Reliability Physics

The study of semiconductor device reliability is motivated by the understanding that as the transistors are turned on and off trillions of times during normal operation of an IC, their parameters (e.g. ON current, ON/OFF ratio, speed, noise, etc.) degrade gradually over time, and if the degradation is excessive, the ICs can stop functioning before the anticipated lifetime of the product. Indeed, if enough customers return these defective parts, the company selling them may go out of business! Therefore, performance and reliability have always been the two primary concerns for an IC designer in choosing technology specifications for a particular product. Reliability physics involves identifying and exploring the physics of these degradation mechanisms using accelerated characterization and sound physical modeling. These models must be accurate so that long term degradation can be predicted based on short-term accelerated tests. Indeed, over the last ten years, as devices have scaled to previously unthinkable dimensions, the concerns regarding reliability, more so than the transistor performance, has defined the technology nodes for each successive generation. Indeed, the ever increasing sophistication of reliability models and the novel characterization techniques to validate them has allowed continued scaling of CMOS technology to sub-100 nm channel dimension. Our research focuses on a broad range of reliability physics issues both for conventional and novel transistors and switching elements. More information...



Novel Devices


 Ultimately Scalable DRAM based on Si-Nanowires

DRAM scaling has been one of the most difficult challenges for modern ICs, requiring innovative and pioneering usage of high-k materials, capacitor geometry, and circuit techniques. According to ITRS roadmap, however, this traditional approach to capacitor scaling may be difficult to implement below 65 nm node, and one must look for alternative cell designs for continued scaling of DRAM circuits. Therefore, we have proposed a Si-Nanowire based capacitor-less, single transistor DRAM element that can scale well into the future, that requires neither high-k gate dielectric nor high aspect ratio trench capacitors, have long-retention time, and does not destroy the stored bit during read operation. We are studying the operation of this novel memory element in the context of the ITRS roadmap and plan to encapsulate our findings in compact model for relative performance analysis of various DRAM architectures.
More information...




Nanocomposites Transistors and Solar Cells


Nano-bundle TFTs for Macroelectronic Applications

Over the past fifteen years, tremendous advances in thin-film transistor (TFT) technology for large-area displays have complemented the equally phenomenal gains in silicon CMOS technology for high performance computation and communication ICs. However, performance limits of a-Si or organic TFT technologies make them unsuitable for a range of potentially exciting and novel applications in sensors, radars, and displays that could dramatically alter the application-landscape of flexible electronic systems. For these applications, researchers are exploring a new class of nano-composites based on bundles of Si nanowires or Carbon Nanotubes. Although initial results promise significant improvement in electro-thermal performance, but lack of adequate transport models have stymied physical understanding and device optimization. We are working on novel transport models for these composites to simulate complete TFT characteristics so that experiments can be interpreted, devices can be optimized, and ultimate limits of this new thin-film material technology can be explored and established. More information...




Nano-bio Sensors


Despite phenomenal increase in processing power of digital computers over the last fifty years, digital technologies' interfaces to the outside world, especially to biomedical systems, remain far from intimate. Although some areas of this interface have benefited from digital technology (e.g., the transducers, image acquisition and processing, and databases for biomedical information), yet the real interface between electronics and biology - the probes that convert the difficult-to-measure chemical, biological, and mechanical signals into electrical or optical signals for convenient digital signal processing - remain bulky, passive, invasive, and expensive. Indeed, real-time information gathering from these systems are often more difficult than processing the information gathered.

Following the pattern of choosing technologically-relevant inter-disciplinary research topics with widening gap between theory and experiments, we have recently began exploring the mechanics, design, and sensitivity of nano-engineered bio-sensors including both passage sensors for genome sequencing and touch sensors for chemical and biological detection. Although the initial experimental results by many groups suggest exciting implications of such sensors for electronics, system biology, personalized medicine, and single-cell expression, the theory of detection of such sensors and the limits of their sensitivity remain poorly understood. As such, it has been impossible to predictively design and comprehensively optimize such nano-bio sensors - a concern that we wish to address in our research. More information...


MATLAB Handle Graphics