Integrated Microsystems Lab

Smart Radiation Detection Readout IC

Significant efforts have been made recently to enhance the effectiveness of nuclear and radiological detection capabilities, specifically in design of portable systems that would be able to localize the threat. A proposed three-dimensional (3D) integration of scintillation-type semiconductor detector pixels provides accurate spectroscopic resolution for isotope discrimination and an accurate determination of the direction to source at the same time. We are investigating the design of integrated readout circuitry for measurement of the optical response of pixelated detector. Large parasitic capacitance imposes stringent constraint in the design of the sensing circuitry and requires development of novel circuit techniques to relax the effect of input parasitic capacitance.

Integrated electronics for biomedical applications

RF ID pill: Drug uptake by patients often needs to be tracked, especially in psychiatric patients, transplant patients and elderly people. The objective of this research is design of an integrated system that would be inserted in a pill and enable monitoring of ingestion of medicine and absorption into the body to insure proper dosage control and usage. We are investigating the design wireless link between RFID tag inside a pill and receiver that is worn on the body. The challenge in the design of wireless link presents extension of distance that can be achieved through near-field UHF, since due to weak coupling between antennas only small part of emitted power by external transmitter antenna reaches RFID tag antenna.

Gas and bio-metabolite detection: The detection and monitoring of gases in exhaled human breath to-date has been limited by the lack of appropriate materials and technologies which could rapidly and selectively identify the presence and monitor the concentration of trace levels of specific analytes, biomarkers. Conventional techniques lack the relative specificity to a mixture of gaseous analytes for these sensing elements, requiring the use of pattern recognition algorithms to process the signals and match the acquired data profile to a known pattern, thus identifying the signature of the odor or gas detected. We are investigating the design of selective sensor arrays systems for discrimination and monitoring of bio-chemical metabolites that signal diseases. The gas sensor behaves electrically as a resistance and therefore a specialized multi-channel instrumentation is required to obtain readouts. The sensitivity required for quantification of the ppb (parts-per-billion) level concentration of different analytes calls for novel circuit and system techniques for readout microsystem.

Neural recording: Recording from large population of neurons using multi-electrodes is becoming a necessary procedure not only to research neuronal activities in central/peripheral nervous system but also in development of neural prosthesis. To better understand how the nervous system functions, we must be able to simultaneously monitor the responses of many neurons in small animals. To utilize advances in fabrication of MEMS structures that enable micro-electrode arrays, there is need for recording systems able to record data and communicate gathered information to recording station in such a way that animal movement is restricted in minimal way. Existing systems for multi-channel recording mainly use analog headstage requiring a number of wires equal to the number of channels, leading to restriction of movement and poor scalability. The design of a headstage with on-chip analog-to-digital conversion enables digital telemetry of neural signals in their entirety, however high-data-rate in multichannel implementation is concern and efficient control of bandwidth is necessary. Our work lead to design of ultra-low-power, low data-rate, wireless 32-channel VLSI microsystem for recording of neural signals from array of micro microelectrodes mounted on a head of a rat-like small animal.

Neurotransmitter concentration monitoring: Design of implanted neural interfaces facilitates understanding of neurological phenomena and allows for automated medical diagnostics. Our research in this direction has led to design of a mixed-signal potentiostatic VLSI chip that performs simultaneous multi-channel acquisition of neurotransmitter concentration from a micro-fabricated sensor array. The current measurement capability spans six orders of magnitude in dynamic range down to hundreds of femtoamperes through innovative micropower design in analog-to-digital conversion. Another step towards integrated implantable neurotransmitter monitoring system was the design of a passive telemetry VLSI chip in a standard CMOS process.

Acoustic localization and separation

The precision of conventional source localization degrades with shrinking dimensions of the sensor array. Our research has led to gradient flow, a new technique for localization of a traveling wave (acoustic, sonar, RF, ...) with accuracy fundamentally independent of aperture, applicable to miniature sensor arrays of sub-wavelength dimensions. A micropower mixed-signal VLSI implementation of gradient flow for 3-D bearing estimation has experimentally demonstrated one degree angular resolution in localizing a broadband acoustic source at 20dB SNR. The chip was integrated in an Acoustic Surveillance Unit (ASU) for autonomous acoustic sensing of the battlefield with superior accuracy in localizing ground vehicles compared to commercial systems of considerably larger size and power budget. Gradient flow in combination with independent component analysis (ICA) enables blind localization and separation of multiple sources. Our work in this area has contributed to the design of intelligent hearing aids with integrated adaptive suppression of unwanted speech, other sources of non-stationary noise, and interfering sounds. This efficient signal representation of gradient flow as a front-end for blind source separation led to implementation of real-time, versatile, reconfigurable hardware implementation of a general class of independent component analysis algorithms. The mixed-signal VLSI implementation uses a combination of analog and digital VLSI technology to maximize flexibility in programming and configuring ICA adaptation rules at minimal power dissipation.