On Modeling the Single Motor Unit Action Potential

A model has been proposed for the generation of single motor unit potentials routinely observed in the clinical EMG examination of the normal biceps brachii muscle. A dipole representation was chosen for the single fiber activity. The motor unit was constructed from a uniform random array of single fibers. Motor unit potentials generated by this array have been observed at various distances both inside and outside the array. The effects of single fiber dipole axial dispersions on the potentials observed at increasing distances from the array have also been investigated. Motor unit potentials generated by the model have been compared with existing data from multielectrode studies in the biceps brachii.     

An Integrated-Circuit Approach to Extracellular Microelectrodes

This paper describes a new multielectrode microprobe which utilizes integrated-circuit fabrication techniques to overcome many of the problems associated with conventional microelectrodes. The probe structure consists of an array of gold electrodes which are supported on a silicon carrier and which project beyond the carrier for a distance of about 50 ? to allow a close approach to active neurons. These electrodes are covered with a thin (0.4-?) layer of silicon dioxide which is selectively removed at the electrode tips using high-resolution photoengraving techniques to define the recording areas precisely. The processing sequence described permits any two-dimensional electrode array to be realized. Interelectrode spacings can be accurately controlled in the range from 10 to 20 ? or larger, and electrode-tip diameters can be as small as 2 ?.     

Influence of tissue inhomogeneities on noninvasive muscle fiberconduction velocity measurements-investigated by physical and numericalmodeling

The determination of conduction velocity in the muscle fibers of single motor units from noninvasive recordings of single motor unit action potentials can be improved by the method of spatially filtering multielectrode EMG. The use of this conduction velocity as a diagnostic tool requires a high reliability of the detected values. However, experiments did reveal that the measured conduction velocity values showed remarkably high fluctuations depending on the recording site along the muscle fibers which could not be attributed to the influence of the endplate and tendon region. The present work examines the hypothesis that the observed fluctuations in propagation velocity were caused by electrically inhomogeneous tissue, regions of different electrical conductivity which are located between the excited muscle fibers and the recording electrodes and which cause a deformation of the extracellular electric current field. The investigation was performed by means of a physical model as well as by finite element model calculations. In both models single, simple shaped (cylindrical) inhomogeneity regions with a conductivity of 0.1 to 10 times that of the surrounding medium and diameters ranging between 1.6 and 2.7 mm were placed between excitation sources and recording site. The results indicate that the observed conduction velocity fluctuations of up to some 10% can be well attributed to inhomogeneity effects of the tissue conductivity. Based on these results, one may look for signal processing methods to cut down such fluctuations in conduction velocity measurements.     

高精度长线列密排光纤阵列的制作研究

本文讨论了利用硅V型槽法制作高精度线性光纤阵列的可行性,介绍了硅V型槽制作机理,对影响器件精度的主要因素进行了分析,并在器件设计、制作中给予充分考虑。根据器件的可靠性要求,分析了用于光纤粘接的粘接剂应满足的特性,并对紫外固化胶和红外粘接剂进行实验对比,结果表明,Norland紫外固化胶和353ND红外粘接剂为最佳选择。用各向异性湿法腐蚀技术制作了硅V型槽,进行了光纤排列、定位及端面处理,制作出了高精度线性光纤阵列。对端面面型误差和表面粗糙度的测试后,结果表明光纤阵列端面纵向位置误差≤324 nm,表面粗糙度均方根值小于3.85nm。     

A high-yield IC-compatible multichannel recording array

This paper reports the development of a multielectrode recording array for use in studies of information processing in the central nervous system and in the closed-loop control of neural prostheses. The probe utilizes a silicon supporting carrier which is defined using a deep boron diffusion and an anisotropic etch stop. This substrate supports an array of polysilicon or tantalum thin-film conductors insulated above and below with silicon nitride and silicon dioxide. Typical probe dimensions include a length of 3 mm, shank width of 50 µm, and a thickness of 15 µm. These structures are capable of simultaneous high-amplitude multichannel recording of neural activity in the cortex. The probe fabrication process requires only four masks and is single-sided using wafers of normal thickness, resulting in yields which exceed 80 percent. The process is also compatible with the inclusion of on-chip MOS circuitry for signal amplification and multiplexing. A complete ten-channel signal processor which requires only three external probe leads is being developed.     

A novel electrode array for diameter-dependent control of axonal excitability: a simulation study

Electrical extracellular stimulation of peripheral nerve activates the large-diameter motor fibers before the small ones, a recruitment order opposite the physiological recruitment of myelinated motor fibers during voluntary muscle contraction. Current methods to solve this problem require a long-duration stimulus pulse which could lead to electrode corrosion and nerve damage. The hypothesis that the excitability of specific diameter fibers can be suppressed by reshaping the profile of extracellular potential along the axon using multiple electrodes is tested using computer simulations in two different volume conductors. Simulations in a homogenous medium with a nine-contact electrode array show that the current excitation threshold (Ith) of large diameter axons (13-17 microm) (0.6-3.0 mA) is higher than that of small-diameter axons (2-7 microm) (0.4-0.7 mA) with 200-microm axon-electrode distance and 10-micros stimulus pulse. The electrode array is also tested in a three-dimensional finite-element model of the sacral root model of dog (ventral root of S3). A single cathode activates large-diameter axons before activating small axons. However, a nine-electrode array activates 50% of small axons while recruiting only 10% of large ones and activates 90% of small axons while recruiting only 50% of large ones. The simulations suggest that the near-physiological recruitment order can be achieved with an electrode array. The diameter selectivity of the electrode array can be controlled by the electrode separation and the method is independent of pulse width.     

A three-dimensional architecture for a parallel processingphotosensing array [silicon retina application]

A three-dimensional architecture for a photosensing array has been developed. This silicon based architecture consists of a 10 x 10 array of photosensors with 80 microns diameter, through chip interconnects to the back side of a 500 microns thick silicon wafer. Each photosensor consists of a 300 x 300 microns pn-junction photodiode. The following processes were used to create this photosensing architecture: 1) thermomigration of aluminum pads through an n-type silicon wafer; 2) creation of pn-junction photosensors on one side of the wafer; and 3) creation of aluminum pad ohmic contacts to the thermomigrated, through chip interconnects and the substrate on the back side of the wafer. The electrical and optical characteristics of the three-dimensional architecture indicates that it should be well suited as a photosensing framework around which a "silicon retina" could be built.     

Performance Improvement of a Power Conversion Module by Liquid Micro-Jet Impingement Cooling

Liquid micro-jet array impingement cooling of a power conversion module with 12 power switching devices (six insulated gate bipolar transistors and six diodes) is investigated. The 1200-V/150-A module converts dc input power to variable frequency, variable voltage three-phase ac output to drive a 50HP three-phase induction motor. The silicon devices are attached to a packaging layer [direct bonded copper (DBC)], which in turn is soldered to a metal base plate. DI water micro-jet array impinges on the base plate of the module targeted at the footprint area of the devices. Although the high heat flux cooling capability of liquid impingement is a well-established finding, the impact of its practical implementation in power systems has never been addressed. This paper presents the first one-to-one comparison of liquid micro-jet array impingement cooling (JAIC) with the traditional methods, such as air-cooling over finned heat sink or liquid flow in multi-pass cold plate. Results show that compared to the conventional cooling methods, JAIC can significantly enhance the module output power. If the output power is maintained constant, the device temperature can be reduced drastically by JAIC. Furthermore, jet impingement provides uniform cooling for multiple devices placed over a large area, thereby reducing non-uniformity of temperature among the devices. The reduction in device temperature, both its absolute value and the non-uniformity, implies multi-fold increase in module reliability. The results thus illustrate the importance of efficient thermal management technique for compact and reliable power conversion application     

Toward the neurocomputer: image processing and pattern recognition with neuronal cultures

Information processing in the nervous system is based on parallel computation, adaptation and learning. These features cannot be easily implemented on conventional silicon devices. In order to obtain a better insight of how neurons process information, we have explored the possibility of using biological neurons as parallel and adaptable computing elements for image processing and pattern recognition. Commercially available multielectrode arrays (MEAs) were used to record and stimulate the electrical activity from neuronal cultures. By mapping digital images, i.e., arrays of pixels, into the stimulation of neuronal cultures, a low and bandpass filtering of images could be quickly and easily obtained. Responses to specific spatial patterns of stimulation were potentiated by an appropriate training (tetanization). Learning allowed pattern recognition and extraction of spatial features in processed images. Therefore, neurocomputers, (i.e., hybrid devices containing man-made elements and natural neurons) seem feasible and may become a new generation of computing devices, to be developed by a synergy of Neuroscience and Material Science.     

Optical frequency spacing tunable four-channel integrated 1.55 mu m multielectrode distributed-feedback laser array

A four-channel integrated 1.55- mu m multielectrode distributed-feedback (DFB) laser array was fabricated using the metallorganic vapor-phase deposition/liquid-phase epitaxy (MOVPE/LPE) hybrid method. Simultaneous single-longitudinal mode operation was achieved in each multielectrode DFB laser on a single chip. Utilizing the frequency tunability of multielectrode DFB lasers, optical frequency spacings were controlled and set to within a few gigahertz. The drift of frequency spacings due to temperature fluctuation was in the range of +or-50 MHz for temperature control of +or-0.1 degrees .<>     

Methods for passive fiber chip coupling of integrated opticaldevices

A useful technique for high precision passive coupling of single mode optical fibers to integrated optical devices is crucial for cost effective packaging especially in multiport devices like switches (N×N) and other WDM components. These devices were fabricated on two different material bases, silicon on insulator (SOI) and polymers. In both cases the waveguides are based on the oversized rib waveguide concept and utilize silicon as a substrate. Two possible fabrication processes for this passive fiber chip coupling IN or ON silicon are presented and compared. The first approach involves a technology similar to flip chip fabrication using a sub- and superstrate, that allows separate processing of v-grooves for fiber alignment and the integrated optical devices. The self aligned mounting of the chip is achieved by a v-shaped rib-groove combination created by wet chemical etching, where the rib is the exact negative of the groove so that the flip chip is put on precisely defined crystal planes rather than on sensitive edges, which would be the case when using rectangular alignment ribs. The second approach utilizes the same chip for waveguides and fiber alignment structures which makes it possible to define both in the same lithographic step and thereby eliminating any vertical displacement. Processing difficulties arise primarily from completely different processing requirements of fiber aligning v-grooves and integrated waveguides. The need to define patterns of the size of only several microns (μm) in the proximity to deep grooves makes the use of an electrophoretic photoresist necessary that is deposited via galvanic means on the extremely nonplanar surface. Both processes allow for fiber chip alignment precisions in the sub-μm range which was also experimentally verified with coupling losses as low as 0.7 dB per end-face. The fabrication processes along with experimental and theoretical results are presented     

8×8 array of thin-film photodetectors vertically electricallyinterconnected to silicon circuitry

This paper reports the integration of an 8×8 array of thin-film GaAs-AlGaAs photodetectors onto a silicon-oscillator array circuit for massively parallel-image processing applications. Each detector was electrically connected to the oscillator below it using vertical electrical interconnections. Both sides of the thin-film devices were metallized for electrical contact, which minimized the interconnection density on the silicon circuit, thereby maximizing the available signal processing area. The yield of this integrated array and associated circuit was 100%, with the majority of pixels demonstrating a dynamic range of 50 dB     

A Two-Wafer Approach for Integration of Optical MEMS and Photonics on Silicon Substrate

This letter reports a novel two-wafer approach which demonstrates an integration of optical microelectromechanical system (MEMS) devices and photonics on a silicon substrate. The great advantage of this novel wafer bonding scheme is the ability to maintain the optical axis of the optical MEMS device at the same axis as the optical components. The bonded two wafers which are partially processed, which allows for further processing on the wafer after bonding. Thus, the critical alignment issue is resolved for devices requiring precise alignment in x-/y-/z-axis. Individual functionalities of optical MEMS device and optical coupling between silicon waveguide, fibers and ball lens are demonstrated. This technology shows the potential for integrating silicon photonics integrated circuit and MEMS components with reconfiguration functions on a single silicon substrate.      

Selectivity of multiple-contact nerve cuff electrodes: a simulation analysis

Advances in functional neuromuscular stimulation (FNS) have increased the need for nerve cuff designs that can control multiple motor functions through selective stimulation of selected populations of axons. This selectivity has proved to be difficult to achieve. Recent experiments suggest that it is possible to slowly reshape peripheral nerve without affecting its physiological function. Using computer simulations we have tested the hypothesis that changing the cross section of a nerve from a round to a flat configuration can significantly improve the selectivity of a nerve cuff. We introduce a new index to estimate selectivity to evaluate the various designs. This index is based on the ability of a nerve electrode to stimulate a target axon without stimulating any other axons. The calculations involve a three-dimensional finite element model to represent the electrical properties of the nerve and cuff and the determination of the firing properties of individual axons. The selectivity rating was found to be significantly higher for the Flat Cuff than the Round Cuff. The result was valid with uniform or random distribution of axons and with a random distribution of fascicles diameters. Flattening of individual fascicles also improved the selectivity of the Flat Cuff but only when the number of contacts used was increased to maintain uniform contact density. Therefore, cuff designs that can reshape the nerve into flatter configurations should yield better cuff performance than the cylindrical cuffs but will require higher contact density.     

The HARPSS process for fabrication of precision MEMS inertial sensors

The high aspect-ratio combined poly- and single-crystal silicon micromachining technology (HARPSS) and its application to fabrication of precision MEMS inertial sensors are presented. HARPSS is a single wafer, all silicon, front-side release process which is capable of producing 10–100's of microns thick, electrically isolated, 3-D poly- and single-crystalline silicon microstructures with various size air-gaps ranging from sub-micron to tens of microns. High aspect-ratio (>50:1) polysilicon structures are created by refilling 100's of microns deep trenches with polysilicon deposited over a sacrificial oxide layer. This technology provides features required for precision micromachined inertial sensors. The all-silicon feature of this technology improves long term stability and temperature sensitivity while fabrication of large area, vertical electrodes with sub-micron gap spacing will increase the sensitivity by orders of magnitude.     

Measurements of the Uptake Area of Small-Size Electromyographic Electrodes

Extracellular action potentials from 76 different muscle fibers in the human brachial biceps were recorded, with a 14 lead multielectrode, each leading-off surface being 25 ?m in diameter. Volume conduction of these action potentials was calculated by representing the low-pass filter characteristics of the muscle tissue by a transfer function with one time constant and an attenuation factor. The radial decline of the action potentials was calculated in steps of 5 ?m, and the pickup radius of the electrode, defined as the distance at which the peak-to-peak amplitude of the action potentials declines to 200 ?V, was computed. The pickup radius for the 25 ?m diameter electrode, assuming an average action potential peak-to-peak amplitude of 6.2 mV was 292 Mm. With this uptake area, a fiber density in the brachial biceps of 1.37 fibers/uptake area (the average number of fibers belonging to one motor unit and included within the electrode uptake area) and a fiber radius of 28 ?m, the electrode "sees" 34 different motor units.     

Self-alignment technique for fiber attachment to guided wave devices

We propose and demonstrate a new multiple fiber-waveguide alignment technique suitable for single-mode and multimode guided wave devices. The method uses an overlap between a precision-etched silicon substrate and the top surface of the waveguide substrate to align all but one of the six degrees of freedom automatically. We have attached six arrays, each with twelve single mode fibers, and have measured an average excess loss of 0.9 dB atlambda = 1.3 mum. The lowest loss array had an average excess loss of 0.4 dB.     

Reducing focused ion beam damage to transmission electron microscopy samples

One of the most important applications of focused ion beam (FIB) systems is sample preparation for transmission electron microscopy (TEM). However, the use of the FIB inherently involves changing and damaging the sample, and thereby degrades the TEM resolution. This paper addresses the beam-induced damage and artifacts, particularly in applications involving silicon semiconductors. The damage appears in the form of amorphization on the surface of the TEM foil. The characteristics of this amorphous damage were studied by making TEM observations of cross sections of the affected foil. The damage is typically 20 to 30 nm thick for a 30 keV FIB, which is generally overly thick for modern silicon devices with feature sizes less than 250 nm. This paper reviews the reported damage depths of FIB-prepared samples, which are determined by experiments and calculations. Several damage reduction techniques, such as the use of gas-assisted etching, low energy FIB, cleaning the FIBfabricated cross section by wet or dry etching and cleaning by broad ion beam (BIB) milling have also been reviewed, with emphasis on applicability to silicon devices. We conclude that the use of low energy FIB and cleaning by argon BIB are particularly efficient techniques.     

Analysis of crosstalk in HgCdTe p-on-n heterojunction photovoltaic infrared sensing arrays

In this paper, an experimental and theoretical study is carried out of crosstalk between nearest-neighbor devices within a backside-illuminated linear HgCdTe photovoltaic infrared sensing array. The dominant form of crosstalk that occurs in high performance photovoltaic arrays is associated with photogenerated minority carriers that diffuse laterally between adjacent devices within the array. To measure crosstalk, a scanning laser microscope is used to obtain a spatial map of spot-scan photoresponse at a temperature of 80K for individual p-on-n photovoltaic devices within the linear array. These experimental results are compared to calculations performed on a commercial two-dimensional device simulation package. The crosstalk measurements and calculations presented in this paper include results on mid-wavelength infrared planar device structures, as well as long-wavelength infrared mesa-isolated devices, which give measured crosstalk values of 6.2 and 8.3%, respectively. The results indicate that the device simulations are in good agreement with experimental results. Further simulations are carried out to determine the sensitivity of crosstalk to various material and device parameters such as epitaxial layer thickness (7 to 25 μm), illumination wavelength (1.047 to 11.0 μm), minority carrier diffusion length (8 to 90 μm), and diode pitch. It is found that the dominant feature influencing the value of crosstalk is the distance between the region of photogeneration and the collecting p-n junction.     

Towards neuromorphic electronics: Memristors on foldable silicon fabric

The advantages associated with neuromorphic computation are rich areas of complex research. We address the fabrication challenge of building neuromorphic devices on structurally foldable platform with high integration density. We present a CMOS compatible fabrication process to demonstrate for the first time memristive devices fabricated on bulk monocrystalline silicon (100) which is next transformed into a flexible thin sheet of silicon fabric with all the pre-fabricated devices. This process preserves the ultra-high integration density advantage unachievable on other flexible substrates. In addition, the memristive devices are of the size of a motor neuron and the flexible/folded architectural form factor is critical to match brain cortex׳s folded pattern for ultra-compact design.