make the molds, micro injection process is used to create the arrays.

21. Biomedical Applications Bio apps – images obtained off the internet.Bio apps – images obtained off the internet.

22. Monolithically Integrated µChemLab™

23. BioMEMS The Overlap between microbiology and microsystem feature sizes makes integration between the two possible Bio-base scale.Bio-base scale.

24. Nanotechnology Meets MEMS The upper images shows carbon nano tubes linked to four electronic leads. The leads were made using standard semiconductor technology (1um leads created through deposition of metal by evaporation, subsequent patterning using lithography and etch to create the pattern). The carbon nanotubes where made separately and attached to t he leads. Lower half of the slide: Real Nano Tech – building structures atom by atom or molecule by molecule. The upper images shows carbon nano tubes linked to four electronic leads. The leads were made using standard semiconductor technology (1um leads created through deposition of metal by evaporation, subsequent patterning using lithography and etch to create the pattern). The carbon nanotubes where made separately and attached to t he leads. Lower half of the slide: Real Nano Tech – building structures atom by atom or molecule by molecule.

25. Cantilever Sensors The next few slides were taken from the Cornell Web site. Many researchers are working on Cantilever based sensors. As mass is added to the cantilever shifts the resonance frequency. The next few slides were taken from the Cornell Web site. Many researchers are working on Cantilever based sensors. As mass is added to the cantilever shifts the resonance frequency.

26. Detection of a single E.coli Cell Single Cell Detection with Micromechanical Oscillators B. Ilic, D. Czaplewski, M. Zalalutdinov, H. G. Craighead, P. Neuzil, C. Campagnolo, C. Batt The ability to detect small amounts of materials, especially pathogenic bacteria, is important for medical diagnostics and for monitoring the food supply. Engineered micro- and nanomechanical systems can serve as multi-functional, highly-sensitive, immunospecific biological detectors. We present a resonant frequency-based mass sensor, comprised of low-stress silicon nitride cantilever beams for the detection of Escherichia coli (E.coli) -cell-antibody binding events (Fig. 1) with detection sensitivity down to a single cell (Fig. 2). The binding events involved the interaction between anti-E. coli O157:H7 antibodies immobilized on a cantilever beam and the O157 antigen present on the surface of pathogenic E.coli O157:H7. Additional mass loading from the specific binding of the E. coli cells was detected by measuring a resonant frequency shift of the micromechanical oscillator (Fig. 3). From the measured resonant frequency spectra of the cantilever, in air, before and after antibody and cell attachment, we calculated that the mass of a single E. coli cell to be 665 fg, which is consistent with other reports and our estimated volume of this cell. In air, where considerable damping occurs, our device mass sensitivities for a 15µm and 25 µm long beam were 1.1 Hz/fg and 7.1 Hz/fg respectively. In both cases, utilizing thermal and ambient noise as a driving mechanism, the sensor was highly effective in detecting immobilized anti-E. coli antibody monolayer assemblies, as well as single E. coli cells. Our results suggest that tailoring of oscillator dimensions is a feasible approach for sensitivity enhancement of resonant mass sensors. Figure 1. 10µm square TappingMode AFM image of several E.coli cells. Analysis of several images measured average 350nm height for more than 50 cells. Figure 2. Scanning electron micrograph of E. Coli cell bound to immobilized antibody layer on ~5µm-wide, 15µm-long cantilever. Figure 3. Cantilever vibrations before ([black]_) and after ( [red]_) cell attachment in Figure 2. Resonance frequencies were determined using a Lorentzian least squares fit. Publication References: B. Ilic, D. Czaplewski, M. Zalalutdinov, H. G. Craighead, P. Neuzil, C. Campagnolo, and C. Batt, "Single Cell Detection With Micromechanical Oscillators", J. Vac. Sci. Technol. B, 19, 2825-2828 (2001). Single Cell Detection Page 2 of 2 http://www.hgc.cornell.edu/biomems/single%20cell%20detection.html 8/22/2006 Single Cell Detection with Micromechanical Oscillators B. Ilic, D. Czaplewski, M. Zalalutdinov, H. G. Craighead, P. Neuzil, C. Campagnolo, C. Batt The ability to detect small amounts of materials, especially pathogenic bacteria, is important for medical diagnostics and for monitoring the food supply. Engineered micro- and nanomechanical systems can serve as multi-functional, highly-sensitive, immunospecific biological detectors. We present a resonant frequency-based mass sensor, comprised of low-stress silicon nitride cantilever beams for the detection of Escherichia coli (E.coli) -cell-antibody binding events (Fig. 1) with detection sensitivity down to a single cell (Fig. 2). The binding events involved the interaction between anti-E. coli O157:H7 antibodies immobilized on a cantilever beam and the O157 antigen present on the surface of pathogenic E.coli O157:H7. Additional mass loading from the specific binding of the E. coli cells was detected by measuring a resonant frequency shift of the micromechanical oscillator (Fig. 3). From the measured resonant frequency spectra of the cantilever, in air, before and after antibody and cell attachment, we calculated that the mass of a single E. coli cell to be 665 fg, which is consistent with other reports and our estimated volume of this cell. In air, where considerable damping occurs, our device mass sensitivities for a 15µm and 25 µm long beam were 1.1 Hz/fg and 7.1 Hz/fg respectively. In both cases, utilizing thermal and ambient noise as a driving mechanism, the sensor was highly effective in detecting immobilized anti-E. coli antibody monolayer assemblies, as well as single E. coli cells. Our results suggest that tailoring of oscillator dimensions is a feasible approach for sensitivity enhancement of resonant mass sensors. Figure 1. 10µm square TappingMode AFM image of several E.coli cells. Analysis of several images measured average 350nm height for more than 50 cells. Figure 2. Scanning electron micrograph of E. Coli cell bound to immobilized antibody layer on ~5µm-wide, 15µm-long cantilever. Figure 3. Cantilever vibrations before ([black]_) and after ( [red]_) cell attachment in Figure 2. Resonance frequencies were determined using a Lorentzian least squares fit. Publication References: B. Ilic, D. Czaplewski, M. Zalalutdinov, H. G. Craighead, P. Neuzil, C. Campagnolo, and C. Batt, "Single Cell Detection With Micromechanical Oscillators", J. Vac. Sci. Technol. B, 19, 2825-2828 (2001). Single Cell Detection Page 2 of 2 http://www.hgc.cornell.edu/biomems/single%20cell%20detection.html 8/22/2006

27. Detection of Single DNA Enumeration of Single DNA Molecules Bound to a Nanomechanical Oscillator Bojan Ilic Resonant nanoelectromechanical systems (NEMS) are being actively investigated as sensitive mass detectors for applications such as chemical and biological sensing. We demonstrate that highly uniform arrays of nanomechanical resonators can be used to detect the binding of individual DNA molecules through resonant frequency shifts resulting from the added mass of bound analyte. Localized binding sites created with gold nanodots create a calibrated response with sufficient sensitivity and accuracy to count small numbers of bound molecules. The amount of nonspecifically bound material from solution, a fundamental issue in any ultrasensitive assay, was measured to be less than the mass of one DNA molecule, allowing us to detect a single 1587 bp DNA molecule. The drive toward ultra-sensitive biochemical assays has motivated significant efforts in single molecule detection and identification. Resonant nanomechanical devices [1-3] provide an alternative approach to techniques such as those using fluorescent labels. The mechanical approaches also have the possibility of quantification of the bound molecules, and can be incorporated in array-based systems for multiplexed biochemical analyses. Carbon nanotubes, attractive because of their uniform diameters and small mass have also been considered as biomolecular detectors, but remain difficult to incorporate in device architectures and have not yet been able to quantify specifically bound biomolecules. We have detected the binding of functionalized 1578 base pair long double-stranded deoxyribonucleic acid (dsDNA) molecules to nanomechanical oscillators by measuring the resonant frequency shift due to the added mass of the bound molecules. The binding of a single DNA molecule could readily be detected [4]. The resonant frequency of individual oscillators in an array of resonator devices was measured by thermo-optically driving the individual devices and detecting their motion by optical interference. The number of bound molecules on each device was quantified as proportional to the measured frequency shift with a proportionality constant determined experimentally and verified by modeling of the mechanical response of the system. For the smallest and most sensitive cantilevers the mass sensitivity was 194Hz/attogram. The resonant frequency shift of the oscillators can be measured with high accuracy, having a practical experimental uncertainty of ~10 Hz corresponding to ~0.05ag. The nonspecific binding of material to the oscillator throughout the process, however, limits the quantification of the specifically bound compounds for a particular analytical process. We measured the effects of non-specific binding of material other than the DNA from our solutions and found this to be approximately 0.43 ± 0.23ag for an oscillator of length L=3.5microns, with 0.23ag therefore being the approximate limiting mass resolution resulting from uncontrolled binding to the surface in our particular process. For the smallest (L=3.5micron), most sensitive oscillator this mass uncertainty corresponds to the mass of ~0.26 DNA molecules, enabling us to be able to resolve a single molecule. With the most sensitive devices and dilute DNA concentrations, we have detected a single dsDNA molecule. Figure 1: Micrographs (a & b) showing arrays of cantilevers of varying lengths. (c) SEM of the 90nm thick SiN cantilever with a 40nm circular Au dot. Figure 2: Schematic of the optical measurement setup and binding strategy of the thiolated dsDNA molecules to the Au dots. Figure 3: Frequency spectra before and after the binding events show a frequency shift due to a single dsDNA molecule bound to the Au surface of the cantilever. References: [1] "Nanoelectromechanical Systems", H. G. Craighead, Science, 290, 1532-1535 (2000). PART 1: Page 2 of 3 http://www.hgc.cornell.edu/Nems%20Folder/Enumeration%20of%20Single%20DNA.html 8/22/2006 [2] "Mechanical Resonant Immunospecific Biological Detector", B. Ilic, D. Czaplewski, H. G. Craighead, P. Neuzil, C. Campagnolo and C. Batt, Applied Physics Letters, 77, 450-452 (2000). [3] N. V. Lavrik, M. J. Sepaniak and P. G. Datskos, Rev. Sci. Inst. 75, 2229 (2004). [4] "Enumeration of DNA Molecules Bound to a Nanomechanical Oscillator", B. Ilic, Y. Yang, K. Aubin, R. Reichenbach, S. Krylov, and H. G. Craighead, Nano Letters, 5, 925-929 (2005). Enumeration of Single DNA Molecules Bound to a Nanomechanical Oscillator Bojan Ilic Resonant nanoelectromechanical systems (NEMS) are being actively investigated as sensitive mass detectors for applications such as chemical and biological sensing. We demonstrate that highly uniform arrays of nanomechanical resonators can be used to detect the binding of individual DNA molecules through resonant frequency shifts resulting from the added mass of bound analyte. Localized binding sites created with gold nanodots create a calibrated response with sufficient sensitivity and accuracy to count small numbers of bound molecules. The amount of nonspecifically bound material from solution, a fundamental issue in any ultrasensitive assay, was measured to be less than the mass of one DNA molecule, allowing us to detect a single 1587 bp DNA molecule. The drive toward ultra-sensitive biochemical assays has motivated significant efforts in single molecule detection and identification. Resonant nanomechanical devices [1-3] provide an alternative approach to techniques such as those using fluorescent labels. The mechanical approaches also have the possibility of quantification of the bound molecules, and can be incorporated in array-based systems for multiplexed biochemical analyses. Carbon nanotubes, attractive because of their uniform diameters and small mass have also been considered as biomolecular detectors, but remain difficult to incorporate in device architectures and have not yet been able to quantify specifically bound biomolecules. We have detected the binding of functionalized 1578 base pair long double-stranded deoxyribonucleic acid (dsDNA) molecules to nanomechanical oscillators by measuring the resonant frequency shift due to the added mass of the bound molecules. The binding of a single DNA molecule could readily be detected [4]. The resonant frequency of individual oscillators in an array of resonator devices was measured by thermo-optically driving the individual devices and detecting their motion by optical interference. The number of bound molecules on each device was quantified as proportional to the measured frequency shift with a proportionality constant determined experimentally and verified by modeling of the mechanical response of the system. For the smallest and most sensitive cantilevers the mass sensitivity was 194Hz/attogram. The resonant frequency shift of the oscillators can be measured with high accuracy, having a practical experimental uncertainty of ~10 Hz corresponding to ~0.05ag. The nonspecific binding of material to the oscillator throughout the process, however, limits the quantification of the specifically bound compounds for a particular analytical process. We measured the effects of non-specific binding of material other than the DNA from our solutions and found this to be approximately 0.43 ± 0.23ag for an oscillator of length L=3.5microns, with 0.23ag therefore being the approximate limiting mass resolution resulting from uncontrolled binding to the surface in our particular process. For the smallest (L=3.5micron), most sensitive oscillator this mass uncertainty corresponds to the mass of ~0.26 DNA molecules, enabling us to be able to resolve a single molecule. With the most sensitive devices and dilute DNA concentrations, we have detected a single dsDNA molecule. Figure 1: Micrographs (a & b) showing arrays of cantilevers of varying lengths. (c) SEM of the 90nm thick SiN cantilever with a 40nm circular Au dot. Figure 2: Schematic of the optical measurement setup and binding strategy of the thiolated dsDNA molecules to the Au dots. Figure 3: Frequency spectra before and after the binding events show a frequency shift due to a single dsDNA molecule bound to the Au surface of the cantilever. References: [1] "Nanoelectromechanical Systems", H. G. Craighead, Science, 290, 1532-1535 (2000). PART 1: Page 2 of 3 http://www.hgc.cornell.edu/Nems%20Folder/Enumeration%20of%20Single%20DNA.html 8/22/2006 [2] "Mechanical Resonant Immunospecific Biological Detector", B. Ilic, D. Czaplewski, H. G. Craighead, P. Neuzil, C. Campagnolo and C. Batt, Applied Physics Letters, 77, 450-452 (2000). [3] N. V. Lavrik, M. J. Sepaniak and P. G. Datskos, Rev. Sci. Inst. 75, 2229 (2004). [4] "Enumeration of DNA Molecules Bound to a Nanomechanical Oscillator", B. Ilic, Y. Yang, K. Aubin, R. Reichenbach, S. Krylov, and H. G. Craighead, Nano Letters, 5, 925-929 (2005).

28. Cantilever sensors Process used to make cantilever sensors – Cornell – Philip S. Waggoner Biological Detection via Resonant Nanoelectromechanical Sensors Philip S. Waggoner In this work, we have taken advantage of the mass-sensing capabilities of nanomechanical resonant devices and their high sensitivity for the detection of particular biological species. We functionalize these nanoscale devices with the appropriate receptor molecules specific to the species we desire to detect, in order to selectively weigh those analytes. From this information we can calculate analyte concentrations. Extending these devices into multiplexed arrays of separately functionalized detectors would allow for compact screening tests to detect disease markers and harmful biological substances or to track disease progression in a low-cost and rapid way. Nanoelectromechanical systems (NEMS) have attracted much attention as candidates for biological sensors due to their compact size, compatibility with semiconductor processing technologies, and sensitive frequency response to added mass [1]. These devices operate as miniaturized analogues of quartz crystal microbalances, exhibiting shifts in their resonant frequencies upon the addition of mass, such as bound biological species. Using nanoscale oscillators, resonant frequencies, and thus sensitivity to added mass, can be increased sufficiently for the detection of single viruses [2] or molecules [3]. Similar detection schemes include static deflection of microcantilevers resulting from induced surface stress of bound objects [4], but require multiple layers of fabrication, and much larger cantilevers, hindering integration in large scale arrays for detection of multiple analytes. For specific detection of biological objects, specificity is required, as resonant NEMS will respond to any adsorbed mass. We have worked to functionalize sensor surfaces using appropriate linking chemistries in order to present a surface of bound receptors, antibodies, for example, specific to a particular analyte. Reducing nonspecific binding of proteins to the nanomechanical biosensors is also an important consideration. With sufficient confidence in the specificity of the cantilever surface, in addition to appropriate controls, resonant frequency shifts can be attributed solely to adsorption and thus detection of the desired analyte. So far, we have measured concentrations as low as ~1 ng/mL using nanomechanical resonators. In this work, we have fabricated NEMS devices from low stress silicon nitride, with thicknesses on the order of 100 nanometers. Typical devices have lengths and widths on the order of a few microns. Derivatives of the standard cantilever beam shape, in addition to other interesting structures, are being investigated to improve sensitivity to low concentration detection. Resonant structures are fabricated by patterning the nitride device layer on a sacrificial layer of thermally oxidized silicon. Isotropic etching of the oxide releases the devices, making them free to oscillate. We use an optical system in order to thermally excite device oscillation and detect the resonant frequencies from modulation of the light reflected from the oscillating structure [4]. Future work will utilize arrays of NEMS biosensors for multiplexed biological detection from a single sample. Biological Detection via Resonant Nanoelectromechanical Sensors Philip S. Waggoner In this work, we have taken advantage of the mass-sensing capabilities of nanomechanical resonant devices and their high sensitivity for the detection of particular biological species. We functionalize these nanoscale devices with the appropriate receptor molecules specific to the species we desire to detect, in order to selectively weigh those analytes. From this information we can calculate analyte concentrations. Extending these devices into multiplexed arrays of separately functionalized detectors would allow for compact screening tests to detect disease markers and harmful biological substances or to track disease progression in a low-cost and rapid way. Nanoelectromechanical systems (NEMS) have attracted much attention as candidates for biological sensors due to their compact size, compatibility with semiconductor processing technologies, and sensitive frequency response to added mass [1]. These devices operate as miniaturized analogues of quartz crystal microbalances, exhibiting shifts in their resonant frequencies upon the addition of mass, such as bound biological species. Using nanoscale oscillators, resonant frequencies, and thus sensitivity to added mass, can be increased sufficiently for the detection of single viruses [2] or molecules [3]. Similar detection schemes include static deflection of microcantilevers resulting from induced surface stress of bound objects [4], but require multiple layers of fabrication, and much larger cantilevers, hindering integration in large scale arrays for detection of multiple analytes. For specific detection of biological objects, specificity is required, as resonant NEMS will respond to any adsorbed mass. We have worked to functionalize sensor surfaces using appropriate linking chemistries in order to present a surface of bound receptors, antibodies, for example, specific to a particular analyte. Reducing nonspecific binding of proteins to the nanomechanical biosensors is also an important consideration. With sufficient confidence in the specificity of the cantilever surface, in addition to appropriate controls, resonant frequency shifts can be attributed solely to adsorption and thus detection of the desired analyte. So far, we have measured concentrations as low as ~1 ng/mL using nanomechanical resonators. In this work, we have fabricated NEMS devices from low stress silicon nitride, with thicknesses on the order of 100 nanometers. Typical devices have lengths and widths on the order of a few microns. Derivatives of the standard cantilever beam shape, in addition to other interesting structures, are being investigated to improve sensitivity to low concentration detection. Resonant structures are fabricated by patterning the nitride device layer on a sacrificial layer of thermally oxidized silicon. Isotropic etching of the oxide releases the devices, making them free to oscillate. We use an optical system in order to thermally excite device oscillation and detect the resonant frequencies from modulation of the light reflected from the oscillating structure [4]. Future work will utilize arrays of NEMS biosensors for multiplexed biological detection from a single sample.

29. Mass Storage - IBM Mass Storage – IBM IBM’s “Millipede” 100 Tera Bit per square inch! This device is due out as a commercial product in 2007. It works by making small indentations in a polymer film. Can read and write Writes divot into polymer by heating tip to 400C Reads by looking at surface with 300C tip (measures resistance change with temp drop) – if the tip is in a divot, the tip cools more than if it is not – therefore, there is a change in resistivity which is measured by the electronics. Erases by making an offset pit, which causes the nearby pit to “pop up” and hence erases it. Principle: Thermomechanical Local Probe Ultimate Density – 1nm bit indentation/pitch = 1Tb/in^2 - Terabit Milestone – Has been Demonstrated 20x higher than the densest magnetic storage currently available! 2002: 100-200Gb/in^2 with the Millipede 32x32 array (1024) - about ~10-15Gbytes Working Prototype demonstrated June, 2002. Resonant frequency limits the data transmission rate of a single cantilever to a few Mb/s, three orders of magnitude (1000x) slower than magnetic read/write systems. - Many cantilevers – faster speeds. Technological background http://www.physorg.com/news3361.html At the heart of the "millipede" technology is a two-dimensional array of V-shaped silicon cantilevers, each 70 micrometers (thousandths of a millimeter) long. At the end of each cantilever there is apart from the tip a micrometer-sized sensor for reading as well as a heating resistor above the tip, which is needed for writing. The cone shaped tip is just under one micrometer in length and has a radius of a few nanometers at its apex. The cantilever cells are arranged in the form of an array on a 10 mm x 10 mm chip. One of the recent array designs comprises a total of 4,096 (64 x 64) cantilevers. The MEMS elements are etched out of a silicon single crystal using existing technologies. The actual data medium is a thin polymer film coated on a silicon substrate. The tips can independently read, write or erase the bits.  Nobel Laureate Gerd Binnig – One of the drivers of Millipede: “Since nanometer-scale tip can address individual atoms, we anticipate further improvements far beyond even this fantastic terabit milestone” “While current storage technologies may be approaching their fundamental limits, this nanomechanical approach is potentially valid for a thousand-fold increase in data storage density.”Mass Storage – IBM IBM’s “Millipede” 100 Tera Bit per square inch! This device is due out as a commercial product in 2007. It works by making small indentations in a polymer film. Can read and write Writes divot into polymer by heating tip to 400C Reads by looking at surface with 300C tip (measures resistance change with temp drop) – if the tip is in a divot, the tip cools more than if it is not – therefore, there is a change in resistivity which is measured by the electronics. Erases by making an offset pit, which causes the nearby pit to “pop up” and hence erases it. Principle: Thermomechanical Local Probe Ultimate Density – 1nm bit indentation/pitch = 1Tb/in^2 - Terabit Milestone – Has been Demonstrated 20x higher than the densest magnetic storage currently available! 2002: 100-200Gb/in^2 with the Millipede 32x32 array (1024) - about ~10-15Gbytes Working Prototype demonstrated June, 2002. Resonant frequency limits the data transmission rate of a single cantilever to a few Mb/s, three orders of magnitude (1000x) slower than magnetic read/write systems. - Many cantilevers – faster speeds. Technological background http://www.physorg.com/news3361.html

30. http://www.nanochip.com/tech.htm Mass Storage - Nanochip Nanochip Technology The core technology contained in our memory chips is created by the use of arrays of atomic force probe tips to write, read, and record data bits in a phase-change memory medium.  Please reference below for a scanning electron microscope photo of one of our tips on the end of a cantilever.  This tip has a radius of about 25 nm and when it comes in contact with the phase change medium a current is passed through the tip into the recording layer.  This current heats the recording layer and is either slowly cooled to form a crystalline bit or rapidly cooled to form an amorphous bit.  In the reading mode, a low level current is passed through the tip to the media, not heating the media, but instead is used to the sense the high resistance amorphous state or the low resistance crystalline state. The structure of the assembled Nanochip is shown in the artist drawing.  The Nanochips are assembled by wafer bonding a media wafer to a tip-array wafer.  As seen in the drawing below, after the wafer bonding process, the Nanochips will be diced and packaged into standard removable memory card packages, such as, Compact Flash, Secure Digital (SD), MMC, Memory Stick, or USB drives.  Writing, reading, and erasing are done while scanning the tip array or the media platform in X and Y. This allows the tips to trace out a raster pattern across the media under each tip, similar to the raster scan pattern used by the electron beam in TVs.Aerial Density:  All storage devices are ultimately limited in capacity by the aerial density of the bits stored, i.e. the number of bits per square inch that can be stored on a disc or in the total area of a semiconductor chip.  In the removable storage chip market today, NAND flash is clearly the market leader in both chip volume and cost per Megabyte.  The latest NAND flash chips use 70 nm lithography to define their bit cells and they typically store two data bits per cell using Multi-Level technology.  With the present Nanochip probe tip technology used we typically record a single bit of data in a 15 nm by 15 nm area. The Nanochip scanning probe technology has a growth path that will lead us in the future to bit cells as small as 5 nm (about 400 times denser than present NAND flash chips). Manufacturing cost:  Today Nanochip uses one micron semiconductor fabs to make our MEMs chips.  This type of equipment was used over ten years ago for most semiconductor products.  Therefore, the cost of building a MEMs fab to make our chips is in the tens of millions of dollars, unlike the several billion dollars needed to make a 70 nm semiconductor fabrication facility.  Furthermore, no matter what aerial density we achieve for the future, we will save enormous capital manufacturing costs since we will use much less expensive fabrication tools and facilities. Interfaces:  Nanochip will use standard NAND flash interfaces so that our customers and the end users can use our products in all consumer electronic devices that presently use NAND flash memory cards. Nanochip Technology The core technology contained in our memory chips is created by the use of arrays of atomic force probe tips to write, read, and record data bits in a phase-change memory medium.  Please reference below for a scanning electron microscope photo of one of our tips on the end of a cantilever.  This tip has a radius of about 25 nm and when it comes in contact with the phase change medium a current is passed through the tip into the recording layer.  This current heats the recording layer and is either slowly cooled to form a crystalline bit or rapidly cooled to form an amorphous bit.  In the reading mode, a low level current is passed through the tip to the media, not heating the media, but instead is used to the sense the high resistance amorphous state or the low resistance crystalline state. The structure of the assembled Nanochip is shown in the artist drawing.  The Nanochips are assembled by wafer bonding a media wafer to a tip-array wafer.  As seen in the drawing below, after the wafer bonding process, the Nanochips will be diced and packaged into standard removable memory card packages, such as, Compact Flash, Secure Digital (SD), MMC, Memory Stick, or USB drives.  Writing, reading, and erasing are done while scanning the tip array or the media platform in X and Y. This allows the tips to trace out a raster pattern across the media under each tip, similar to the raster scan pattern used by the electron beam in TVs.

31. What is a Cantilever?

32. Cantilever Cantilevers are used as Sensors Cantilevers are used as Switches Many MEMS Sensors use the principles of Cantilevers as well as RF Swtiches

33. Cantilever A cantilever is supported at one end (fixed). It has a length, thickness and width (geometry) When a force is applied to the end, it deflects

34. Cantilevers as Sensors As sensors, Cantilevers can react to the environment in two ways: The resonance frequency can shift (due to a change in loading mass) The deflection can shift (due to stress)

35. Common Observations Consider this about a diving board What happens when a little kid bounces on the end of the diving board? What happens when his large dad bounces on the end of the diving board? Which one has a higher resonance frequency?

36. MEMS Cantilever sensors In MEMS Cantilever sensors, the ends of the cantilevers are coated with a layer of probe molecules. When a target molecule is present, it attaches to the probe molecule, thereby increasing the mass. The resonant frequency goes down. You just detected the presence of a molecule!

37. Actual System

39. Resonance Shift

40. Resonance Frequency Shift as a Function of Mass Have the students plot this data!Have the students plot this data!

42. Two Concepts of Cantilevers as Sensors Response to Stress Use a laminate cantilever of dissimilar materials. One material expands or contracts at a different rate as another due to absorption, adsorption, heat… Resulting stress gradient (difference in stress) causes the cantilever to bend. Response to Mass Cantilevers are coated with a material which is selective to what can adhere to it. When the target material adheres to the cantilever, its mass changes resulting in a shift of the cantilevers resonance (natural) frequency.

43. Applications of MEMS cantilever beams

48. E. ColiHow big is this? See the scaleSee the scale