Boosting Machine Vision Efficiency with Memristors
As machine vision technology advances, there is a growing need for more efficient systems that can process images and learn from visual data. Traditional systems face challenges due to separate components for sensors, memory, and computing, leading to high energy consumption and delays. This study introduces a new visual recognition system using memristor-based reservoir computing (RC) with integrated sensing and memory. By utilizing a 2D material (WS2) with light-responsive properties, the system efficiently transforms image data into high-dimensional information, simplifying classification with minimal training. The system achieved 88.3% accuracy in recognizing handwritten digits and 100% accuracy in identifying traffic signal colors, paving the way for more efficient and sustainable machine learning applications in artificial vision.
Carbon Nanotubes for Better Signal Shielding
As electronic devices become smaller and more powerful, managing electromagnetic interference (EMI) is a growing challenge. To address this, we have developed a new method to enhance electromagnetic isolation while occupying minimal space on device boards. The approach uses carbon nanotubes (CNTs), which are highly conductive and offer strong potential for reducing interference, but have been limited by their complex manufacturing needs. This study presents a technique for transferring carbon nanotube fence walls (CNTFW) onto existing structures, boosting their ability to block unwanted signals. The result is an up to 16 dB improvement in isolation across a broad frequency range (0-50 GHz), offering a promising solution for more efficient and compact electronic devices.
Faster Computing with Low-Power Ternary Logic
Ternary logic, which uses three states instead of the usual two, is gaining attention for its potential to improve the efficiency of digital circuits, particularly in data-heavy tasks like artificial intelligence. This research focuses on optimizing ternary multiply-accumulate (MAC) units, crucial for tasks like neural network processing. The study introduces new ternary algorithms that reduce power consumption by up to 30%, with minimal errors. It also presents advanced ternary circuits that outperform traditional binary systems, offering up to 80% lower power-delay product (PDP) and 45% less area usage. These improvements were confirmed through simulations using both carbon nanotube and CMOS technologies, showing promising practical applications for future high-performance, low-power computing systems.
Thermoelectric Gains in Ultra-Thin Penta-PdPS
Thermoelectric materials, which convert heat into electricity, are a promising technology for energy harvesting. This study investigates a novel class of 2D materials, penta-PdPS, which are made up of thin layers of palladium, phosphorus, and sulfur arranged in a pentagonal pattern. By examining how layer thickness and applied voltage affect their thermoelectric properties, the research reveals that thinner layers of PdPS significantly boost its performance. The thermoelectric power factor, a key indicator of efficiency, was found to be twice as high for 11-layer PdPS compared to thicker 88-layer samples. These findings highlight the potential of ultra-thin 2D materials in advancing energy conversion technologies, with implications for developing more efficient thermoelectric devices.
Cleaner Thin Films with FCVA
Filtered Cathodic Vacuum Arc (FCVA) is a breakthrough technique in material deposition that combines high-energy ions with a magnetic filter to eliminate unwanted particles, a common problem in traditional methods. By effectively removing these macroparticles, FCVA produces cleaner, high-quality thin films, which are crucial in industries like electronics and optics. This method has been particularly successful for depositing metal oxide films, such as titanium oxide and zinc oxide, which have applications in everything from solar cells to transparent electronics. Additionally, we have used FCVA for creating nanostructured carbon films, making it a versatile tool for advancing electronics, and nanotechnology.
Scalable MoSe2 Monolayers for Superior Electronics
Two-dimensional materials like MoS2 and WS2 are key to the development of electronic and optoelectronic devices. MoSe2, a promising alternative, offers superior electron mobility and a smaller band gap, making it more suitable for practical applications. However, producing large-scale, high-quality MoSe2 has been a significant challenge. In this study, we successfully grew uniform MoSe2 monolayers using chemical vapor deposition (CVD) at ambient pressure. These monolayers demonstrated a direct band gap of 1.5 eV and excellent electrical performance with a mobility of 50 cm²/V·s, surpassing previous MoS2 reports. This breakthrough opens up new possibilities for MoSe2 in advanced electronic and optical technologies.
Accurate Band Gap Forecasting for TMDs
Predicting the band gap of new materials is essential for discovering their potential uses, especially in electronics and optoelectronics. This study presents a new method called GVJ-2e for accurately predicting the band gap of transition metal dichalcogenides (TMDs), such as MoS2, MoSe2, WS2, and WSe2. The method, based on total energies and without adjustable parameters, delivers results that closely match experimental values for both bulk and monolayer TMDs. It predicts direct band gaps for monolayers and indirect gaps for bulk materials. With an error margin significantly smaller than other popular methods, GVJ-2e offers a highly reliable tool for exploring the electronic properties of 2D and 3D materials, supporting the development of new technologies.
Raman Scattering from Bulk to Monolayer MoS2
Molybdenum disulfide (MoS2) has unique properties for electronics and optoelectronics that improve as it becomes thinner. This study explores how the behavior of MoS2 changes as it is reduced to just a few layers, using Raman spectroscopy to track these changes. It shows that key Raman peaks shift predictably with the number of layers in MoS2, providing a reliable way to identify the thickness of ultrathin flakes. The study also reveals that the interaction between electronic transitions and vibrations (phonons) becomes weaker as MoS2 gets thinner, offering insights into how its electronic properties evolve. These findings are crucial for the development of MoS2-based devices, improving their performance in applications like transistors and solar cells.
Professor Tay Beng Kang
School of Electrical & Electronic Engineering
Nanyang Technological University
S1-B1a-22, 50 Nanyang Avenue
Singapore 639798
Email: EBKTay@ntu.edu.sg
Tel: (+65) 6790 4533