Category Archives: MEMS

NVIDIA and Taiwan’s Ministry of Science and Technology today announced an extensive collaboration that will advance Taiwan’s artificial intelligence capabilities.

Announced at the start of Computex 2018, the partnership will extend over the next decade to build up local deep learning and associated AI technologies.

“Taiwan was at the center of the PC revolution and now it is investing to play an important role in the next era of computing,” said Jensen Huang, founder and chief executive officer of NVIDIA. “With the essential infrastructure and tools, the rich talent in Taiwan’s schools and industry will create world-changing breakthroughs in science and society.”

Taiwan Premier Lai Ching-te expressed enthusiasm for the collaboration, which he called essential to sharpening national competitiveness.

“Taiwan is committed to be an important global player in the AI ecosystem,” Premier Lai said. “NVIDIA is the leader of AI computing in the world. By collaborating with NVIDIA, we will gain the expertise and technical platforms to train AI talents, build the strongest AI ecosystem of both software and hardware, and further reshape the world with our own technologies and services of AI.”

The collaboration is focused in five key areas:

  • Supercomputing infrastructure. NVIDIA and Taiwan government agencies will co-invest to bring NVIDIA’s most advanced technology to Taiwan, including the new NVIDIA® HGX-2™, which fuses AI and high performance computing into a single platform.
  • Research. NVIDIA Research, a global organization that includes some of the world’s best computer scientists, will collaborate with Taiwan researchers and startups to exchange best practices.
  • Training. NVIDIA will expand its Deep Learning Institute — which has provided developers worldwide with hands-on training for beginning and advanced AI techniques — to train thousands of Taiwanese developers on the latest AI capabilities.
  • Startups. Taiwan agencies and NVIDIA will work together to help Taiwan AI startups through NVIDIA’s Inception startup accelerator program, which is helping more than 2,800 young companies globally.
  • Innovation. Joint investment in developing AI solutions for key vertical markets for Taiwan, including manufacturing, healthcare, safe cities and transportation.

Building on Grand Plan
The announcement extends the Taiwan Ministry of Science and Technology’s “AI Grand Plan,” which was unveiled last year. Last month, MOST unveiled its Taiwania HPC supercomputer powered by NVIDIA technology. And last week, it selected NVIDIA for an AI supercomputer powered by 2,000 NVIDIA Tesla® V100 32GB Tensor Core GPUs with access to the NVIDIA GPU Cloud™ (NGC) container registry of AI-optimized software.

Speaking last Wednesday to more than 2,200 technologists, developers, researchers and business executives at NVIDIA’s GPU Technology Conference Taiwan, Huang described a series of AI initiatives underway in Taiwan. These address a range of pressing domestic issues in such fields as manufacturing, healthcare and transportation, which align with the government’s focus on furthering AI.

Among the five examples he cited:

  • Foxconn drives superhuman inspection accuracy in manufacturing. Using GPU-powered deep learning with NVIDIA HGX-1 and Tesla P4 GPUs, Foxconn is slashing its manufacturing defect detecting “escape rate.” It has cut the rate to 0.015 percent from the 4.3 percent rate expert human inspectors can achieve — a 287x performance improvement.
  • China Medical University Hospital attacks Asia’s highest cancer fatality rate. Using the NVIDIA DGX-1™ supercomputer, CMUH and Eddie Huang — a post-doc student from MOST — developed an AI to detect liver cancer. The AI diagnostic “super assistant” is especially important on Taiwan, which has Asia’s highest cancer fatality rate.
  • National Taiwan University addresses locally acute cancer type. Working with Dr. Winston Hsu, NTU has made breakthroughs in detecting nasopharyngeal carcinoma, a rare head and neck cancer that’s locally prevalent due to diet and environmental factors. NVIDIA DGX-1 enabled Dr. Hsu to combine CT scans with AI-generated MRI images into one algorithm — improving detection rates by as much as 36 percent.
  • Taoyuan City makes its streets safer. Taiwan’s third-largest city is pushing development of autonomous vehicles to cut back on accidents and carbon emissions, while improving the productivity of trucks, taxis and buses. It is using the NVIDIA DGX Station™ deskside supercomputer for AI model training and the NVIDIA DRIVE™ PX2 autonomous driving computer as it works to have 30 percent of its fixed-route buses equipped with autonomous capabilities by the start of the new decade.
  • Tainan City girds against typhoons. The municipal government of Taiwan’s fourth-largest city is deploying drones, with AI software developed using NVIDIA DGX-1 systems, to monitor the structural integrity of the city’s 1,650 bridges. By evaluating their risk to potential damage from flooding, earthquakes and mudslides, it can fix bridges before the next typhoon hits.

Researchers at Seoul National University and Stanford University developed artificial mechanosensory nerves using flexible organic devices to emulate biological sensory afferent nerves. They used the artificial mechanosensory nerves to control a disabled insect leg and distinguish braille characters.

Compared to conventional digital computers, biological nervous system is powerful for real-world problems, such as visual image processing, voice recognition, tactile sensing, and movement control. This inspired scientists and engineers to work on neuromorphic computing, bioinspired sensors, robot control, and prosthetics. The previous approaches involved implementations at the software level on conventional digital computers and circuit designs using classical silicon devices which have shown critical issues related to power consumption, cost, and multifunction.

The research describes artificial mechanosensory nerves based on flexible organic devices to emulate biological mechanosensory nerves. “The recently found mechanisms of information processing in biological mechanosensory nerves were adopted in our artificial system,” said Zhenan Bao at Stanford University.

The artificial mechanosensory nerves are composed of three essential components: mechanoreceptors (resistive pressure sensors), neurons (organic ring oscillators), and synapses (organic electrochemical transistors). The pressure information from artificial mechanoreceptors can be converted to action potentials through artificial neurons. Multiple action potentials can be integrated into an artificial synapse to actuate biological muscles and recognize braille characters.

Devices that mimic the signal processing and functionality of biological systems can simplify the design of bioinspired system or reduce power consumption. The researchers said organic devices are advantageous because their functional properties can be tuned, they can be printed on a large area at a low cost, and they are flexible like soft biological systems.

Wentao Xu, a researcher at Seoul National University, and Yeongin Kim and Alex Chortos, graduate students at Stanford University, used their artificial mechanosensory nerves to detect large-scale textures and object movements and distinguish braille characters. They also connected the artificial mechanosensory nerves to motor nerves in a detached insect leg and control muscles.

Professor Tae-Woo Lee, a Professor at Seoul National University said, “Our artificial mechanosensory nerves can be used for bioinspired robots and prosthetics compatible with and comfortable for humans.” Lee said, “The development of human-like robots and prosthetics that help people with neurological disabilities can benefit from our work.”

BISTel, a provider of intelligent, real-time data management, advanced analytics and predictive solutions for smart manufacturing announced today its first adaptive intelligence (A.I.) based applications to enable the smart connected factory or industry 4.0 as some call it. Called Dynamic Fault Detection (DFD), BISTel’s new fault detection and classification solution offers customers full sensor trace data analysis to detect and classify faults real-time, improving quality and yield significantly.

Today, customers rely on legacy FDC systems for accurate fault detection. These systems offer only summary data analysis from sensors for fault detection. Consequently, small changes in sensor behavior can go undetected, resulting in a negative impact on yield. BISTel’s new Dynamic Fault Detection (DFD®) system overcomes these challenges by offering full trace analysis. Because BISTel’s new DFD® system establishes trace references dynamically and does not rely on the traditional control limiting methods used by FDC, it eliminates modeling completely. DFD also uses smarter algorithms to better distinguish between real alarms and false alarms resulting in 10 times fewer alarms than FDC systems.

“DFD is the first of several intelligent manufacturing applications with new machine learning that will help our customers to start to realize the full potential of A.I. for smart manufacturing,” commented W.K. Choi, Founder and CEO, BISTel. “DFD enables customers to quickly and accurately identify and classify faults. DFD helps our customers create early identification of yield related issues so that they can quickly execute the fastest possible response to solving these issues.” added Choi.

Sensor trace data contains a wealth of information that helps manufacturers identify potential yield issues, including ramp rate changes, spikes, glitches, shift and drift. BISTel’s first of its kind, online Dynamic Fault Detection (DFD®) system lowers these risks by offering manufacturers real-time monitoring and detection of full sensor trace data. Customers can now quickly detect, and analyze yield impacting events and quickly resolving yield issues. DFD® also integrates seamlessly to legacy FDC systems.

Key Features and Benefits

  • Real time monitoring Improves quality and yield.
  • Reduces risk by protecting against yield impacting events.
  • Real-time fault detection with dynamic references instead of static control limits.
  • DFD’s sensor behavior analysis enables best system drift detection
  • Intelligent alarming reduces alarms by more than 10X

Upon the proposal of ST’s new President & CEO Jean-Marc Chery, the Supervisory Board has approved the establishment of a newly formed Executive Committee, entrusted with the management of the Company and led by Mr. Chery as its Chairman.

The other members of ST’s Executive Committee are:

  • Orio Bellezza, President, Technology, Manufacturing and Quality
  • Marco Cassis, President, Sales, Marketing, Communications and
    Strategy Development
  • Claude Dardanne, President, Microcontrollers and Digital ICs Group
  • Lorenzo Grandi, President, Finance, Infrastructure and Services and Chief Financial Officer
  • Marco Monti, President, Automotive and Discrete Group
  • Georges Penalver, President, Human Resources and Corporate Social Responsibility
  • Steven Rose, President, Legal Counsel
  • Benedetto Vigna, President, Analog, MEMS and Sensors Group.

“ST’s new Executive Committee is a team of strong and experienced semiconductor industry leaders. Our first priority is to deliver on our 2018 business and financial objectives and continue on our path of sustainable and profitable growth. Customers choose ST because we are able to bring them innovation in technology and products. We will keep pushing in this direction, with a focus on fast time-to-market and strong execution, to create value for customers and for all of our stakeholders.” said Jean-Marc Chery, President & CEO of STMicroelectronics.

Developing new medicines to treat pulmonary fibrosis, one of the most common and serious forms of lung disease, is not easy.

One reason: it’s difficult to mimic how the disease damages and scars lung tissue over time, often forcing scientists to employ a hodgepodge of time-consuming and costly techniques to assess the effectiveness of potential treatments.

Now, new biotechnology reported in the journal Nature Communications could streamline the drug-testing process.

The innovation relies on the same technology used to print electronic chips, photolithography. Only instead of semiconducting materials, researchers placed upon the chip arrays of thin, pliable lab-grown lung tissues — in other words, its lung-on-a-chip technology.

“Obviously it’s not an entire lung, but the technology can mimic the damaging effects of lung fibrosis. Ultimately, it could change how we test new drugs, making the process quicker and less expensive,” says lead author Ruogang Zhao, PhD, assistant professor in the Department of Biomedical Engineering at the University at Buffalo.

The department is a multidisciplinary unit formed by UB’s School of Engineering and Applied Sciences and the Jacobs School of Medicine and Biomedical Sciences at UB.

With limited tools for fibrosis study, scientists have struggled to develop medicine to treat the disease. To date, there are only two drugs — pirfenidone and nintedanib — approved by the U.S. Food and Drug Administrations that help slow its progress.

However, both drugs treat only one type of lung fibrosis: idiopathic pulmonary fibrosis. There are more than 200 types of lung fibrosis, according to the American Lung Association, and fibrosis also can affect other vital organs, such as the heart, liver and kidney.

Furthermore, the existing tools do not simulate the progression of lung fibrosis over time — a drawback that has made the development of medicine challenging and relatively expensive. Zhao’s research team, which included past and present students, as well as a University of

Toronto collaborator, created the lung-on-a-chip technology to help address these issues.

Using microlithography, the researchers printed tiny, flexible pillars made of a silicon-based organic polymer. They then placed the tissue, which acts like alveoli (the tiny air sacs in the lungs that allow us to consume oxygen), on top of the pillars.

Researchers induced fibrosis by introducing a protein that causes healthy lung cells to become diseased, leading to the contraction and stiffening of the engineered lung tissue. This mimics the scarring of the lung alveolar tissue in people who suffer from the disease.

The tissue contraction causes the flexible pillars to bend, allowing researchers to calculate the tissue contraction force based on simple mechanical principles.

Researchers tested the system’s effectiveness with pirfenidone and nintedanib. While each drug works differently, the system showed the positive results for both, suggesting the lung-on-a-chip technology could be used to test a variety of potential treatments for lung fibrosis.

Leti, a technology research institute of CEA Tech, today announced its annual flagship event, Leti Innovation Days, July 4-5 in Grenoble.

This year, the institute will address how microelectronics, Leti’s core activities, are empowering new technological revolutions within industry, changing our daily lives in ways that will shape tomorrow’s global, post-modern society – in other words, how humans interact, commute, consume and much more. This two-day event gathers each year hundreds of top executives for presentations and discussions of the latest tech trends and the outlook for the future. 

Program 2018

From microelectronics to markets and end-users

–        Quantum computing: from lab to fab

–        New advances in materials

–        The virtues of photons

–        Bio-inspired circuits

–        5G: Towards less redundant processing

Sessions during the two-day event also will present novel use cases in personalized healthcare and other fields in a hyper-connected world, as well as live tech demonstrations from Renault, Rossignol and other global industrials.

On the evening of July 4, Arianespace CEO Stéphane Israël will headline a special Leti Innovation Days event about trends and visions for the space industry.

Technical Workshops

In addition, there will be seven satellite workshops on design for 3D, lithography, quantum engineering, silicon photonics, memory, 5G, and MEMS on July 2, 3 and 6.

The full program can be found here.

For free registration, please contact [email protected]

Researchers using powerful supercomputers have found a way to generate microwaves with inexpensive silicon, a breakthrough that could dramatically cut costs and improve devices such as sensors in self-driving vehicles.

“Until now, this was considered impossible,” said C.R. Selvakumar, an engineering professor at the University of Waterloo who proposed the concept several years ago.

High-frequency microwaves carry signals in a wide range of devices, including the radar units police use to catch speeders and collision-avoidance systems in cars.

The microwaves are typically generated by devices called Gunn diodes, which take advantage of the unique properties of expensive and toxic semiconductor materials such as gallium arsenide.

When voltage is applied to gallium arsenide and then increased, the electrical current running through it also increases – but only to a certain point. Beyond that point, the current decreases, an oddity known as the Gunn effect that results in the emission of microwaves.

Lead researcher Daryoush Shiri, a former Waterloo doctoral student who now works at Chalmers University of Technology in Sweden, used computational nanotechnology to show that the same effect could be achieved with silicon.

The second-most abundant substance on earth, silicon would be far easier to work with for manufacturing and costs about one-twentieth as much as gallium arsenide.

The new technology involves silicon nanowires so tiny it would take 100,000 of them bundled together to equal the thickness of a human hair.

Complex computer models showed that if silicon nanowires were stretched as voltage was applied to them, the Gunn effect, and therefore the emission of microwaves, could be induced.

“With the advent of new nano-fabrication methods, it is now easy to shape bulk silicon into nanowire forms and use it for this purpose,” said Shiri.

Selvakumar said the theoretical work is the first step in a development process that could lead to much cheaper, more flexible devices for the generation of microwaves.

The stretching mechanism could also act as a switch to turn the effect on and off, or vary the frequency of microwaves for a host of new applications that haven’t even been imagined yet.

“This is only the beginning,” said Selvakumar, a professor of electrical and computer engineering. “Now we will see where it goes, how it will ramify.”

The 64th annual IEEE International Electron Devices Meeting(IEDM), to be held at the Hilton San Francisco Union Square hotel December 1-5, 2018, has issued a Call for Papers seeking the world’s best original work in all areas of microelectronics research and development.

The paper submission deadline this year is Wednesday, August 1, 2018. Authors are asked to submit four-page camera-ready papers. Accepted papers will be published as-is in the proceedings. A limited number of late-news papers will be accepted. Authors are asked to submit late-news papers announcing only the most recent and noteworthy developments. The late-news submission deadline is September 10, 2018.

At IEDM each year, the world’s best scientists and engineers in the field of microelectronics gather to participate in a technical program consisting of more than 220 presentations, along with a variety of panels, special sessions, Short Courses, a supplier exhibit, IEEE/EDS award presentations and other events highlighting leading work in more areas of the field than any other conference.

This year, special emphasis is placed on the following topics:

  • Neuromorphic computing/AI
  • Quantum computing devices and links
  • Devices for RF, 5G, THz and mmWave
  • Advanced memory technologies
  • More-than-Moore devices and integrations
  • Technologies for advanced logic nodes
  • Non-charge-based devices and systems
  • Sensors and MEMS devices
  • Package-device level interactions
  • Electron device simulation and modeling
  • Advanced characterization, reliability and noise
  • Optoelectronics, displays and imaging systems

Overall, papers in the following areas of technology are encouraged:

  • Circuit and Device Interaction
  • Characterization, Reliability and Yield
  • Compound Semiconductor and High-Speed Devices
  • Memory Technology
  • Modeling and Simulation
  • Nano Device Technology
  • Optoelectronics, Displays and Imagers
  • Power Devices
  • Process and Manufacturing Technology
  • Sensors, MEMS and BioMEMS

Further information

For more information, interested persons should visit the IEDM 2018 home page at www.ieee-iedm.org.

BY DAVID W. JIMENEZ, CEO, Wright Williams & Kelly, Inc.

For 27 years Wright Williams & Kelly, Inc. (WWK) has developed strategies and operational products and services proven to produce significant results. Over the course of nearly three decades, WWK has saved its clients over $10 billion and led the way in cost modeling, capacity planning, and operational efficiency; however, sometimes a company gets ahead of its markets. It has been 15 years since WWK launched its first online subscription-based product…and 13 years since it stopped offering it. Today, WWK returns to the cloud.

The cloud is an innovation fueled by advanced chip technology, but it has also been a model the industry hesitated to embrace. Much of this had to do with limited data protection schemes. Intellectual property (IP) is at the core of a successful integrated circuit business and letting key information leave the confines of the organization has traditionally been a forbidden proposition.

Fast forward a decade and a half and cloud-based services are now the norm. Fears over IP theft remain, but the protections have greatly improved. Further, the offerings that add value to cloud-based solutions have also greatly expanded. The move to the cloud now has less to do with a reduction in paranoia and more to do with the advantages of cloud computing. IBM breaks down the advantages into three areas; flexibility; efficiency; and strategic value.

Flexibility allows the scaling of computing power to the task at hand regardless of the local machine used to connect. Efficiency is accessing the needed applications from anywhere in the world from any connected device. Strategic value comes from being able to move faster than competitors by not being tied to existing infrastructure and the hesitancy to obsolete major IT investments. Michael Wright and Walter Ferguson in their 2005 treatise “The New Business Normal” predicted strategic advantage would accrue to those who could access, collate, analyze, and act on information faster than the competition, anywhere in the world and at any time.

WWK has leveraged these advantages by moving its complete suite of manufacturing optimization applica- tions to the cloud. In addition to the advantages inherent in cloud computing, this move provides WWK’s clients substantial cost advantages by lowering up front licensing costs and shifting from capital budgeting to more flexible expense accounting.

Cloud-based solutions: Developed with DARPA/SEMATECH, TWO COOL® is a cost of ownership (COO) and overall equipment efficiency (OEE) modeling platform designed to help equipment and process engineers as well as suppliers understand process step level impacts of changes in operating parameters.

Initially developed by Sandia National Laboratories, Factory Commander® is a cost and resource analysis platform. It analyzes overall factory and individual product costs, manufacturing capacity, and return on investment.

Factory Explorer® is an integrated capacity, cost, and discrete-event simulation tool which predicts factory capacity and bottleneck resources, product cost and gross margins, and dynamic measures such as cycle time and work-in-process.

Advantages put into practice: One advantage in moving these applications to the cloud is users benefit from a state-of-the art computing system. Modeling and simulation apps are computing power intensive. Instead of each user requiring a high-end workstation, the cloud allows users to share a virtual machine(s) (VM). When needs increase, upgrading the VM is quick and low-cost. This keeps the total cost of ownership (TCO) for IT infrastructure at a minimum.
Another advantage is updates happen behind the scenes and for all users at the same time. Traditional software maintenance costs disappear. No more scenarios where users are operating on different revision levels nor lose data due to forgotten backups.

Remote computing has always been a better solution, but there were reasons behind the slow acceptance. Even before the term cloud computing came to the fore, WWK understood this. It offered a remote server-based product before anyone knew what the cloud was. WWK was early to market, but the understanding it gained pointed it in the right direction. Like most market windows you can be early but never late. The arrival of the breadth of solutions needed to offer cloud-based applications has enabled WWK to scrap client-side software licensing and provide a robust, low cost manufacturing optimization software suite with all the advantages it envisioned 15 years ago. I guess we are back to the future.

SEMI, the global industry association representing the electronics manufacturing supply chain, today announced that the WT | Wearable Technologies Conference 2018 USA will co-locate July 11-12 with SEMICON West 2018 in San Francisco. The electronics industry’s premier U.S. event, SEMICON West — July 10-12 at Moscone North and South — will highlight engines of industry expansion including smart transportation, smart manufacturing, smart medtech, smart data, big data, artificial intelligence, blockchain and the Internet of Things (IoT). Click here to register.

“We are excited that the WT | Wearables Technologies Conference has joined SEMICON West to co-locate in 2018,” said David Anderson, president of SEMI Americas. “Our strategic partnership brings new content and more value to our extended supply chain. Every day the semiconductor industry makes chips smaller and faster with ever-higher performance. These innovations enable new wearable applications for smart living, smart medtech and healthcare that are continuously improving our lives. The WT | Wearable Technologies Conference speakers at SEMICON West 2018 will demonstrate just how they use semiconductor technology to deliver leading-edge wearables.”

“It is a great pleasure to collaborate with the leading global electronics manufacturing association and its successful SEMICON West event,” said Christian Stammel, CEO of WT | Wearables Technologies. “Since the beginning of our platform in 2006, the semiconductor industry has been a major driver of wearables and IoT innovation. All major developments in the WT application markets like healthcare (smart patches), safety and security (tracking solutions), lifestyle and sport (smartwatches and wristbands) and in the industrial field (AR / VR) were driven by semiconductor and MEMS innovations. Our program of expert speakers at SEMICON West will share the latest insights in the wearables market as the SEMI and WT ecosystems explore collaboration and innovation opportunities.”