Material engineering for the next generation of GaN-based high-electron-mobility transistors grown by metal-organic chemical vapor deposition

(Al,Ga)N material system have great capacity in high-power, high-frequency applications due to their high breakdown field and electron saturation velocity. One particularly attractive application is the high-electron-mobility transistor (HEMT) with a two-dimensional electron gas (2DEG) formed at the AlGaN/GaN hetero-interface. While the majority of today’s commercial HEMTs are grown in the (0001) metal-polar direction, (000-1) N-polar HEMT structures are advantageous in device scalability, electron confinement and low contact resistance. In this work, we focused on two crucial aspects of N-polar AlGaN/GaN HEMT structures grown by metal-organic chemical vapor deposition (MOCVD), the AlN interlayer and the channel thickness scaling.
The AlN interlayer between the AlGaN barrier and GaN channel is critical to achieve high electron mobility (µ) of the 2DEG by suppressing alloy scattering. However, ~ 50% Ga was previously observed in metal-polar AlN films grown by MOCVD. In this work, both metal- and N-polar AlN films grown under various conditions were investigated. While 40~50% of Ga was again observed in the metal-polar AlN layers, only 1~5% Ga was measured for the N-polar AlN films regardless of the growth conditions, indicating another advantage of N-polar HEMTs compared to the metal-polar ones.
In order to scale down channel thickness of N-polar HEMT structures without sacrificing the sheet charge density (ns) and µ, we proposed a novel design of channel structure and demonstrated it by MOCVD. A thin InGaN layer was inserted between the AlGaN cap and the GaN channel, introducing net positive polarization charge at the InGaN/GaN interface, reducing the electric field in the channel, and therefore increasing ns. The 2DEG also moved away from the AlN/GaN interface, reducing the interface related scattering and improving µ. By replacing the conventional 6 nm thick pure GaN channel with a composite channel of 2 nm In0.1Ga0.9N + 4 nm GaN, the ns increased from 7.5×1012 to 1.2×1013 cm-2 with an improving µ from 606 to 1141 cm2/Vs.

Posted in Defenses | Comments Off on Material engineering for the next generation of GaN-based high-electron-mobility transistors grown by metal-organic chemical vapor deposition

ECE Professor B.S. Manjunath and researchers receive a $3.4 million grant from the NSF’s Office of Advanced Cyberinfrastructure to improve scientific image processing

scientific image processing
The Office of Advanced Cyberinfrastructure (OAC) supports and coordinates the development, acquisition, and provision of state-of-the-art cyberinfrastructure resources, tools and services essential to the advancement and transformation of science and engineering. OAC also supports forward-looking research and education to expand the future capabilities of cyberinfrastructure specific to science and engineering.

Working with scientific images is arduous, and no simple platform exists for sharing them. That is all about to change.

UC Santa Barbara engineers and researchers have been awarded a $3.4 million grant from the National Science Foundation’s Office of Advanced Cyberinfrastructure to build a large-scale distributed image-processing infrastructure (LIMPID) through a broad, interdisciplinary collaboration. Encompassing databases, image analysis and various scientific disciplines, their creation, BisQue, is an image informatics platform that makes it easy to share, distribute and collaborate with large image datasets.

“Think of BisQue as Google Docs for scientific images,” said UCSB principal investigator B. S. Manjunath, who directs the campus’s Center for Multimodal Big Data Science and Healthcare. “Imaging data is ubiquitous and much of big-data science is image-centric. Working with such data should be as simple as working with text files in Google Docs.”

BisQue is unique in its ability to handle a wide range of imaging data across diverse scientific applications, ranging from marine and materials science to neuroscience and medical imaging. For example, Manjunath is working with co-PI Tresa Pollock, UCSB materials chair, to integrate algorithms developed specifically for processing materials imaging data into BisQue. Recent advances in materials tomography (cross-sectional imaging) are generating an enormous quantity of imaging data that must be reconstructed, shared with the community and further analyzed.

Explained Pollock, “LIMPID will greatly enhance our ability to work with large material data sets and will leverage advances made in computer vision and machine learning.”

“In marine science, and particularly marine ecology, the technology to capture underwater images is growing exponentially, but most of the imaging data is manually processed,” said co-PI Robert Miller, a research biologist in UCSB’s Marine Science Institute. “In the Santa Barbara Channel Marine Biodiversity Observation Network, which is supported by NASA and the Bureau of Ocean Energy Management, we are developing image-analysis pipelines and models to process underwater imagery and automate the processes of identifying and quantifying marine organisms. LIMPID will expand that work dramatically to the point where UCSB will become the epicenter of image analysis technology for marine science.”

A team at UC Riverside, the home campus of LIMPID collaborator Amit Roy-Chowdhury, will work with neuroscience researchers to analyze large volumes of live imaging data that capture neuronal activities in the Drosophila nervous system. The UCSB scientists also are collaborating with Nirav Merchant of the University of Arizona, where BisQue and the cyberinfrastructure CyVerse will be leveraged to further enable image-based scientific discoveries.

The potential impacts of the project are significant, ranging from wide dissemination of novel computer vision and deep learning methods to development of automatic methods that can leverage data and human feedback from large data sets for software training and validation.

“The main goal of LIMPID is to provide specific user communities — materials science, marine science and neuroscience — with the ability to share, test and refine methods that have common underlying algorithms and procedures,” Merchant said.

The UCSB Current – ‘Share, Test and Refine’ (full article)

Posted in News | Comments Off on ECE Professor B.S. Manjunath and researchers receive a $3.4 million grant from the NSF’s Office of Advanced Cyberinfrastructure to improve scientific image processing

ECE Professor Chris Palmstrom & UCSB scientists are on the cusp of a major advance in topological quantum computing.

Deterministic growth of InSb nanowire networks
Illustration on the right shows deterministic growth of InSb nanowire networks

UC Santa Barbara scientists are on the cusp of a major advance in topological quantum computing.

In a paper that appears in the journal Nature, Chris Palmstrøm, a UCSB professor of electrical and computer engineering and of materials, and colleagues describe a method by which “hashtag”– shaped nanowires may be coaxed to generate Majorana quasiparticles. These quasiparticles are exotic states that if realized, can be used to encode information with very little risk of decoherence — one of quantum computing’s biggest challenges — and thus, little need for quantum error correction.

“This was a really good step toward making things happen,” said Palmstrøm. In 2012, Dutch scientists Leo Kouwenhoven and Erik Bakkers (also authors on the paper) from the Delft and Eindhoven Universities of Technology in the Netherlands, reported the first observation of states consistent with these quasiparticles. At the time, however, they stopped short of definitive proof that they were in fact the Majoranas, and not other phenomena.

Under the aegis of Microsoft Corporation’s Research Station Q headquartered on the UCSB campus, this team of scientists is part of a greater international effort to build the first topological quantum computer.

“Quantum technology is now being advanced through large academic – industry collaborations,” said Michael Freedman, Fields Medal-winning mathematician and director of Station Q. “The scale of the work, in most cases, is too large for university labs alone, but the imagination and inventiveness of these labs make them essential partners in any industrial effort. Inventiveness and imagination is precisely what is on display in this recent collaboration involving UCSB, Delft, Eindhoven, Copenhagen, and Microsoft. The ‘hashtag’ structures whose quantum properties are studied in this paper have an unworldly beauty and look nearly as impossible as a tower by Escher. They are single crystals with the topology of a circle. Hats off to the grows and the experimentalists.”

The quasiparticles are named for Italian physicist Ettore Majorana, who predicted their existence in 1937, around the birth of quantum mechanics. They have the unique distinction of being their own antiparticles — they can annihilate one another. They also have the quality of being non-Abelian, resulting in the ability to “remember” their relative positions over time — a property that makes them central to topological quantum computation.

“If you are to move these Majoranas physically around each other, they will remember if they were moved clockwise or anticlockwise,” said Mihir Pendharkar, a graduate student researcher in the Palmstrøm Group. This operation of moving one around the other, he continued, is what is referred to as “braiding.” Computations could in theory be performed by braiding the Majoranas and then fusing them, releasing a poof of energy — a “digital high” — or absorbing energy — a “digital low.” The information is contained and processed by the exchange of positions, and the outcome is split between the two or more Majoranas (not the quasiparticles themselves), a topological property that protects the information from the environmental perturbations (noise) that could affect the individual Majoranas.

However, before any braiding can be performed, these fragile and fleeting quasiparticles must first be generated. In this international collaboration, semiconductor wafers started their journey with patterning of gold droplets at the Delft University of Technology. With the gold droplets acting as seeds, Indium antimonide (InSb) semiconductor nanowires were then grown at the Eindhoven University of Technology. Next, the nanowires traveled across the globe to Santa Barbara, where Palmstrøm Group researchers carefully cleaned and partially covered them with a thin shell of superconducting aluminum. The nanowires were returned to the Netherlands for low temperature electrical measurements.

“The Majorana has been predicted to occur between a superconductor and a semiconductor wire,” Palmstrøm explained. Some of the intersecting wires in the infinitesimal hashtag-shaped device are fused together, while others barely miss one another, leaving a very precise gap. This clever design, according to the researchers, allows for some regions of a nanowire to go without an aluminum shell coating, laying down ideal conditions for the measurement of Majoranas.

“What you should be seeing is a state at zero energy,” Pendharkar said. This “zero-bias peak” is consistent with the mathematics that results in a particle being its own antiparticle and was first observed in 2012. “In 2012, they showed a tiny zero-bias blip in a sea of background,” Pendharkar said. With the new approach, he continued, “now the sea has gone missing,” which not only clarifies the 2012 result and takes the researchers one step closer to definitive proof of Majorana states, but also lays a more robust groundwork for the production of these quasiparticles.

Majoranas, because of their particular immunity to error, can be used to construct an ideal qubit (unit of quantum information) for topological quantum computers, and, according to the researchers, can result in a more practicable quantum computer because its fault-tolerance will require fewer qubits for error correction.

“All quantum computers are going to be working at very low temperatures,” Palmstrøm said, “because ‘quantum’ is a very low energy difference.” Thus, said the researchers, cooling fewer fault-tolerant qubits in a quantum circuit would be easier, and done in a smaller footprint, than cooling more error-prone qubits plus those required to protect from error.

The final step toward conclusive proof of Majoranas will be in the braiding, an experiment the researchers hope to conduct in the near future. To that end, the scientists continue to build on this foundation with designs that may enable and measure the outcome of braiding.

“We’ve had the funding and the expertise of people who are experts in the measurements side of things, and experts in the theory side of things,” Pendharkar said, “and it has been a great collaboration that has brought us up to this level.”

The UCSB Current – “Finding Majoranas” (full article)

Posted in News | Comments Off on ECE Professor Chris Palmstrom & UCSB scientists are on the cusp of a major advance in topological quantum computing.

The Association for Computing Machinery (ACM) interviews ECE Professor Yuan Xie in their November 2017 “People of ACM – Bulletin”

photo of yuan xie
“People of ACM” highlights the unique scientific accomplishments and compelling personal attributes of ACM members who are making a difference in advancing computing as a science and a profession. These bulletins feature ACM members whose personal and professional stories are a source of inspiration for the larger computing community.

What research area(s) is receiving the most of your attention right now?
I am looking at application-driven and technology-driven novel circuits/architectures and design methodologies. My current research projects include novel architecture with emerging 3D integrated circuit (IC) and nonvolatile memory, interconnect architecture, and heterogeneous system architecture. In particular, my students and I have put a lot of effort into novel architectures for emerging workloads with an emphasis on artificial intelligence (AI). These novel architectures include computer architectures for deep learning neural networks, neuromorphic computing, and bio-inspired computing.

In your recent book Die-Stacking Architecture co-authored with Jishen Zhao, you predict that 3D memory stacking will be a computer architecture design that will become prevalent in the coming years. Will you tell us a little about 3D memory stacking?
Die-stacking technology is also called three-dimensional integrated circuits (3D ICs). The concept is to stack multiple layers of integrated circuits vertically, and connect them together with vertical interconnections called through-silicon vias (TSVs). My research group has been working on die-stacking architecture for more than a decade. We’ve been looking at different ways to innovate the processor architecture designs with this revolutionary technology. Recently, memory vendors have developed multi-layer 3D stacked DRAM products, such as Samsung’s High-bandwidth Memory (HBM) and Micron’s Hybrid-Memory Cube (HMC). Using interposer technologies, processors can be integrated with 3D stacked memory into the same package, increasing the in-package memory capacity dramatically. The first commercial die-stacking architecture is the AMD Fury X graphic processing unit (GPU) with 4GB HBM die-stacking memory, which was officially released in 2015. Since then, we have seen many other products that integrate 3D memory, such as Nvidia’s Volta GPU, Google’s TPU2, and, most recently, Intel and AMD’s partnership on Intel’s Kaby Lake G series, which integrates AMD’s Radeon GPU and 4GB HBM2.

More questions & answers and Xie’s ACM Bio

  • How might the introduction of radically new hardware impact the existing ecosystem of software?
  • What are the possible architectural innovations in the AI era?
Posted in News, Press Releases | Comments Off on The Association for Computing Machinery (ACM) interviews ECE Professor Yuan Xie in their November 2017 “People of ACM – Bulletin”

Multimodal Analytics for Healthcare

The ailing healthcare system demands effective, autonomous solutions to improve services and provide individualized care, while reducing the burden on the scarce healthcare workforce. Most of these solutions require a multidisciplinary approach that combines healthcare with computational abilities. Intensive Care Unit (ICU) rooms are of particular interest due to their strategic importance, workflow controls, and potential for dissemination of clinical findings and developments. The work presented in this thesis introduces a multimodal, multiview sensor network along with methods and solutions to monitor mock-up and real medical ICU rooms. One prominent outcome of this work includes enabling the medical analysis of preventable ICU conditions such as sleep disorders, decubitus ulcerations, and hospital-acquired infections. Some of the challenges include illumination variations and partial and complete occlusions, such as blankets or privacy curtains. In addition, proper monitoring of human environments requires person identification, which is prohibited by healthcare privacy-protection stipulations and can be limited by scene constraints. The problems tackled include patient-pose classification, pose-motion analysis and summarization, role representation and identification, and activity and event logging. These problems are addressed via a non-intrusive, non-disruptive, multimodal, multiview sensor network (i.e., Medical Internet-of-Things). The multimodal data is combined with coupled optimization to estimate source weights and accurately classify patient poses. Pose transitions are represented using deep convolutional features and pose durations are modelled via segments. The described techniques serve to differentiate between poses and pseudo-poses (transitions) and create effective motion summaries. Role representations are tackled using novel appearance and semantic interaction maps to assign generic labels to individuals (e.g. doctors, nurses, or visitors) without using identifiable information (e.g., face tracking or badges), which is prohibited in healthcare applications. Finally, activities and events are analyzed using contextual aspects, where aspect bases and weights are learned and then used to classify activities. The objective of this thesis is to enable the development, evaluation, and optimization of individualized therapies, standards-of-care, infrastructural designs, and clinical workflows and procedures.

Posted in Defenses | Comments Off on Multimodal Analytics for Healthcare

Robotics meets Wireless Communications: Opportunities and Challenges

The 2017-2018 Pacific Views – Library Speaker Series kicks off with a talk by Professor Yasamin Mostofi (UCSB Department of Electrical & Computer Engineering) to be followed by a public reception

Radio Frequency signals are everywhere these days. Can they be used for sensing? Do they carry useful information about the objects they visit? For instance, imagine two unmanned vehicles arriving behind thick concrete walls. They have no prior knowledge of the area behind these walls. But they are able to see every square inch of the invisible area through the walls, fully imaging what is on the other side with high accuracy. Can the robots achieve this with only WiFi signals and no other sensors? In another example, consider the WiFi network of a building. Can it estimate the occupancy level of the building and the spatial concentration of the people, without relying on people to carry a device? In this talk, Mostofi will discuss her latest results to achieve these goals. More specifically, she will show that it is possible to achieve x-ray vision with only WiFi signals and drones, and image details through thick concrete walls. Furthermore, she will discuss occupancy estimation where I show how to count people with only WiFi measurements

Posted in Events | Comments Off on Robotics meets Wireless Communications: Opportunities and Challenges

Combinatorial Designs for Distributed Storage, Function Computation and Coded Caching

Combinatorial design theory has its roots in recreational mathematics and is concerned with the arrangement of the elements of a finite set into subsets, such that the collection of subsets has certain “nice” properties. In this talk we shall demonstrate that interpreting designs in the right manner yields improved solutions for distributed storage and content caching and novel impossibility results for distributed function computation.

Regenerating codes have been proposed as an efficient mechanism for dealing with the problem of reliability in large scale distributed storage systems. These systems also have additional requirements pertaining to repair. When nodes fail, the system needs to be repaired in a speedy manner by consuming as few resources (number of drives accessed, energy etc.) as possible. We will demonstrate that combinatorial designs allow us to design efficient systems that upon failure can be repaired by simply downloading packets from the surviving nodes.

Next, we will show that designs can be used to construct a family of directed acyclic networks that have several interesting properties. In particular, our work shows that the computation rate of such networks depends significantly on the source alphabets. This is in stark contrast with multiple unicast networks where the rate is independent of the source alphabet. We will conclude with an overview of the role of coding in content caching networks and our recent results on how combinatorial designs can play a central role in making coded caching more practical.

The talk will be self-contained; no background in combinatorial designs and/or network coding will be assumed.

Posted in Events | Comments Off on Combinatorial Designs for Distributed Storage, Function Computation and Coded Caching

Integrated SI3N4 Waveguide Circuits for Single- and Multi-Layer Applications

Photonic integrated circuits are key building blocks for ever increasing range of applications including optical communications, sensing, and position and navigation. A key challenge to today’s photonics integration is realizing circuits and function that require low loss waveguides on chip while balancing the waveguide loss with device function and footprint. The Si­3N4 waveguide low loss platform serves as the third platform that complements silicon photonics and III/V semiconductor based photonics. Incorporating the low loss of Si3N4 waveguides into a photonic circuit to realize varying functions requires tuning the properties of the waveguide through parameters like waveguide core geometry and upper cladding design. The design, fabrication, and optimization of low loss Si3N4 waveguides for multiple applications is described and several devices are demonstrated. First a high-extinction tunable third-order resonator filter with record extinction ratio is presented utilizing a waveguide geometry for compact and higher FSR devices. Next, rotational sensors using low loss, large area designs are demonstrated in the form of both a waveguide gyroscope coil and high Q resonator. Lastly, a method for vertically integrating multiple waveguide layers, capable of integrating devices with different loss and footprint requirements, is demonstrated in the form of a multi-layer delay spiral.

Posted in Defenses | Comments Off on Integrated SI3N4 Waveguide Circuits for Single- and Multi-Layer Applications

Semiconductor Quantum Dot Lasers: Why are they so quantum?

By the reduction of semiconductor structure dimensionality, from bulk to quantum-well and finally to quantum-dots (QDs), we gain in control over the energetic distribution of carriers and their interaction with light. While in bulk material many electron states, which are distributed in energy and momentum, have to be occupied, only a few states near the bandgap can contribute to the optical gain in lasers. Over the last past years, tremendous efforts have been carried out towards the improvement of quantum confined devices. All these achievements are related to the invention of novel in plane semiconductor materials like those based on QD nanostructures. Such self-organized nanostructures are one of the best practical examples of emerging nanotechnologies. Indeed, owing to the atom-like discrete energy levels, all carriers in the ground state contribute to the optical amplification. Practically, QD lasers exhibit meaningful properties resulting from the three-dimensional confinement of carriers, like a high stability against temperature variation and a low-threshold lasing operation, which is of paramount importance to feature low energy consumption photonic integrated circuits [2]. In addition, QD mode-locked lasers can be used as efficient sources for tunable optical and electrical pulse combs. In this seminar, I will review our recent advances in nonlinear and dynamics properties of QD lasers for high-speed operation and narrow linewidth emission. I will also report on the optical feedback dynamics of QD lasers emitting exclusively on sole ground and excited lasing states. Lastly, I will discuss the dynamical response of QD silicon lasers epitaxially grown onto silicon and show that the large robustness against optical perturbations is attributed to the ultra small linewidth enhancement factor (α-factor) These last results demonstrate the ability of silicon lasers to operate without an isolator that is important for the development of future silicon photonics systems.

Posted in Events | Comments Off on Semiconductor Quantum Dot Lasers: Why are they so quantum?

Rescuing Moore’s Law with Two-Dimensional Van der Waals Materials

Moore’s law, which has been guiding the development of integrated circuits (IC) industry over six decades, is facing unprecedented challenges, because of severe short-channel effects and unaffordable fabrication cost of silicon based sub-10 nm metal-oxide-semiconductor field-effect transistors (MOSFETs). The emerging two-dimensional (2D) Van der Waals materials provide an ideal platform for ultra-short channel FETs due to their unique properties, and hence can potentially extend Moore’s law into forthcoming sub-10 nm technology nodes. My doctoral research is focused on exploring, both theoretically and experimentally, the performance and scalability of 2D materials based logic/memory devices for next-generation very-large-scale-integration (VLSI) circuits.

The theoretical work can be divided into two parts. The first is analytical modeling of 2D FETs and tunnel FETs, which provides straightforward physical insight/guidance into the device design and optimization, and paves the way for large-scale 2D circuit exploration. The second is developing a non-equilibrium Green function (NEGF) based quantum transport simulator (over 3000 lines of Matlab code) to evaluate/predict the intrinsic or upper-limit performance and scalability of 2D logic (FETs and tunnel FETs) and memory (floating gate transistors) devices. It is found that the ultrathin body and unique properties of 2D materials allow FETs, tunnel FETs and floating gate transistors made of them scalable up to 5-6 nm channel length, without compromising device performance. By judiciously selecting 2D materials for the channel and floating gate, a novel retention mechanism is found achievable and can improve the lifetime of NAND FLASH by more than one order. Moreover, by taking key parameters for transport, such as effective mass, and mobility etc., as variables for extensive multi-dimensional simulations, a performance phase diagram is constructed. Through this diagram, it is found that among various 2D semiconductors, black phosphorus (BP) and WSe2 are the most promising candidates for high-performance (HP) and low-standby-power (LSTP) applications, respectively, at 6 nm channel length scale. This work unambiguously charts the research direction and paves the pathway for experimentalists in the 2D electronics arena.

The experimental contribution also has two parts. The first is demonstrating a 10-nm-scale top gated single layer MoS2 FET without resorting to the high-cost and/or low-throughput electron beam lithography, for the first time. The gate/channel of this device is defined with 6-nm Al2O3 wrapped metallic Si2Co nanowire, which is obtained through chemical synthesis. The scalability of this device is evaluated with quantum transport simulation, and proved to be scalable to 6 nm. The second contribution is realizing a novel polarity programmable CMOS device that is entirely made of 2D materials, specifically tungsten diselenide (WSe2), graphene, and hexagonal boron nitride (h-BN). The polarity programmability is achieved by adding an extra control (or programming) gate on one of the two contacts of a Schottky barrier FET, to modulate the Schottky barrier contact to n-type or p-type. Demonstrated device shows clear unipolarity and high current level. Device performance can be further improved by suppressing defect density in WSe2, and by using 2D high-k gate dielectric. This polarity programmable 2D device renders exponential functionality increase to circuits/chips, i.e., the cost per functionality is significantly reduced, thereby opening up a revolutionarily new pathway to sustain Moore’s law forever.

Posted in Defenses | Comments Off on Rescuing Moore’s Law with Two-Dimensional Van der Waals Materials