Bulletin of the American Physical Society
APS March Meeting 2019
Volume 64, Number 2
Monday–Friday, March 4–8, 2019; Boston, Massachusetts
Session K33: Phase Change Materials, Memristors, and Neuromorphic Computing |
Hide Abstracts |
Sponsoring Units: FIAP Chair: Ali Gokirmak Room: BCEC 204B |
Wednesday, March 6, 2019 8:00AM - 8:12AM |
K33.00001: Evidence of Charge Trapping Giving Rise to Resistance Drift of Metastable Amorphous Ge2Sb2Te5 Raihan Sayeed Khan, Sadid Muneer, Nafisa Noor, Helena Silva, Ali Gokirmak Phase change memory (PCM) is a high speed, high density, scalable non-volatile memory that utilizes the resistivity contrast between amorphous (high resistivity) and crystalline (low resistivity) phases of chalcogenides like Ge2Sb2Te5 (GST) [1]. The large resistivity window enables programming to intermediate states and hence multi bit storage; however, PCM suffers from spontaneous resistance drift after amorphization [2,3] potentially resulting in overlapping resistance levels. It is difficult to be understand resistance drift from room temperature measurements as multiple processes may be occurring simultaneously, presenting a need for cryogenic measurements. We monitor resistance drift in GST line cells for up to ~104 s from 175K to 300K at 25K intervals with varying light exposure. The resistance drift coefficients decrease with decreasing temperature. Light exposure alters cell resistivity at a time scale longer than expected from thermal effects at these temperatures, suggesting a charge-trapping mechanism. |
Wednesday, March 6, 2019 8:12AM - 8:24AM |
K33.00002: Heat transfer as coupling mechanism in VO2-based neurons Javier Del Valle Granda, Yoav Kalcheim, Pavel Salev, Ivan Schuller Neuromorphic computing is a new computation paradigm which imitates the architecture of biological brain. One of the basic tasks is to find materials that mimic the functionality of the two basic components of neural networks: neurons and synapses. Resistive switching offers a unique opportunity to do this. Non-volatile resistive switching has been extensively studied and successfully used to mimic synaptic behavior. On the other hand, finding resistive-switching based neurons has remained an elusive task. Current hardware neurons are based on complex, not-scalable CMOS circuits. |
Wednesday, March 6, 2019 8:24AM - 8:36AM |
K33.00003: Electrical Characterization of Ge2Sb2Te5 Phase Change Memory Cells at Cryogenic Temperatures to Investigate the Physical Phenomena that Give Rise to Resistance Drift of the Amorphous Phase ABM Hasan Talukder, Raihan Sayeed Khan, Sadid Muneer, Kimberly Nguyen, Madison Nadolny, Ali Gokirmak, Helena Silva Electrical measurements of metastable amorphous Ge2Sb2Te5 in the temperature range of 300 - 675 K using a waveform tailored to melt, quench and characterize cells in a single shot [1] have resulted in characterization of temperature dependent carrier activation energies that follow a parabolic behavior with a peak value of ∼377 meV at ∼465 K [2]. Measurements of cells amorphized at cryogenic temperatures give insights about the physical phenomena that give rise to resistance drift of the amorphous phase as well as the activation energy in a broader temperature range. We have performed electrical measurements in 85 – 300 K temperature range and observed resistance drift over time. As the viscosity of the material is expected to be extremely high at 85 K, we expect charge trapping to be the dominant factor for drift in amorphous Ge2Sb2Te5. |
Wednesday, March 6, 2019 8:36AM - 8:48AM |
K33.00004: Characteristics and applications of carbon-based ReRAM Pei-Fang Chung, Mon-shu Ho In this work, we anticipate the performance of fullerene molecules could provide a new prototype of the Carbon-based ReRAM model owing to its outstanding electric and magnetic characteristic. Few layers of fullerene molecules (C84) ultrathin film were assembled thermally on Si (111)-7×7 substrates under ultra-high vacuum. The top-electrode Pt (50nm) was deposited by RF magnetron sputtering on C84 film. The surface electronic states and magnetic characteristics of C84 film were studied by UHV-STM and MFM. The STS has revealed the wide band gap of fullerene molecule in which the LDOS can be tuned by different cover density and different combination of fullerene molecules. The ferromagnetic domain has been discovered along the domain boundaries on proposed Si substrate due to the quantum confinement along with localized unpaired electrons in fullerene molecules. The resistance switching behavior of the memory devices were attributed to the formation and rupture of conductive filaments. Thus, we characterized the binding feature and the photoelectric properties with Raman, XPS and VT-PL. We could declare fullerene molecules have potential as a promising candidate for next generation nonvolatile memory and other semiconductor applications. |
Wednesday, March 6, 2019 8:48AM - 9:00AM |
K33.00005: Finite Element Modeling of Ovonic Switching Jake Scoggin, Helena Silva, Ali Gokirmak Ovonic switching in amorphous semiconductors was first reported in the 1960s [1] and has since spawned phase change memory, ovonic threshold switches, and a rich debate on the physical phenomena underlying the rapid transition from low to high electrical conductivity in these materials once a sufficient field is applied [2]. We model ovonic switching as a field-assisted thermal phenomenon in 2-D, 2-D rotational, and 3-D finite element simulations. We vary fields, geometries, and transient conditions and obtain I-V characteristics in good agreement with experimental data. We also show 2-D simulations of ovonic switches as current-limiting access devices for reset and set of phase change memory cells in a crossbar array using our finite element phase change model [3], [4]. |
Wednesday, March 6, 2019 9:00AM - 9:12AM |
K33.00006: On attractor states in the dynamics of pulse-driven memristors Yuriy Pershin, Valeriy Slipko We predict the existence of attractor states in the dynamics of certain pulse-driven memristors (resistors with memory) and memristive networks [1]. A general approach to identify the attractor points in two-terminal memristive circuits driven by alternating polarity voltage or current pulses has been developed and applied to several model cases including the ideal and threshold-type memristors, as well as certain memristive networks. Potential applications of dynamical memristor attractors include the memristor initialization, associative memory, and validation of memristor models, to name a few. In particular, based on specific cases we have demonstrated that the presence or absence of attractor dynamics can be directly related to the window function used in the memristor model. This result paves the way towards a simple experimental test of memristor models, which is just one of several promising applications of dynamical memristor attractors. |
Wednesday, March 6, 2019 9:12AM - 9:24AM |
K33.00007: Finite Element Simulation of Phase Change Memory Cell MD TASHFIQ BIN KASHEM, Jake Scoggin, Sadid Muneer, Helena Silva, Ali Gokirmak We perform finite element simulations of read, reset and set operations on Ge2Sb2Te5 phase change memory cells using the model developed in [1], [2]. An electrical circuit model is included to utilize a transistor as an access device. Thermoelectric effects (Thomson and Peltier heat) are incorporated in the current continuity and heat transfer physics which are solved self-consistently to obtain potential and temperature distributions across the device. We use temperature dependent material parameters (Seebeck coefficient, thermal conductivity and electrical resistivity) to precisely model electrothermal phenomena in the cell. We perform successive read, reset and set operations and analyze nucleation, growth and amorphization as well as temperature and resistance profiles during the simulations. |
Wednesday, March 6, 2019 9:24AM - 9:36AM |
K33.00008: Enhanced Reset Variability in Phase Change Memory for Hardware Security Applications Nafisa Noor, Sadid Muneer, Raihan Sayeed Khan, Ali Gokirmak, Helena Silva Phase change memory (PCM) relies on the change of resistance of a chalcogenide material, that can be reversibly switched between amorphous and crystalline states by applying appropriate electrical pulses. We have experimentally shown enhanced variability during the amorphization hexagonal close packed (hcp) Ge2Sb2Te5 line cells by using long-narrow cell dimensions and voltage pulses with narrow rise and fall times. Percolation-blocked conduction in long-narrow cells, parasitic capacitive current, and thermal runaway at the onset of melting from the initial hcp state are the possible physical mechanisms behind the significant increase of reset uncertainty, which is useful for implementing hardware security primitives1–4. |
Wednesday, March 6, 2019 9:36AM - 9:48AM |
K33.00009: A simple and highly-scalable artificial neuron using an Ovonic Threshold Switch (OTS) Suyoun Lee, Milim Lee, Seong Won Cho, Joon Young Kwak, Hyunsu Ju, Yeonjin Yi, Byung-ki Cheong A scalable and low power-consuming artificial neuron is an essential building block for developing a brain-inspired computing system. Among various features of a biological neuron in the mammalian cortex, the spike-frequency adaptation and chaotic activities are very important ingredients for the realization of the energy-efficient signal processing, learning, and adaptation to environments, which are hard to be achieved up to now. In this work, we have demonstrated those features in a simple artificial neuron device composed of an Ovonic Threshold Switch (OTS) and a few passive electrical components. Furthermore, with our OTS-based neuron device employing the reservoir computing technique combined with delayed feedback dynamics, spoken-digit recognition task has been performed with a considerable degree of recognition accuracy. These results show that our OTS-based artificial neuron device is promising for the application in the development of a large-scale brain-inspired computing system. |
Wednesday, March 6, 2019 9:48AM - 10:00AM |
K33.00010: Development of Self-rectifying TaOy/Nanoporous TaOx Memristor Synapse for Suppressing Non-neural Signal in the Large-scale Neuromorphic Array System Sanghyeon Choi, Seonghoon Jang, Jung-Hwan Moon, Jong Chan Kim, Hu Young Jeong, Peonghwa Jang, Kyung-Jin Lee, Gunuk Wang Memristor that consists of a metal-oxide layer sandwiched between two conductors is being greatly envisioned as a platform to imitate the principal of biological synapses due to its nonlinear and dynamic electrical properties depending on the history of applied electrical programming[1]. In this study, we fabricated the nanoporous (NP) TaOx memristor device by an anodic treatment in the room temperature and utilized the device as the two-terminal artificial synapses[2]. The device exhibits a stable self-rectifying I-V switching behavior with ~104 nonlinearity, which can effectively prevent the undesired neural signals in the densely-integrated synaptic array. Based on analog shift of the Ohmic-contact site by diverse electrical stimuli, the essential synaptic functions were successfully mimicked. A 16 × 16 crossbar array with only our device was fabricated and statistically evaluated. In addition, we investigated the effect of the nonlinearity of the synaptic device on the accuracy of the pattern recognition using artificial neural network simulation. Taken all together, we believe the designed device can provide a route toward the large-scale neuromorphic computing technology. |
Wednesday, March 6, 2019 10:00AM - 10:12AM |
K33.00011: Current-induced resistivity switching in VO2 micro-electronic oscillators Milinda Pattanayak, Md Nadim F Hoque, Zhaoyang Fan, Ayrton A Bernussi Functional metal oxides are a class of emerging materials with novel applications in tunable and reconfigurable microelectronics beyond the CMOS technology. Vanadium dioxide (VO2) is one such material that is attracting considerable attention for its reversible insulator-to-metal phase transition (IMT). Under electrically controlled IMT process VO2 devices exhibit electrical switching and negative differential resistance (NDR). Such attributes make VO2 the ideal candidate for designing tunable micro oscillators for emerging applications such as neuro-morphic computing. We designed, fabricated, and characterized planar VO2 micro-channel devices with different dimensions. The intrinsic parasitic capacitance of the fabricated devices allowed for stable spontaneous electrical oscillations under current actuation. Our novel approach for controlling the current in the NDR region eliminates the need of an external pulsed power source or external passive components to generate self-sustained electrical oscillations. Such reduction in circuitry complexity is significant for prospective large scale on-chip integration of micro- or nano- oscillators that simplifies the power source design and the connection between oscillators and metal interconnects carrying power to the oscillators. |
Wednesday, March 6, 2019 10:12AM - 10:24AM |
K33.00012: Scalable Free-Space Optical Neural Networks Liane Bernstein, Alexander Sludds, Ryan Hamerly, Dirk R. Englund The transformative impact of deep neural networks (DNNs) in many fields has motivated the development of hardware accelerators to improve speed and power consumption. We present a novel photonic approach based on homodyne detection where inputs and weights are encoded optically and can be reprogrammed and trained on the fly. This architecture is naturally adapted to free-space optics where both fully-connected and convolutional networks can be implemented and scaled to millions of neurons. By utilizing passive optical fan-out and performing arithmetic coherently with optical interference, this scheme circumvents fundamental limits of irreversible electronic processing. We study the effect of detector shot noise on neural-network accuracy to establish a “standard quantum limit” for this system. This bound, which can be as low as 50 zJ/FLOP, suggests performance below the Landauer (thermodynamic) limit is theoretically possible with photonics. |
Wednesday, March 6, 2019 10:24AM - 10:36AM |
K33.00013: Synaptic Barristor Based on Phase-Engineered Two-Dimensional Heterostructures Woong Huh, Seonghoon Jang, Jae Yoon Lee, Donghun Lee, Jung Min Lee, Hong-Gyu Park, Jong Chan Kim, Hu Young Jeong, Gunuk Wang, Chul-Ho Lee Heterostructures built from various two-dimensional (2D) layered materials are emerging material platforms for low-power and high-performance electronic devices because of their high-quality heterointerfaces. |
Wednesday, March 6, 2019 10:36AM - 10:48AM |
K33.00014: Resolving the Metal-Insulator Transition Mechanism of NbO2 for Memristor Applications Matthew Wahila, Galo Paez, Christopher Singh, Shawn Sallis, Jerzy T. Sadowski, Brooks Tellekamp, Alan Doolittle, Wei-Cheng Lee, Louis F. J. Piper Electroformed NbOx films show potential for memristor applications, however, their memristive mechanism is not understood. NbO2 filaments with a temperature-dependent MIT have been theorized to be the culprit,[1,2] but this has not been proven due in part to a lack of understanding regarding the MIT nature. Crystalline NbO2 undergoes an MIT with accompanying structural transition at ~810°C, but there are competing claims as to whether it is Mott, Peierls, or cooperative Mott-Peierls like VO2.[3,4] We have investigated crystalline NbO2 using synchrotron-based temperature-dependent and variable energy XPS, µ-LEED, and LEEM/PEEM, and have conclusively found it be a gradual 2nd-order Peierls transition. The gradual nature is then attributed to weakening Nb-Nb dimers causing the formation of a pseudogap as temperature increases. Moreover, calculations fully reproduce experimental spectra without explicit consideration of correlations, indicating minimal Mott character. |
Wednesday, March 6, 2019 10:48AM - 11:00AM |
K33.00015: Nb-doped TiO2: Effect of interstitial oxygen atom on charge state of Nb Wei Yan, Xiaojie Liu The structural and electronic properties of Nb doped Rutile TiO2 are investigated by first-principles calculations based on density functional theory. Several doping models including 2(NbTi)+Oi, NbTi+Oi, 2(NbTi), and NbTi are investigated. We find that all the doped systems preserve the semiconductor character but have an impurity state inside the band gap. Therefore, the band gaps of the doped systems are reduced by 0.15~1.41 eV, compared to that of the pure rutile TiO2. These results suggest that Nb-doped TiO2 would have higher photocatalytic activity than pure TiO2. The origin of the gap state is different in different doping models. The calculation results show that interstitial oxygen atom play an essential role in manipulating the valence state of Nb impurity. The charge density analysis show that Nb atom in 2(NbTi)+Oi and NbTi+Oi systems can be pentavalent since the Nb atom loses much more electrons due to the interaction with the interstitial oxygen atoms. On the other hand Nb atoms in the NbTi and 2(NbTi) doped systems would be tetravalent. |
Follow Us |
Engage
Become an APS Member |
My APS
Renew Membership |
Information for |
About APSThe American Physical Society (APS) is a non-profit membership organization working to advance the knowledge of physics. |
© 2024 American Physical Society
| All rights reserved | Terms of Use
| Contact Us
Headquarters
1 Physics Ellipse, College Park, MD 20740-3844
(301) 209-3200
Editorial Office
100 Motor Pkwy, Suite 110, Hauppauge, NY 11788
(631) 591-4000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 20045-2001
(202) 662-8700