Nanometer resolution stress measurement of the Si gate using illumination-collection-type scanning near-field Raman spectroscopy with a completely metal-inside-coated pyramidal probe.

We have proposed an illumination-collection-type scanning near-field Raman spectroscopy (SNRS) with a completely gold metal-inside-coated (MIC) pyramidal probe without an optical aperture in order to detect the Raman spectra of fine Si devices for local stress measurements. The gold MIC pyramidal probe has been studied to act as a plasmon resonance near-field optical probe with high power using a finite differential time domain (FDTD) simulation and the prototyped SNRS. In the simulation, the propagated optical power can be made available for SNRS. In the experiments, it is clear that the prototyped SNRS enhanced the Si Raman peak signal by plasmon resonance and could measure the Si Raman peak shift by line scanning the Si gate region and the Si active layer. Furthermore, compressive and tensile stresses localized around the Si gate were demonstrated by the Si Raman peak shift with a resolution of about 10 nm. It is clarified that the proposed SNRS has the possibility of detecting the Raman spectra of a local area.

Observation of Si pattern sidewall using inclination atomic force microscope for evaluation of line edge roughness.

Inclination atomic force microscope (AFM) imaging has been studied on the possibility to observe a pattern sidewall in contact mode or digital probing (step-in) mode for a line edge roughness (LER) or line width roughness (LWR). Analysis of the AFM tip bending and slipping indicates that it is serious problem to measure and control very fine patterns within an error of less than 1 nm in contact of the tip on the steep slop of the pattern, and it is very important directly to observe the sidewall at inclination angle. In experiments using pyramidal tip and steep Si pattern with about 90 degrees slop, it has demonstrated that the inclination angle is 35-40 degrees for faithful observation of the sidewall. We have observed the etched strip lines on the sidewall with a width of about 100 nm and a depth of about 6.4 nm. We have demonstrated that the inclination AFM is very useful for evaluation of the LER or LWR.

Handwritten-Digit Recognition by Hybrid Convolutional Neural Network based on HfO2 Memristive Spiking-Neuron.

Although there is a huge progress in complementary-metal-oxide-semiconductor (CMOS) technology, construction of an artificial neural network using CMOS technology to realize the functionality comparable with that of human cerebral cortex containing 1010-1011 neurons is still of great challenge. Recently, phase change memristor neuron has been proposed to realize a human-brain level neural network operating at a high speed while consuming a small amount of power and having a high integration density. Although memristor neuron can be scaled down to nanometer, integration of 1010-1011 neurons still faces many problems in circuit complexity, chip area, power consumption, etc. In this work, we propose a CMOS compatible HfO2 memristor neuron that can be well integrated with silicon circuits. A hybrid Convolutional Neural Network (CNN) based on the HfO2 memristor neuron is proposed and constructed. In the hybrid CNN, one memristive neuron can behave as multiple physical neurons based on the Time Division Multiplexing Access (TDMA) technique. Handwritten digit recognition is demonstrated in the hybrid CNN with a memristive neuron acting as 784 physical neurons. This work paves the way towards substantially shrinking the amount of neurons required in hardware and realization of more complex or even human cerebral cortex level memristive neural networks.

Scanning electron microscope for in situ study of crystallization of Ge2Sb2Te5 in phase-change memory.

By introducing electrical connections into the chamber of a scanning electron microscope (SEM) via its holder assembly, it has become feasible to in situ observe and electrically characterize electronic devices. The in situ SEM was applied to investigate electric-pulse-induced behavior of Ge(2)Sb(2)Te(5) in a lateral phase-change memory cell. Randomly distributed nuclei with sizes from 20 to 80 nm were initiated at a low voltage pulse. Initially, grain growth depended strongly on pulse amplitude at around 60.3 nm/V and then a weak pulse amplitude dependence was observed at around 13.5 nm/V. Device resistance during crystallization dropped by two to three orders of magnitude with two falling steps, which probably resulted from amorphous to face-centered-cubic and subsequently to hexagonal transitions, respectively.

Associative memory realized by a reconfigurable memristive Hopfield neural network.

Although synaptic behaviours of memristors have been widely demonstrated, implementation of an even simple artificial neural network is still a great challenge. In this work, we demonstrate the associative memory on the basis of a memristive Hopfield network. Different patterns can be stored into the memristive Hopfield network by tuning the resistance of the memristors, and the pre-stored patterns can be successfully retrieved directly or through some associative intermediate states, being analogous to the associative memory behaviour. Both single-associative memory and multi-associative memories can be realized with the memristive Hopfield network.

Controlling of 6 nm sized and 10 nm pitched dot arrays ordered along narrow guide lines using PS-PDMS self-assembly.

We have studied graphoepitaxy to make nanodots or nanolines ordered along electron beam (EB)-drawn resist guide using block copolymers (BCPs) of polystyrene-polydimethylsiloxane (PS-PDMS). We found out that the number n of ordered molecular dot arrays in the line gap increases stepwise with the gap between guide lines. The n self-assembled dot arrays were ordered in a gap between n and n+1 times the mean PDMS pitch and self-assembled with no guide pattern. According to the ordering characteristics, 6 nm sized and 10 nm pitched PDMS dot arrays were formed using the BCP self-assembly with the guide lines.

Self-assembled incorporation of modulated block copolymer nanostructures in phase-change memory for switching power reduction.

Phase change memory (PCM), which exploits the phase change behavior of chalcogenide materials, affords tremendous advantages over conventional solid-state memory due to its nonvolatility, high speed, and scalability. However, high power consumption of PCM poses a critical challenge and has been the most significant obstacle to its widespread commercialization. Here, we present a novel approach based on the self-assembly of a block copolymer (BCP) to form a thin nanostructured SiOx layer that locally blocks the contact between a heater electrode and a phase change material. The writing current is decreased 5-fold (corresponding to a power reduction by 1/20) as the occupying area fraction of SiOx nanostructures is increased from a fill factor of 9.1% to 63.6%. Simulation results theoretically explain the current reduction mechanism by localized switching of BCP-blocked phase change materials.

Formation of dot arrays with a pitch of 20 nm × 20 nm for patterned media using 30 keV EB drawing on thin calixarene resist.

We studied the possibility of achieving very fine-pitch dot arrays with a pitch of 20 nm × 20 nm using 30 keV electron beam (EB) drawing on negative calixarene resist. In order to form such patterns, we studied the dependence on resist thickness of the dot size and the packing. We propose EB drawing on an extremely thin film for very highly packed dot-array formation. Our experimental results demonstrate the possibility of forming highly packed dot-array patterns with a pitch of 20 nm × 20 nm and a resist thickness of about 13 nm, which corresponds to about 1.6 Tbits in(-2).