A5009973223- Phase-change materials and chalcogenides
- Advanced Memory and Neural Computing
- Chalcogenide Semiconductor Thin Films
- Ferroelectric and Negative Capacitance Devices
- Nonlinear Optical Materials Studies
- Semiconductor materials and devices
- Quantum Dots Synthesis And Properties
- Neuroscience and Neural Engineering
- Advanced Semiconductor Detectors and Materials
- Neural Networks and Reservoir Computing
- Advanced Thermoelectric Materials and Devices
- Neural Networks and Applications
- Transition Metal Oxide Nanomaterials
- 2D Materials and Applications
- Electronic and Structural Properties of Oxides
- Machine Learning and ELM
- CCD and CMOS Imaging Sensors
- Topological Materials and Phenomena
- Liquid Crystal Research Advancements
- Adhesion, Friction, and Surface Interactions
- Electrochemical Analysis and Applications
- Advancements in Semiconductor Devices and Circuit Design
- Terahertz technology and applications
- Solid-state spectroscopy and crystallography
- Integrated Circuits and Semiconductor Failure Analysis
IBM Research - Zurich
2020-2025
Paul Drude Institute for Solid State Electronics
2014-2022
Forschungsverbund Berlin
2018-2022
Abstract Phase Change Materials (PCMs) are unique compounds employed in non-volatile random access memory thanks to the rapid and reversible transformation between amorphous crystalline state that display large differences electrical optical properties. In addition amorphous-to-crystalline transition, experimental results on polycrystalline GeSbTe alloys (GST) films evidenced a Metal-Insulator Transition (MIT) attributed disorder phase. Here we report fundamental advance fabrication of GST...
Neuromorphic computing architectures enable the dense colocation of memory and processing elements within a single circuit. This removes communication bottleneck transferring data between separate units as in standard von Neuman for data-critical applications including machine learning. The essential building blocks neuromorphic systems are nonvolatile synaptic such memristors. Key memristor properties include suitable resistance range, continuous linear modulation, symmetric switching. In...
Sb2Te3 films are used for studying the epitaxial registry between two-dimensionally bonded (2D) materials and three-dimensional (3D) substrates. In contrast to growth of 3D materials, it is found that formation coincidence lattices Si(111) depends on geometry dangling bonds reconstructed substrate surface. Furthermore, we show can be influenced by controlling surface reconstruction confirm results ultrathin films.
Abstract The technological success of phase-change materials in the field data storage and functional systems stems from their distinctive electronic structural peculiarities on nanoscale. Recently, superlattice structures have been demonstrated to dramatically improve optical electrical performances these chalcogenide based materials. In this perspective, unravelling atomistic structure that originates improvements switching time energy is paramount order design nanoscale with even enhanced...
In the present work, GeTe and Sb2Te3 van der Waals bonded superlattices epitaxially grown by molecular beam epitaxy are investigated. These structures on passivated Si substrates, resulting in one single epitaxial domain its twinned domain, both sharing same out-of-plane orientation. Supported X-ray diffraction Raman spectroscopy, attention is called to thermodynamically driven tendency of intermix into a Ge–Sb–Te (GST) alloy at interfaces. A growth model proposed explain how these GST formed.
Abstract A combination of far-infrared and Raman spectroscopy is employed to investigate vibrational modes the carrier behavior in amorphous crystalline ordered GeTe-Sb 2 Te 3 alloys (GST) epitaxially grown on Si(111). The infrared active GST mode not observed spectra vice versa, indication fact that inversion symmetry preserved metastable cubic phase accordance with F"Equation missing"<!-- image only, no MathML or LaTex -->m3 space group. For trigonal phase, instead, a partial break due...
The thermal and electrical properties of phase change materials, mainly GeSbTe alloys, in the crystalline state strongly depend on their associated degree order. switching Ge atoms superlattice structures with trigonal has been recently proposed to develop memories reduced energy, which two differently ordered phases are logic states. A detailed knowledge stacking plane sequence, local composition vacancy distribution is therefore crucial order understand underlying mechanism transformations...
Abstract The in‐memory computing paradigm aims at overcoming the intrinsic inefficiencies of Von‐Neumann computers by reducing data‐transport per arithmetic operation. Crossbar arrays multilevel memristive devices enable efficient calculations matrix‐vector‐multiplications, an operation extensively called on in artificial intelligence (AI) tasks. Resistive random‐access memories (ReRAMs) are promising candidate for such applications. However, they generally exhibit large stochasticity and...
Abstract Brain-inspired computing emerged as a forefront technology to harness the growing amount of data generated in an increasingly connected society. The complex dynamics involving short- and long-term memory are key undisputed performance biological neural networks. Here, we report on sub-µm-sized artificial synaptic weights exploiting combination ferroelectric space charge effect oxidation state modulation oxide channel field transistor. They lead quasi-continuous resistance tuning...
Abstract With remarkable electrical and optical switching properties induced at low power near room temperature (68 °C), vanadium dioxide (VO 2 ) has sparked rising interest in unconventional computing among the phase-change materials research community. The scalability potential to compute beyond von Neumann model make VO especially appealing for implementation oscillating neural networks artificial intelligence applications, solve constraint satisfaction problems, pattern recognition. Its...
Analog in-memory computing is an emerging paradigm designed to efficiently accelerate deep neural network workloads. Recent advancements have demonstrated significant improvements in throughput and efficiency, focusing independently on either inference or training acceleration. However, a unified analog technology platform-capable of performing on-chip training, retaining the weights, sustaining long-term acceleration-has yet be reported. In this work, all-in-one AI accelerator presented...
The structural change upon annealing of an amorphous GeSbTe (GST) film deposited by molecular beam epitaxy on a Si(111) substrate is studied means X-ray diffraction (XRD), reflectivity (XRR), and atomic force microscopy (AFM). XRD profiles reveal that both metastable cubic stable hexagonal phases are obtained with single out-of-plane orientation. XRR study shows density increase consequent thickness decrease annealing, in accordance literature. From both, the AFM study, it emerges...
A critical bottleneck for the training of large neural networks (NNs) is communication with off-chip memory. promising mitigation effort consists integrating crossbar arrays analogue memories in Back-End-Of-Line, to store NN parameters and efficiently perform required synaptic operations. The "Tiki-Taka" algorithm was developed facilitate presence device nonidealities. However, so far, a resistive switching exhibiting all fundamental Tiki-Taka requirements, which are many programmable...
With remarkable electrical and optical switching properties induced at low power near room temperature (68C), vanadium dioxide (VO2) has sparked rising interest in unconventional computing among the phase-change materials research community. The scalability potential to compute beyond von Neumann model make VO2 especially appealing for implementation oscillating neural networks artificial intelligence (AI) applications, solve constraint satisfaction problems, pattern recognition. Its...
Abstract In this study we report on the investigation of epitaxially grown Sb 2 Te 3 by employing Fourier-Transform transmission Spectroscopy (FTS) with laser-induced Coherent Synchrotron Radiation (CSR) in Terahertz (THz) spectral range. Static spectra range between 20 and 120 cm −1 highlight a peculiar softening an in-plane IR-active phonon mode upon temperature decrease, as opposed to all Raman active modes which instead show hardening decrease same energy The is found be accompanied...
We show for the first time in hardware that contrast to conventional stochastic gradient descent (SGD), our modified SGD algorithm (TTv2) together with a co-optimized ReRAM material achieves respectable accuracy (98%) on reduced MNIST classification (0 & 1), approaching floating point (FP) baseline. To extrapolate these insights towards larger DNN training workloads simulations, we establish an analog switching test sequence and extract key device statistics from 6T1R arrays (up 2k devices)...
Abstract Phase change alloys are used for non-volatile random-access memories exploiting the conductivity contrast between amorphous and metastable, crystalline phase. However, this has never been directly related to electronic band structure. Here we employ photoelectron spectroscopy map relevant bands epitaxial GeSbTe films. The constant energy surfaces of valence close Fermi level hexagonal tubes with little dispersion perpendicular (111) surface. electron density responsible transport...
Phase change materials such as pseudobinary GeTe-Sb2Te3 (GST) alloys are an essential part of existing and emerging technologies. Here, we investigate the electrical optical properties epitaxial phase materials: α-GeTe, Ge2Sb2Te5 (GST225), Sb2Te3. Temperature-dependent Hall measurements reveal a reduction hole concentration with increasing temperature in Sb2Te3 that is attributed to lattice expansion, resulting non-linear increase resistivity also observed GST225. Fourier transform infrared...
Phase-change materials based on Ge-Sb-Te alloys are widely used in industrial applications such as nonvolatile memories, but reaction pathways for crystalline-to-amorphous phase-change picosecond timescales remain unknown. Femtosecond laser excitation and an ultrashort x-ray probe is to show the temporal separation of electronic thermal effects a long-lived ($>$100 ps) transient metastable state Ge$_{2}$Sb$_{2}$Te$_{5}$ with muted interatomic interaction induced by weakening resonant...
<title>Abstract</title> Advancements in memristive devices have given rise to a new generation of specialized hardware for bio-inspired computing. However, the majority these implementations only draw partial inspiration from architecture and functionalities mammalian brain. Moreover, use is typically restricted specific elements within learning algorithm, leaving computationally expensive operations be executed software. Here, we demonstrate actor-critic temporal difference (TD) on analogue...
Abstract Phase change memory (PCM) is being actively explored for in-memory computing and neuromorphic systems. The ability of a PCM device to store continuum resistance values can be exploited realize arithmetic operations such as matrix-vector multiplications or the synaptic efficacy in neural networks. However, variations arising from structural relaxation, 1/f noise, changes ambient temperature pose key challenge. recently proposed projected concept helps mitigate these by decoupling...
Abstract In this study, a generalized guideline is identified to predict the interaction between two‐dimensional (2D) layered materials and substrate surfaces. Additionally, van der Waals (vdW) heterostructures commensurability, phase formation strain relaxation are during interface growth. To achieve such general overview, case of Ge‐Sb‐Te (GST) alloys on InAs(111) studied. system, low‐lattice mismatch conditions fulfilled avoid due misfit dislocations allow correctly identify vdW epitaxy....
GeSbTe-based materials exhibit multiple crystalline phases, from disordered rocksalt, to rocksalt with ordered vacancy layers, and the stable trigonal phase. In this paper we investigate role of interfaces on structural electrical properties Ge2Sb2Te5. We find that site nucleation metastable phase is crucial in determining evolution towards ordering By properly choosing substrate capping sites engineering can be obtained, thus promoting or preventing structure conversion into The occurs at...
A Ferroelectric, two-terminals, analog memristive device is fabricated with a Back-End-Of-Line, CMOS compatible process. bilayer composed of ferroelectric material, HfZrO4 (HZO) and semiconducting oxide, WOx layer comprised between two TiN electrodes. The devices demonstrate reversible remanent resistive switching, record endurance (>1010 switching cycles) ON/OFF ratio up to 10. obtained, cycle-to-cycle reproducibility 90%. synaptic behavior explored pulses varying sign, length, or...