Search

This paper describes the fabrication and characterization of Lithium Niobate (LiNbO3) memristor devices that have the ability to be tuned to a specific resistance state within a continuous resistance range. This is essential for programming neuromorphic systems based on memristor crossbars in order to achieve best deep learning capability. The memristor devices were formed using a 42 nm layer of LiNbO3 sandwiched between two metal electrodes. I-V curves demonstrate a typical and repeatable memristor characteristic from − 3 V to 3 V. Such devices have a continuous resistance range that has a maximum to minimum resistance ratio of about 100, and the ability to program intermediate resistance states. The results also show the ability to read the device symmetrically with a positive or negative voltage, and strong data retention after the programming phase.

In this paper, parallel plate varactor based phase shifter circuits using Barium Strontium Titanate (BST) thin films are presented. A cascade of 10, 15 and 20 parallel plate varactors were able to produce (150°, 258°, 297°), (218°, 381°,443°) and (227°, 402°, 464°) phase shift at (5, 10, 12 GHz) respectively. The 360° phase shift is achieved with small device size, low bias voltages (0-8 V), low leakage currents, low insertion loss and high figure of merit (FOM). FOM of 40, 33, 24 Degrees/dB is achieved with 10, 15 and 20 shunt varactors at 12 GHz respectively. The proposed circuit is very easy to fabricate which uses a CPW transmission line configuration and can be easily integrated with other circuits on chip.

This paper describes the fabrication and characterization process used to develop a series of lithium niobate memristors. A common approach for the development of memristor-based neuromorphic circuits is to store synaptic weight values within memristors as resistance states. Therefore, memristors for these systems must be stable, symmetric, and programmable with a significant bit resolution. In other words, a continuous resistance range must be available in these memristors to store the weight matrix produced by a learning algorithm. Furthermore, it is important to be able to iteratively program a target resistance through a number of feedback controlled voltage pulses as opposed to abruptly switching the device between two binary states. This paper describes the fabrication and characterization results for a set of six different memristor wafers. These results are used to decide which device composition is the best for neuromorphic computing applications through the properties of symmetry, reliability, stability, and programmability. Once the best device for neuromorphic programmability was selected, we also show the potential programming resolution available in this device using a voltage pulse characterization.

This work presents electrical characterization and analysis of the dominant charge transport mechanism suggesting inhomogeneous, filamentary conduction for a lithium niobate switching layer based memristor for use in neuromorphic computing. Memristor conductivity has been investigated both for the high and low resistance states. It is suggested that when the device is in a high resistance state, deep trap energy level within the switching layer initiate the device conduction process. The elastic trap assisted tunneling mechanism with a simple steady state approach agrees with the experimental measurements in the high resistance state. This work considers existence of inhomogeneously distributed positively charged oxygen ions/vacancies (within the oxygen deficient switching layer) as the deep trap energy level, required for electron tunneling from memristor electrode. Alternatively, ohmic conduction was found to be the main mechanism for the memristor on state conductivity at room temperature. Existence of intermediate resistive states in the memristor’s high resistive region was experimentally investigated and the elastic trap assisted tunneling mechanism for such phenomena was validated through simulation.

Memristor devices have the potential to drive a new class of specialized low power embedded hardware. The unique characteristics of these non-volatile and nanoscale devices allow them to perform parallel analog computing with extreme efficiency. To help facilitate the design of such systems, this paper describes the fabrication and characterization process used to develop memristors that are strong candidates for use in neuromorphic systems. In this work two different types of memristor devices, those with a GeTe switching layer, and those with a VO 2 switching layer, are characterized and analyzed. These results are used to determine device suitability for use in neuromorphic computing applications through the properties of symmetry, reliability, stability, and programmability. In short, repeatable multi-level resistive switching has been investigated and the results have been summarized.

Memristor crossbars are capable of implementing learning algorithms in a much more energy- and area-efficient manner compared with traditional systems. However, the programmable nature of memristor crossbars must first be explored on a smaller scale to see if physical devices are suitable for applications of reconfigurable computing. The reconfigurability of these devices through small scale memristor crossbar implementations is demonstrated. It is shown that a crossbar containing eight memristors is capable of learning several different two-input Boolean logic functions. A strong foundation is provided to build on demonstrating that physical memristor crossbars can be programmed as linear classifiers.

Tunable coplanar waveguide interdigital capacitors (IDC) designed with vanadium dioxide (VO2) thin film are presented in this paper. Two different configurations, series IDC and shunt IDC, are proposed. Tunable capacitance can be implemented by the thermally controllable VO2 thin film. The tunability of IDC structures are 95.6% and 85.4% corresponding to the series IDC and shunt IDC, respectively.

Memristor crossbars are capable of implementing learning algorithms in a much more energy and area efficient manner compared to traditional systems. However, the programmable nature of memristor crossbars must first be explored on a smaller scale to see which memristor device structures are most suitable for applications in reconfigurable computing. In this paper, we demonstrate the programmability of memristor devices with filamentary switching based on LiNbOx, a new resistive switching oxide. We have performed several characterization experiments that demonstrate the high resolution programmability of these devices. Once the optimal programming voltages and conductivity ranges for these memristor devices are determined, a method is presented to determine how many unique states can be achieved in these devices. Parameters obtained from this analysis can greatly strengthen memristor device models that implement realistic stochastic switching behavior. We show that 11 unique states can be stored in this device with 95% confidence, but many more states are achievable using feedback write techniques or in-situ learning. Therefore, we have determined that Lithium Niobate memristors are strong candidates for use in neuromorphic computing.

We present a design of a quantum information processing C-phase (Controlled-phase) gate applicable for generating cluster states that has a form of integrated photonic circuits assembled with cascaded directional couplers on a Ti in-diffused Lithium Niobate (Ti-LN) platform where directional couplers as the integrated optical analogue of bulk beam splitters are used as fundamental building blocks. Based on experimentally optimized fabrication parameters of Ti-LN optical waveguides operating at an 810nm wavelength, an integrated Ti-LN quantum C-phase gate is designed and simulated. Our proposed C-phase gate consists of three tunable directional couplers cascaded together with having different weighted switching ratios for providing a tool of routing vertically- and horizontally-polarized photons independently. Its operation mechanism relies on selectively controlling the optical coupling of orthogonally polarized modes via the change in the index of refraction, and its operation is confirmed by the BPM simulation.

This paper describes the fabrication and characterization process used to develop memristors that are strong candidates for use in neuromorphic systems. A common approach for the development of memristor-based neuromorphic circuits is to store synaptic weight values within memristors as resistance values. This requires use of the continuous resistance range available in the memristors to store the weight matrix produced by a learning algorithm. More specifically, these devices will be used in systems that implement supervised learning algorithms such as single and multilayer perceptrons. It is important to be able to iteratively program a target resistance through a number of feedback controlled voltage pulses as opposed to abruptly switching the device between two binary states. This paper shows how TiO 2 memristor devices placed in a crossbar arrangement are capable of implementing Boolean logic functions using a perceptron algorithm. Resistances within a memristor crossbar can be tuned and reconfigured so that a neuron circuit based on this crossbar can represent multiple different functions.

A resonant circuit combining a 3D inductor with the barium strontium titanate thin film varactor, is presented in this work. The filter was fabricated using a 3D inductor fabrication process. The modeling of the filter was examined using the Advanced Design System (ADS). The measurement results showed that the resonant frequencies were around 10 GHz and correlated to different number of turns of 3D inductor. The inductances extracted from the equivalent circuit varied from 0.28 nH to 0.37 nH.

The effort of investigating memristor material continues. This paper demonstrates the latest results of such effort by our group. These include memristor device design and measurement based on stacked lithium niobate and aluminum oxide. I-V sweeping results show good switch memristive characteristic. Other preliminary results in this paper include temperature dependency study, pulse voltage write/read and retention characterization. From these results, lithium niobate based device show the temperature stability and multiple stages of write/read and long retention period. All these demonstrate its potential for the application of neuromorphic computing in the future.

This paper describes the fabrication of memristor devices based on titanium and hafnium oxides. The device cross sectional area is varied to observe the impact this has on the current-voltage characteristic. A modeling technique is then utilized that is capable of matching the current-voltage characteristics of memristor devices. The model was able to match the titanium oxide device described in this paper with 13.58% error. The device model was then used in a neuromorphic simulation showing that a circuit based on this device is capable of learning logic functions.

Adaptive coded aperture (diffraction) sensing, an emerging technology enabling real-time, wide-area IR/visible sensing and imaging, could benefit from new high performance biologically inspired image processing architectures. The memristor, a novel two terminal passive device can enable significantly powerful biologically inspired processing architectures. This device was first theorized by Dr. Leon Chua in 1971. In 2008, HP Labs successfully fabricated the first memristor devices. Due to its unique properties, the memristor can be used to implement neuromorphic functions as its dynamics closely model those of a synapse, and can thus be utilized in biologically inspired processing architectures. This paper uses existing device models to determine how device parameters can be tuned for the memristor to be used in neuromorphic circuit design. Specifically, the relation between the different models and the number of states the device can hold are examined.

The memristor, experimentally verified for the first time in 2008, is one of four fundamental passive circuit elements (the others being resistors, capacitors, and inductors). Development and characterization of memristor devices and the design of novel computing architectures based on these devices can potentially provide significant advances in intelligence processing systems for a variety of applications including image processing, robotics, and machine learning. In particular, adaptive coded aperture (diffraction) sensing, an emerging technology enabling real-time, wide-area IR/visible sensing and imaging, could benefit from new high performance biologically inspired image processing architectures based on memristors. In this paper, we present results from the fabrication and characterization of memristor devices utilizing titanium oxide dielectric layers in a parallel plate conuration. Two versions of memristor devices have been fabricated at the University of Dayton and the Air Force Research Laboratory utilizing varying thicknesses of the TiO 2 dielectric layers. Our results show that the devices do exhibit the characteristic hysteresis loop in their I-V plots.

As semiconductor devices have shrunk further into the nanoscale regime, a new device, the memristor, has been discovered that has the potential to transform neuromorphic computing systems. This device is considered as the fourth fundamental circuit element. It was first theorized by Dr. Leon Chua in 1971 and has been discovered by HP labs in 2008. This paper describes initial efforts at fabricating the memristor devices and examining their properties. Two versions of memristor devices have been fabricated at the University of Dayton and the Air Force Research Laboratory utilizing varying thicknesses of the TiO 2 dielectric layers. Our results show that the devices do exhibit the characteristic hysteresis loop in their I-V plots. Further refinement in the devices to achieve stronger hysteresis will be carried out as future work.

There are several applications that require a thin layer of nanoparticles that are spread uniformly and without agglomeration over a surface. Among these applications are carbon nanotube synthesis and the development of fuel cells, advanced sensors, and optoelectronic devices. There is a need for improved processes of forming these nanoparticles. Among the numerous methods that have been developed, pulsed laser ablation has proved (ElShall et al., 1995; Burr et al., 1997, Seraphin et al., 1997, Geohegan et al., 1998; Lowndes et al, 1998, Makimura et al., 1998; Becker et al., 1998; Lowndes et al., 1999; Makino et al., 1999; Link and El-Sayed, 2000; Mafune et al., 2000; Mafune et al., 2001; Ogawa et al., 2000; Tang et al., 2001; Ozawa et al., 2001; Hata et al., 2001; Barnes et al., 2002) to be especially effective because of the potential for congruent ablation, the ability to produce nanoparticles of high purity, the ability to deposit nanoparticles on room temperature substrates, and the relative simplicity of the process. One problem with conventional pulsed laser ablation is the observation that the deposited material, in addition to containing nanoparticles, also contains large, m-sized particles. These are formed through a process denoted splashing (Ready, 1963), which occurs when a transient liquid layer is formed in the irradiated volume of the target; liquid droplets can be ejected from the target by the recoil pressure of the expanding gas on the transient liquid layer. Traditionally, splashing has been minimized (but not eliminated) by ablating very smooth targets or by using low laser energy densities. Target splashing is an issue in nanoparticle synthesis since the deposition of large particles within a field of nanoparticles makes it considerably more difficult to exploit the unique properties of isolated, non-agglomerated nanoparticles. Target splashing is also an issue because it represents material waste. There is, therefore, a need for a laser ablation process that further reduces or, ideally, eliminates splashing in nanoparticle synthesis. In addition, there is the further need for a process that minimizes nanoparticle agglomeration

We have developed a laser-based technique for fabricating thin films, nanoparticles, and nanocomposite thin films. The process is denoted Through Thin Film Ablation (TTFA) and entails using a thin film target that is ablated from the backside. The deposits produced by this process show neither evidence of the larger (micrometer-sized) particles nor the extent of agglomeration that are produced by conventional laser ablation. The TTFA technique offers the potential for fabricating a wide variety of materials in thin film, nanoparticle, and nanocomposite form. In this paper we present the results of Fe nanoparticle synthesis by TTFA including nanoparticle size distributions, time of flight distributions, and time resolved spectral data. These data provide a more complete understanding of the TTFA process.

One of the critical aspects of nanotechnology is to assemble different nanoscale components into a single system. In such a multicomponent system, the overall functionality depends strongly on the precise location and structural characteristics of each of the constituent components. In this context, we have prepared multicomponent micropatterns of silica particles interposed within the discrete areas of aligned multiwall carbon nanotubes. The patterns were fabricated by dry contact transferring aligned carbon nanotubes onto a tape pre-patterned with a thin layer of gold structure, followed by region-specific adsorption of thiol-modified silica particles onto the gold surface from solution. The dry contact transfer technique has further enabled us to develop micropatterns of aligned single-wall carbon nanotubes with interdispersed non-aligned multiwall carbon nanotubes and microsized carbon fibers sheathed with micropatterned aligned carbon nanotubes.

The interaction of low-frequency electromagnetic waves with metallic nanostructure consisting of nanoparticles has been investigated. The existence of helicons in metallic nanostructure is predicted based on the enhancement of effective mass of the electrons in metallic mesostructures in low-frequency electromagnetic field. The enhancement of the effective mass of electrons subjected to low-frequency electromagnetic field results in the suppression of the cyclotron frequency. When the excitation frequency approaches the cyclotron frequency, helicons can be observed, which in typical metals occur at MHz frequencies. With the decrease in the cyclotron frequency in nanostructures, the helicons can be observed at low frequencies. We present an experimental setup to detect and image helicon resonances in metallic nanostructures at low frequencies and room temperature conditions. The approach is based on modifying an existing atomic force microscope to image surface topography and the magnetic field images of h...