A5046869634- Advanced Memory and Neural Computing
- Ferroelectric and Negative Capacitance Devices
- Semiconductor materials and devices
- Advancements in Semiconductor Devices and Circuit Design
- Neuroscience and Neural Engineering
- Neural dynamics and brain function
- Integrated Circuits and Semiconductor Failure Analysis
- Physical Unclonable Functions (PUFs) and Hardware Security
- Electronic and Structural Properties of Oxides
- Advanced Malware Detection Techniques
- Transition Metal Oxide Nanomaterials
- Neural Networks and Applications
- CCD and CMOS Imaging Sensors
- Adversarial Robustness in Machine Learning
- Anomaly Detection Techniques and Applications
- Thin-Film Transistor Technologies
- Ferroelectric and Piezoelectric Materials
- ZnO doping and properties
- Acoustic Wave Resonator Technologies
- Neural Networks and Reservoir Computing
- CRISPR and Genetic Engineering
- VLSI and Analog Circuit Testing
- Nanowire Synthesis and Applications
- Photoreceptor and optogenetics research
- Software Engineering Research
University of Cincinnati
2016-2025
KIIT University
2025
Wright-Patterson Air Force Base
2022
Cincinnati Children's Hospital Medical Center
2022
University of Toledo
2010-2015
IBM (United States)
2008-2013
North Carolina State University
2004-2006
For the first time, we have demonstrated a 32nm high-k/metal gate (HK-MG) low power CMOS platform technology with standby leakage transistors and functional high-density SRAM cell size of 0.157 μm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> . Record NMOS/PMOS drive currents 1000/575 μA/μm, respectively, been achieved at 1 nA/μm off-current 1.1V V <inf xmlns:xlink="http://www.w3.org/1999/xlink">dd</inf> cost process. With this high...
This report discusses the electrical characteristics of two-terminal synaptic memory devices capable demonstrating an analog change in conductance response to varying amplitude and pulse-width applied signal. The are based on Mn doped HfO2 material. mechanism behind reconfiguration was studied a unified model is presented explain underlying device physics. then utilized show application these speech recognition. A comparison between 20 nm × sized with that state-of-the-art VLSI SRAM synapse...
This letter presents a methodology to accurately extract the work function of metal electrodes on high-/spl kappa/ dielectrics with various charge distributions. A mathematical analysis including sources errors was used study effect distribution in gate dielectric stacks flatband voltage device. The calculations are verified by experimental results obtained for Ru-Ta alloys HfO/sub 2/ and SiO/sub stacks. It is shown that accounting appropriate model imperative accurate calculation...
We report the switching characteristics of RRAM devices consisting Ru/HfO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> /TiO xmlns:xlink="http://www.w3.org/1999/xlink">2-x</sub> /Ru stacks with and without an external access device. In addition to bistable switching, we also achieved analog reconfiguration resistance by controlling compliance current or reset voltage achieve a low-resistance state (LRS) high-resistance (HRS),...
We discuss the theory, design, fabrication, and testing of a three-terminal memristor based on thin-film metal oxides. The fabricated device modifies traditional SrTiO3 to include third control terminal. results show conductance is continuous over three orders magnitude, with significant retention endurance, comparatively low set reset currents. gate allows for state tuning, flexible architectures compared two-terminal memristors by separating read write terminals.
Many currently available deep neural network (DNN) accelerators are highly application specific and have focused on supervised learning. In addition, many rigid architectures algorithms that prevent adapting to dynamic environments. this work, we propose a neuromorphic architecture implementing self-organizing feature map (SOFM) using ferroelectric field-effect transistors (FeFETs) for in-memory error computation. The takes inspiration from biological networks is able grow neurons adapt the...
In this paper, we report multi-level cell (MLC) switching characteristics of resistive random access memory devices with a W/Zr/HfO2/TiN stack. A multi-step forming technique was implemented in work which efficiently suppressed the current overshoot and allowed device at low set/reset voltage current. Four distinct resistance states, achieved by controlling reset stop voltages, showed excellent endurance. Write/read/erase energy values for different states were also calculated. Amongst four...
We report the switching dynamics and charge transport studies on Ru/HfO2/TiOx/Ru resistive random access memory devices in low resistance state (LRS), high (HRS), virgin (VRS). The LRS is governed by Ohmic conduction of electrons through local filamentary paths while it a combination Frenkel-Poole emission trap assisted tunneling process HRS VRS. area filament extracted related to compliance current. thickness re-oxidized reset voltage HRS. energy consumed during was analyzed time-scale...
Machine learning approaches are gaining popularity in the medical field for diagnostics, predictive analytics and general research. With data often being unlabeled or sparse to collect, there is a need unsupervised networks field. Self-Organizing Feature Maps (SOFM) common application of allow use their training. We applied chest x-ray images COVID-19 patients an SOFM network found distinct classification between sick healthy with average euclidean distance 1.1 1st 2nd winning neurons our...
We report the switching characteristics of Mg-doped HfO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> -based ReRAM devices consisting Ru/ Mg:HfO /TiN/W stacks. The concentration Mg dopant was varied from 0% to 20% for four samples show impact on forming voltage. In addition reducing voltage 2.8 1.7 V, doping in also improved repeatability initial device characteristics, and retention. mechanism conduction identified as Frenkel-Poole...
Repetitive occupational lifting has been shown to create an increased risk for incidence of back pain. Ergonomic workstations that promote proper technique can reduce risk, but it is difficult assess the without constant monitoring. Machine learning systems using inertial measurement unit (IMU) data have successful in various human activity recognition (HAR) applications, limited work done regarding tasks which collect significant amounts data, such as manual tasks. In this article, we...
Occupationally-induced back pain is a leading cause of reduced productivity in industry. Detecting when worker lifting incorrectly and at increased risk injury presents significant possible benefits. These include quality life for the due to lower rates fewer workers’ compensation claims missed time employer. However, recognizing provides challenge typically small datasets subtle underlying features accelerometer gyroscope data. A novel method classify dataset using 2D convolutional neural...
Dual metal electrodes such as Ru, Ru-Ta alloy, TaN and TaSiN were investigated on low EOT single layer HfO/sub 2/ stacked 2//SiO/sub gate dielectrics. It was found that the work function values of gates SiO/sub are similar. Thermal anneal studies selected metals above dielectrics also performed to evaluate change in V/sub FB/ values.
We report an interface engineering approach to achieve a self-compliance controlled and forming-free switching in ReRAM devices based on MgO layers. The proposed showed scalability current with low set reset voltages as the interfacial layer thickness was increased. write endurance up 1000 cycles under demonstrated potential for ultra-fast energies. mechanism behind self-compliant is explained supported capacitance-voltage measurement. believe that these will not require compliance limiting...
This letter presents a route for tuning the metal gate effective work function via interface dipoles formed using AlTa and AlTaN alloys. It was found that alloy has higher (4.45 eV) compared to either Al (~ 4.1 or Ta (4.2 gates on SiO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> at 400 degC. increase in attributed electrode dielectric interface. The origin of this dipole is reaction between layer. Similar also observed high-k...
CMOS devices with high-k/metal gate stacks have been fabricated using a gate-first process flow and conventional stressors at lengths of 25 nm, highlighting the scalability this approach for high performance SOI technology. AC drive currents 1630muA/mum 1190muA/mum demonstrated in 45 nm ground-rules 1V 200nA/mum off current nFETs pFETs, Tinv 14 15 respectively. The were achieved simplified integration scheme embedded SiGe dual stress liners (DSL) without utilizing additional enhancers....
In this paper, we investigated the performance of SrTiO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sub> (STO) resistive random-access memory (RRAM) as a candidate to realize short-term synaptic elements in neuromorphic circuits. Ionic defect movement within oxides is postulated play significant role state reconfiguration these devices. This relatively small was sensed by developing novel measurement technique using differential...
This paper presents a simulation based study of the impact nanoscale bidirectional diode reverse saturation current, junction capacitance, and crossbar coupling capacitances on read operation ReRAM devices in 1Diode 1ReRAM memory architectures. Our results show that maximum achievable density is strong function which governs sneak current steady-state mode. For operations with ultrafast voltage pulses, additional paths due to line-to-line arrays come into picture. causes transient spike can...
Impedance spectra of multiple resistive states in a stable-switching Resistive Random-Access Memory device based on stack Ru/HfO2/Zr/W was studied this work. Using these observations, reflection-coefficients were extracted for different states. Clear changes the reflection coefficient observed. The low state showed significantly higher compared to its high state. An increasing trend observed as gradually reconfigured towards lower resistances. Maximum frequency increased with decrease area....
Memory is an integral and important component of both general-purpose embedded systems. It widely acknowledged that energy the memory structure a major contributor in overall system energy. Recent advances with emerging non-volatile (NVM) technologies can potentially alleviate issue leakage power. However, they introduce new challenges opportunities for dynamic power management memory. In this paper, we consider resistive random access (RRAM), promising NVM technology, observe specific...
We report the synaptic characteristics of novel two-terminal reconfigurable devices fabricated using a doped transition metal oxide. These demonstrate short-term plasticity, frequency-dependent augmentation, long-term potentiation, and depression, have potential to show spike timing-dependent plasticity that are macroscopically similar biological synapse. The underlying mechanism behind observed was studied charge transport characterization. Based on this study, fundamental correlation...
Resistive Random-Access Memory (ReRAM) devices have caught significant research attention as scalable nonvolatile memory technology for high-density data storage in 3-D crossbar architectures. ReRAM can switch with low programming voltages (<; ±1 V) at fast time-scales (≈10-100 ns) that make them an attractive option not only off-chip but also on-chip embedded applications. been explored applications reconfigurable switches field programmable gate arrays (FPGAs). The random variabilities the...