Publikationen 2021

(1) Tackling the Low Conductance State Drift through Incremental Reset and Verify in RRAM Arrays
A. Baroni, C. Zambelli, P. Olivo, E. Perez, Ch. Wenger, D. Ielmini
Proc. IEEE International Integrated Reliability Workshop (IIRW 2021), (2021)
(KI-IoT)
Resistive switching memory (RRAM) is a promising technology for highly efficient computing scenarios. RRAM arrays enabled the acceleration of neural networks for artificial intelligence and the creation of In-Memory Computing circuits. However, the arrays are affected by several issues materializing in conductance variations that might cause severe performance degradation in those applications. Among those, one is related to the drift of the low conductance states appearing immediately at the end of program and verify algorithms that are fundamental for an accurate Multi-level conductance operation. In this work, we tackle the issue by developing an Incremental Reset and Verify technique showing enhanced variability and reliability features compared with a traditional refresh-based approach.

(2) Biomolecule Sensing in THz Range with N-Ge/Si Antennas
C.A. Chavarin, A.A. Wiciak, E. Hardt, S. Gruessing, O. Skibitzki, I. Costina, W. Seifert, W.M. Klesse, C.L. Manganelli, C. You, J. Flesch, J. Piehler, M. Missori, W. Koczorowski, L. Baldassarre, B. Witzigmann, G. Capellini, D. Spirito
Proc. 46^th International Conference on Infrared, Millimeter, and Terahertz Waves (IRMMW-THz 2021), (2021)
DOI: 10.1109/IRMMW-THz50926.2021.9567579, (DFG-ESSENCE)

(3) N-Type Ge/Si Antennas for THz Sensing
C.A. Chavarin, E. Hardt, S. Gruessing, O. Skibitzki, I. Costina, D. Spirito, W. Seifert, W.M. Klesse, C.L. Manganelli, C. You, J. Flesch, J. Piehler, M. Missori, L. Baldassarre, B. Witzigmann, G. Capellini
Optics Express 29(5), 7680 (2021)
DOI: 10.1364/OE.418382, (DFG-ESSENCE)
Ge-on-Si plasmonics holds the promise for compact and low-cost solutions in the manipulation of THz radiation. We discuss here the plasmonic properties of doped Ge bow-tie antennas made with a low-point cost CMOS mainstream technology. These antennas display resonances between 500 and 700 GHz, probed by THz Time Domain Spectroscopy. We show surface functionalization of the antennas with a thin layer of α-lipoic acid, that red-shifts the antenna resonances by about 20 GHz. Moreover, we show that also antennas covered with a protective silicon nitride cap layer exhibit a comparable red-shift when covered with the biolayer. This suggests that the electromagnetic fields at the hotspot extend well beyond the cap layer, enabling the possibility to use the antennas with an improved protection of the plasmonic material in conjunction with microfluidics.

(4) Perspectives on Electrically Pumped Ge/SiGe QW Emitters at THz Frequencies
C. Ciano, M. Montanari, L. Persichetti, D. Stark, G. Scalari, J. Faist, L. Di Gaspare, G. Capellini, C. Corley, T. Grange, S. Birner, M. Virgilio, L. Baldassarre, M. Ortolani, M. De Seta
Proc. 45^th International Conference on Infrared, Millimeter, and Terahertz Waves (IRMMW-THz 2020), (2021)
DOI: 10.1109/IRMMW-THz46771.2020.9370671, (FLASH)

(5) Liquid-Phase Exfoliated Gallium Selenide for Light-Driven Thin-Film Transistors
N. Curreli, M. Serri, M.I. Zappia, D. Spirito, G. Bianca, J. Buha, L. Najafi, Z. Sofer, R. Krahne, V. Pellegrini, F. Bonaccorso
Advanced Electronic Materials 7(3), 2001080 (2021)
DOI: 10.1002/aelm.202001080
Gallium selenide (GaSe), a layered semiconductor of Group‐III monochalcogenides, has been recognized by the scientific community in recent years as an appealing material in the fields of photonics and (opto)electronics. Thanks to its pseudodirect bandgap and its thickness‐dependent (opto)electronic properties, GaSe has emerged as a promising candidate for the implementation of thin‐film transistors (TFTs) and photodetectors with fast response and high sensitivity. Solution processing of 2D materials provides low‐cost inks that allow the design and realization of printed electronic devices, enabling this technology to move from the laboratory to the industry. In this work, a solution‐processed GaSe‐based light‐driven transistor is presented. Liquid phase exfoliation (LPE) is used to exfoliate bulk GaSe in isopropanol, formulating a functional ink that is subsequently deposited by spray coating onto Si/SiO₂ substrates. The GaSe phototransistor exhibits a p‐channel behavior with a high on/off ratio (≈10³) that is gate‐voltage dependent. Moreover, the device response also depends on the illumination with a maximum responsivity of 13 A W^–1 to UV–visible light and a fast response time of 35 ms. This study demonstrates that liquid phase exfoliated GaSe is a promising candidate for the design and realization of next‐generation (opto)electronic devices.

(6) Mechanical Switching of Orientation-Related Photoluminescence in Deep-Blue 2D Layered Perovskite Ensembles
B. Dhanabalan, A. Castelli, L. Ceseracciu, D. Spirito, F. Di Stasio, L. Manna, R. Krahne, M.P. Arciniegas
Nanoscale 13(7), 3948 (2021)
DOI: 10.1039/D0NR08043H
The synergy between the organic component of two-dimensional (2D) metal halide layered perovskites and flexible polymers offers an unexplored window to tune their optical properties at low mechanical stress. Thus, there is a significant interest in exploiting their PL anisotropy by controlling their orientation and elucidating their interactions. Here, we apply this principle to platelet structures of micrometre lateral size that are synthesized in situ into free-standing polymer films. We study the photoluminescence of the resulting films under cyclic mechanical stress and observe an enhancement in the emission intensity up to ∼2.5 times along with a switch in the emission profile when stretching the films from 0% to 70% elongation. All the films recovered their initial emission intensity when releasing the stress throughout ca. 15 mechanical cycles. We hypothesize a combined contribution from reduced reabsorption, changes on in-plane and out-of-plane dipole moments that stem from different orientation of the platelets inside the film, and relative sliding of platelets within oriented stacks while stretching the films. Our results reveal how low-mechanical stress affects 2D layered perovskite aggregation and orientation, an open pathway toward the design of strain-controlled emission.

(7) Three-Dimensional Interfacing of Cells with Hierarchical Silicon Nano/Microstructures for Midinfrared Interrogation of In Situ Captured Proteins
J. Flesch, M. Bettenhausen, M. Kazmierczak, W.M. Klesse, O. Skibitzki, O.E. Psathaki, R. Kurre, G. Capellini, S. Guha, T. Schroeder, B. Witzigmann, C. You, J. Piehler
ACS Applied Materials & Interfaces 13(7), 8049 (2021)
DOI: 10.1021/acsami.0c22421, (DFG-ESSENCE)
Label-free optical detection of biomolecules is currently limited by a lack of specificity rather than sensitivity. To exploit the much more characteristic refractive index dispersion in the mid-infrared (IR) regime, we have engineered three-dimensional IR-resonant silicon micropillar arrays (Si-MPAs) for protein sensing. By exploiting the unique hierarchical nano- and microstructured design of these Si-MPAs attained by CMOS-compatible silicon-based microfabrication processes, we achieved an optimized interrogation of surface protein binding. Based on spatially resolved surface functionalization, we demonstrate controlled three-dimensional interfacing of mammalian cells with Si-MPAs. Spatially controlled surface functionalization for site-specific protein immobilization enabled efficient targeting of soluble and membrane proteins into sensing hotspots directly from cells cultured on Si-MPAs. Protein binding to Si-MPA hotspots at submonolayer level was unambiguously detected by conventional Fourier transform IR spectroscopy. The compatibility with cost-effective CMOS-based microfabrication techniques readily allows integration of this novel IR transducer into fully fledged bioanalytical microdevices for selective and sensitive protein sensing.

(8) Mitigating the Effects of RRAM Process Variation on the Accuracy of Artifical Neural Networks
M. Fritscher, J. Knödtel, M. Mallah, S. Pechmann, E. Perez-Bosch Quesada, T. Rizzi, Ch. Wenger, M. Reichenbach
Proc. 21^st International Conference on Embedded Computer Systems: Architectures, Modeling and Simulation (SAMOS 2021), (2021)
(NeuroMem)
In this work, three different RRAM compact models implemented in Verilog-A are analyzed and evaluated in order to reproduce the multilevel approach based on the switching capability of experimental devices. These models are integrated in 1T-1R cells to control their analog behavior by means of the compliance current imposed by the NMOS select transistor. Four different resistance levels are simulated and assessed with experimental verification to account for their multilevel capability. Further, an Artificial Neural Network study is carried out to evaluate in a real scenario the viability of the multilevel approach under study.

(9) Mitigating the Effects of RRAM Process Variation on the Accuracy of Artifical Neural Networks
M. Fritscher, J. Knödtel, M. Mallah, S. Pechmann, E. Perez-Bosch Quesada, T. Rizzi, Ch. Wenger, M. Reichenbach
Proc. 21^st International Conference on Embedded Computer Systems: Architectures, Modeling and Simulation (SAMOS 2021), (2021)
(Total Resilience)
In this work, three different RRAM compact models implemented in Verilog-A are analyzed and evaluated in order to reproduce the multilevel approach based on the switching capability of experimental devices. These models are integrated in 1T-1R cells to control their analog behavior by means of the compliance current imposed by the NMOS select transistor. Four different resistance levels are simulated and assessed with experimental verification to account for their multilevel capability. Further, an Artificial Neural Network study is carried out to evaluate in a real scenario the viability of the multilevel approach under study.

(10) Mitigating the Effects of RRAM Process Variation on the Accuracy of Artifical Neural Networks
M. Fritscher, J. Knödtel, M. Mallah, S. Pechmann, E. Perez-Bosch Quesada, T. Rizzi, Ch. Wenger, M. Reichenbach
Proc. 21^st International Conference on Embedded Computer Systems: Architectures, Modeling and Simulation (SAMOS 2021), (2021)
(KI-PRO)
In this work, three different RRAM compact models implemented in Verilog-A are analyzed and evaluated in order to reproduce the multilevel approach based on the switching capability of experimental devices. These models are integrated in 1T-1R cells to control their analog behavior by means of the compliance current imposed by the NMOS select transistor. Four different resistance levels are simulated and assessed with experimental verification to account for their multilevel capability. Further, an Artificial Neural Network study is carried out to evaluate in a real scenario the viability of the multilevel approach under study.

(11) The Formation of a Sn Monolayer on Ge(100) Studied at the Atomic Scale
E.V.S. Hofmann, E. Scalise, F. Montalenti, T.J.Z. Stock, S.R. Schofield, G. Capellini, L. Miglio, N.J. Curson, W.M. Klesse
Applied Surface Science 561, 149961 (2021)
DOI: 10.1016/j.apsusc.2021.149961
The growth of multi-layer germanium-tin (GeSn) quantum wells offers an intriguing pathway towards the integration of lasers in a CMOS platform. An important step in growing high quality quantum well interfaces is the formation of an initial wetting layer. However, key atomic-scale details of this process have not previously been discussed. We use scanning tunneling microscopy combined with density functional theory to study the deposition of Sn on Ge(1 0 0) at room temperature over a coverage range of 0.01 to 1.24 monolayers. We demonstrate the formation of a sub-2% Ge content GeSn wetting layer from three atomic-scale characteristic ad-dimer structural components, and show that small quantities of Sn incorporate into the Ge surface forming two atomic configurations. The ratio of the ad-dimer structures changes with increasing Sn coverage, indicating a change in growth kinetics. At sub-monolayer coverage, the least densely packing ad-dimer structure is most abundant. As the layer closes, forming a two-dimensional wetting layer, the more densely packing ad-dimer structure become dominant. These results demonstrate the capability to form an atomically smooth wetting layer at room temperature, and provide critical atomic-scale insights for the optimization of growth processes of GeSn multi-quantum-wells to meet the quality requirements of optical GeSn-based devices.

(12) AC Electrokinetic on the Nanoscale: Immobilisation of Nanoparticles and Molecules
R. Hölzel, X. Knigge, E.-M. Laux, M. Noffke, S. Stanke, Ch. Wenger, F.F. Bier
Proc. 3^rd European BioSensor Symposium (EBS 2021), abstr. (2021)
(exosurf)

(13) Oxides based Resistive Switching Memories
S. Kalem, S. B. Tekin, Z. E. Kaya, E. Jagaluier, R. Roelofs, S. Yildirim, O. Yavuzcetin, Ch. Wenger
Proc. SPIE Oxide-based Materials and Devices XII (SPIE OPTO 2021), 11687, 116871L (2021)
DOI: 10.1117/12.2585681, (Panache)

(14) Precipitation of Suboxides in Silicon
G. Kissinger, D. Kot, A. Huber, R. Kretschmer, T. Müller, A. Sattler
ECS Meeting Abstracts MA2021-01(34), 1093 (2021)
DOI: 10.1149/MA2021-01341093mtgabs, (Future Silicon Wafers)

(15) Fast Scatterometric Measurement of Periodic Surface Structures in Plasma-Etching Processes
W.M. Klesse, A. Rathsfeld, C. Groß, E. Malguth, O. Skibitzki, L. Zealouk
Measurement 170, 108721 (2021)
DOI: 10.1016/j.measurement.2020.108721
To satisfy the continuous demand of ever smaller feature sizes, plasma etching technologies in microelectronics enable the fabrication of device structures in the nanometer range. To control these processes, real-time access to the structure’s dimensions is needed. We develop a special method of optical critical dimension metrology and evaluate the feasibility of reconstructing the etched dimensions from experimental reflectivity spectra of the surface, taken about every second. For a periodic 2D model structure etched into a silicon, we develop and test a fast algorithm.
To reduce the computing time, we generate a library of spectra before the etching. We demonstrate that, by replacing the numerically simulated spectra in the reconstruction algorithm by spectra interpolated from the library, it is possible to compute the geometry parameters in times less than a second. Finally, to also reduce memory size and computing time for the library, we reduce the scanning of the parameter values to a sparse grid.

(16) Adaptation Strategies for Personalized Gait Neuroprosthetics
A.D. Koelewijn, M. Audu, A.J. del-Ama, A. Colucci, J.M. Font-Llagunes, A. Gogeascoechea Hernandez, S. Hnat, N. Makowski, J.C. Moreno, M. Nandor, R. Quinn, M. Reichenbach, R.-D. Reyes, M. Sartori, S. Soekader, R.J. Triolo, M. Vermehren, Ch. Wenger, U.S. Yavuz, D. Fey, P. Beckerle
Frontiers in Neurorobotics 15, 750619 (2021)

(17) AC Electrokinetic Immobilization of K562 Exosomes on Nanoelectrode Arrays
E.-M. Laux, Ch. Wenger, R. Hölzel
Proc. 3^rd European BioSensor Symposium (EBS 2021), abstr. (2021)
(exosurf)

(18) Influence of Plasma Treatment on SiO₂/Si and Si₃N₄/Si Substrates for Large-Scale Transfer of Graphene
R. Lukose, M. Lisker, F. Akhtar, M. Fraschke, T. Grabolla, A. Mai, M. Lukosius
Scientific Reports 11, 13111 (2021)
DOI: 10.1038/s41598-021-92432-4, (GIMMIK)
One of the limiting factors of graphene integration into electronic, photonic, or sensing devices is the unavailability of large-scale graphene directly grown on the isolators. Therefore, it is necessary to transfer graphene from the donor growth wafers onto the isolating target wafers. In the present research, graphene was transferred from the chemical vapor deposited 200mm Germanium/Silicon (Ge/Si) wafers onto isolating (SiO₂/Si and Si₃N₄/Si) wafers by electrochemical delamination procedure, employing poly(methylmethacrylate) as an intermediate support layer. In order to influence the adhesion properties of graphene, the wettability properties of the target substrates were investigated in this study. To increase the adhesion of the graphene on the isolating surfaces, they were pre-treated with oxygen plasma prior the transfer process of graphene. The wetting contact angle measurements revealed the increase of the hydrophilicity after surface interaction with oxygen plasma, leading to improved adhesion of the graphene on 200 mm target wafers and possible proof-of-concept development of graphene-based devices in standard Si technologies.

(19) Low-Temperature Atomic Layer Deposition of Indium Oxide Thin Films using Trimethylindium and Oxygen Plasma
A. Mahmoodinezhad, C. Morales, F. Naumann, P. Plate, R. Meyer, C. Janowitz, K. Henkel, M. Kot, M.H. Zoellner, Ch. Wenger, J.I. Flege
Journal of Vacuum Science and Technology A 36(6), 062406 (2021)
DOI: 10.1116/6.0001375
Indium oxide (In_xO_y) thin films were deposited by plasma-enhanced atomic layer deposition (PEALD) using trimethylindium and oxygen plasma in a low-temperature range of 80–200 °C. The optical properties, chemical composition, crystallographic structure, and electrical characteristics of these layers were investigated by spectroscopic ellipsometry (SE), x-ray photoelectron spectroscopy (XPS), x-ray diffraction (XRD), as well as current-voltage and capacitance-voltage measurements. The SE results yielded a nearly constant growth rate of 0.56 Å per cycle and a thickness inhomogeneity of ≤1.2% across 4-in. substrates in the temperature range of 100–150 °C. The refractive index (at 632.8 nm) was found to be 2.07 for the films deposited at 150 °C. The PEALD-In_xO_y layers exhibit a direct (3.3 ± 0.2 eV) and an indirect (2.8 ± 0.1 eV) bandgap with an uptrend for both with increasing substrate temperature. Based on XPS characterization, all InxOy samples are free of carbon impurities and show a temperature-dependent off-stoichiometry indicating oxygen vacancies. XRD diffraction patterns demonstrate an onset of crystallization at 150 °C. Consistent with the optical, XPS, and XRD data, the films deposited at ≥150 °C possess higher electrical conductivity. Our findings prove that a low-temperature PEALD process of In_xO_y is feasible and promising for a high-quality thin-film deposition without chemical impurities on thermally fragile substrates.

(20) Towards CMOS Compatible Materials for Surface Enhanced Raman Spectroscopy (SERS)
C.L. Manganelli, H. Matbaechi Ettehad, M. Masood, D. Spirito, Ch. Wenger
Proc. 3^rd European BioSesnsor Symposium (EBS 2021), abstr. (2021)
(exosurf)

(21) Tensile Strained Germanium Microstructures: A Comprehensive Analysis of Thermo-Opto-Mechanical Properties
C.L. Manganelli, M. Virgilio, M. Montanari, I. Zaitsev, N. Anriolli, S. Faralli, S. Tirelli, F. Dagnano, W.M. Klesse, D. Spirito
Physica Status Solidi A 218(21), 2100293 (2021)
DOI: 10.1002/pssa.202100293
We investigate the influence of the thermo-mechanical effects on the optical properties of Germanium microstructures. Finite element method (FEM) calculations allow a complete spatial assessment of mechanical deformations induced by a stressor layer deposited on Ge micro-pillars. Simulated strain maps are in excellent agreement with experimental maps obtained by Raman spectroscopy. The theoretical investigation on strain-dependent band-structure, including the presence of a strain gradient along the longitudinal direction, is exploited to fully capture photoluminescence spectroscopy experiments. Finally, we also quantify the joint effect of temperature and strain on the fundamental band gap.

(22) Phase Transitions in Low-Dimensional Layered Double Perovskites: The Role of the Organic Moieties
B. Martin-Garcia, D. Spirito, G. Biffi, S. Artyukhin, F. Bonaccorso, R. Krahne
The Journal of Physical Chemistry Letters 12(1), 280 (2021)
DOI: 10.1021/acs.jpclett.0c03275
Halide double perovskites are an interesting alternative to Pb-containing counterparts as active materials in optoelectronic devices. Low-dimensional double perovskites are fabricated by introducing large organic cations, resulting in organic/inorganic architectures with one or more inorganic octahedra layers separated by organic cations. Here, we synthesized layered double perovskites based on 3D Cs2AgBiBr6, consisting of double (2L) or single (1L) inorganic octahedra layers, using ammonium cations of different sizes and chemical structures. Temperature-dependent Raman spectroscopy revealed phase transition signatures in both inorganic lattice and organic moieties by detecting variations in their vibrational modes. Changes in the conformational arrangement of the organic cations to an ordered state coincided with a phase transition in the 1L systems with the shortest ammonium moieties. Significant changes of photoluminescence intensity observed around the transition temperature suggest that optical properties may be affected by the octahedral tilts emerging at the phase transition.

(23) Characterization and Manipulation of Yeast Cells using Microfluidic-based Interdigitated Biosensor
H. Matbaechi Ettehad, Ch. Wenger
Proc. 3^rd European BioSensor Symposium (EBS 2021), abstr. (2021)
(BioBic)

(24) Characterization and Separation of Live and Dead Yeast Cells using CMOS-Based DEP Microfluidics
H. Matbaechi Ettehad, Ch. Wenger
Micromachines 12(3), 270 (2021)
DOI: 10.3390/mi12030270, (BioBic)
This study aims at developing a miniaturized CMOS integrated silicon-based microfluidic system, compatible with a standard CMOS process, to enable the characterization, and separation of live and dead yeast cells (as model bio-particle organisms) in a cell mixture using the DEP technique. DEP offers excellent benefits in terms of cost, operational power, and especially easy electrode integration with the CMOS architecture, and requiring label-free sample preparation. This can increase the likeliness of using DEP in practical settings. In this work the DEP force was generated using an interdigitated electrode arrays (IDEs) placed on the bottom of a CMOS-based silicon microfluidic channel. This system was primarily used for the immobilization of yeast cells using DEP. This study validated the system for cell separation applications based on the distinct responses of live and dead cells and their surrounding media. The findings confirmed the device’s capability for efficient, rapid and selective cell separation. The viability of this CMOS embedded microfluidic for dielectrophoretic cell manipulation applications and compatibility of the dielectrophoretic structure with CMOS production line and electronics, enabling its future commercially mass production.

(25) Accurate Program/Verify Schemes of Resistive Switching Memory (RRAM) for In-Memory Neural Network Circuits
V. Milo, A. Glukhov, E. Perez, C. Zambelli, N. Lepri, M.K. Mahadevaiah, E. Perez-Bosch Quesada, P. Olivo, Ch. Wenger, D. Ielmini
IEEE Transactions on Electron Devices 68(8), 3832 (2021)
DOI: 10.1109/TED.2021.3089995, (Neutronics)
Resistive switching memory (RRAM) is a promising technology for embedded memory and their application in computing. In particular, RRAM arrays can provide a convenient primitive for matrix vector multiplication (MVM) with strong impact on the acceleration of neural networks for artificial intelligence (AI). At the same time, RRAM is affected by intrinsic conductance variations which might cause a degradation of accuracy in AI inference hardware. This work provides a detailed study of the multilevel-cell (MLC) programming of RRAM for neural network applications. We compare three MLC programming schemes and discuss their variations in terms of the different slope in the programming characteristics. We test the accuracy of a 2-layer fully-connected neural network (FC-NN) as a function of the MLC scheme, the number of weight levels, and the weight mapping configuration. We find a trade-off between the FC-NN accuracy, size and current consumption. This work highlights the importance of a holistic approach to AI accelerators encompassing the device properties, the overall circuit performance, and the AI application specifications.

(26) Optimized Programming Algorithms for Multilevel RRAM in Hardware Neural Networks
V. Milo, F. Anzalone, C. Zambelli, E. Perez, M.K. Mahadevaiah, O.G. Ossorio, P. Olivo, Ch. Wenger, D. Ielmini
Proc. International Reliability Physics Symposium (IRPS 2021), (2021)
DOI: 10.1109/IRPS46558.2021.9405119, (KI-PRO)

(27) Optimized Programming Algorithms for Multilevel RRAM in Hardware Neural Networks
V. Milo, F. Anzalone, C. Zambelli, E. Perez, M.K. Mahadevaiah, O.G. Ossorio, P. Olivo, Ch. Wenger, D. Ielmini
Proc. International Reliability Physics Symposium (IRPS 2021), (2021)
DOI: 10.1109/IRPS46558.2021.9405119, (Total Resilience)

(28) Optimized Programming Algorithms for Multilevel RRAM in Hardware Neural Networks
V. Milo, F. Anzalone, C. Zambelli, E. Perez, M.K. Mahadevaiah, O.G. Ossorio, P. Olivo, Ch. Wenger, D. Ielmini
Proc. International Reliability Physics Symposium (IRPS 2021), (2021)
DOI: 10.1109/IRPS46558.2021.9405119, (NeuroMem)

(29) Accurate Program/Verify Schemes of Resistive Switching Memory (RRAM) for In-Memory Neural Network Circuits
V. Milo, A. Glukhov, E. Perez, C. Zambelli, N. Lepri, M.K. Mahadevaiah, E. Perez-Bosch Quesada, P. Olivo, Ch. Wenger, D. Ielmini
IEEE Transactions on Electron Devices 68(8), 3832 (2021)
DOI: 10.1109/TED.2021.3089995, (Total Resilience)
Resistive switching memory (RRAM) is a promising technology for embedded memory and their application in computing. In particular, RRAM arrays can provide a convenient primitive for matrix vector multiplication (MVM) with strong impact on the acceleration of neural networks for artificial intelligence (AI). At the same time, RRAM is affected by intrinsic conductance variations which might cause a degradation of accuracy in AI inference hardware. This work provides a detailed study of the multilevel-cell (MLC) programming of RRAM for neural network applications. We compare three MLC programming schemes and discuss their variations in terms of the different slope in the programming characteristics. We test the accuracy of a 2-layer fully-connected neural network (FC-NN) as a function of the MLC scheme, the number of weight levels, and the weight mapping configuration. We find a trade-off between the FC-NN accuracy, size and current consumption. This work highlights the importance of a holistic approach to AI accelerators encompassing the device properties, the overall circuit performance, and the AI application specifications.

(30) A Compact Optical Sensor for Explosive Detection Based on NIR Luminescent Quantum Dots
F. Mitri, A. De Iacovo, S. De Santis, C. Giansante, D. Spirito, G. Sotgiu, L. Colace
Applied Physics Letters 119(4), 041106 (2021)
DOI: 10.1063/5.0060400
Detection of explosive traces in the vapor phase is of primary importance for safety and security in several environments. Different detection methods with high sensitivity are available in the market, but they are typically expensive and require specialized personnel to be operated. Here, we propose a compact, low-cost sensor for explosive detection based on the photoluminescence (PL) quenching of solid-state PbS quantum dot solids cast from the solution phase on a silicon substrate. We demonstrate the sensor capability to detect nitrobenzene vapor at a concentration as low as 445 ppb in air at room temperature, overcoming the performance of other state-of-the-art quantum dot-based PL sensors for nitroaromatic compounds. Moreover, the proposed system can be realized with off-the-shelf electronics and does not need any additional laboratory equipment to be operated, thus paving the way for its deployment in distributed sensor networks.

(31) THz Intersubband Absorption in N-Type Si_1-xGe_x Parabolic Quantum Wells
M. Montanari, C. Ciano, L. Persichetti, C. Corley, L. Baldassarre, M. Ortolani, L. Di Gaspare, G. Capellini, D. Stark, G. Scalari, M. Virgilio, M. De Seta
Applied Physics Letters 118(16), 163106 (2021)
DOI: 10.1063/5.0048344, (FLASH)
High-quality n-type continuously graded Ge-rich Si_1-xGe_x parabolic quantum wells with different doping levels were grown by using ultrahigh-vacuum chemical vapor deposition on Si(001) substrates. A thorough structural characterization study highlights an ideal parabolic compositional profile. THz intersubband absorption has been investigated in modulation-doped samples and samples directly doped in the wells. The comparison of experimental absorption data and theoretical calculations allowed us to quantify the impact of electron correlation effects on the absorption resonances in the different doping conditions and for electron sheet densities in the (1/6) x 10¹¹ cm^-2 range. A single optical resonance is present in modulation doped samples. Its peak energy and line shape are independent of temperature-induced variations of the electron distribution in the subbands up to 300 K, in agreement with the generalized Kohn theorem. This achievement represents a relevant step forward for the development of CMOS compatible optoelectronic devices in the THz spectral range, where thermal charge fluctuations play a key role.

(32) Simulation of Nitrogen Indiffusion and its Impact on Silicon Wafer Strength
T. Müller, G. Kissinger, D. Kot, M. Gehmlich, M. Boy, A. Vollkopf, A. Sattler, A. Miller
Physica Status Solidi A 2100210 (2021)
DOI: 10.1002/pssa.202100210, (Future Silicon Wafers)
In this work, we simulate the concentration profiles of nitrogen after nitriding RTA (rapid thermal annealing) of Si wafer surfaces. We use a model describing the fundamental partial differential equations for diffusion and interaction of nitrogen, vacancies, interstitials, and NV complexes for this thermal process and solve them simultaneously. In addition, thermal stress tests via intentionally induced delta T stress in an RTA process are conducted. By these wafer strength tests, which apply a controlled stress level, the relative slip robustness of wafers with nitrogen in-diffused RTA and polished wafers are compared quantitatively.

(33) Ge(001) Surface Reconstruction with Sn Impurities
K. Noatschk, E.V.S. Hofmann, J. Dabrowski, N.J. Curson, T. Schroeder, W.M. Klesse, G. Seibold
Surface Science 713, 121912 (2021)
DOI: 10.1016/j.susc.2021.121912
Defects play an important role for surface reconstructions and therefore also influence the substrate growth. In this work we present a first principle calculation for the Ge(001) surface without and with tin impurities incorporated into the top surface layer. By mapping the system onto an Ising-type model, with interaction constants taken from density functional theory, the stability of the surface reconstructions under the influence of dierent concentrations of tin impurities is explored. This approach allows us to simulate the possible phase transitions for the dierent surface reconstructions including the local structure around the tin impurity atoms. In addition, we compare our theoretical results with experimental STM images on clean and Sn-doped Ge(100) surfaces.

(34) AC Electric Field Mediated Preparation of Regular Enzyme Arrays and their Functional Characterization
M. Noffke, Ch. Wenger, F.F. Bier, R. Hölzel
Proc. 3^rd European BioSensor Symposium (EBS 2021),abstr.  (2021)
(exosurf)

(35) Performance Assessment of Amorphous HfO₂-based RRAM Devices for Neuromorphic Applications
O.G. Ossorio, G. Vinuesa, H. Garcia, B. Sahelices, S. Dueñas, H. Castan, E. Perez, M.K. Mahadevaiah, Ch. Wenger
ECS Journal of Solid State Science and Technology 10, 083002 (2021)
Proc. 239^th ECS Meeting (2021)
(NeuroMem)

(36) Performance Assessment of Amorphous HfO₂-based RRAM Devices for Neuromorphic Applications
O.G. Ossorio, G. Vinuesa, H. Garcia, B. Sahelices, S. Dueñas, H. Castan, E. Perez, M.K. Mahadevaiah, Ch. Wenger
ECS Journal of Solid State Science and Technology 10, 083002 (2021)
Proc. 239^th ECS Meeting (2021)
(Total Resilience)

(37) Performance Assessment of Amorphous HfO₂-based RRAM Devices for Neuromorphic Applications
O.G. Ossorio, G. Vinuesa, H. Garcia, B. Sahelices, S. Dueñas, H. Castan, E. Perez, M.K. Mahadevaiah, Ch. Wenger
ECS Journal of Solid State Science and Technology 10, 083002 (2021)
Proc. 239^th ECS Meeting (2021)
(KI-PRO)

(38) Performance Assessment of Amorphous HfO₂-Based RRAM Devices for Neuromorphic Applications
O.G. Ossorio, G. Vinuesa, H. Garcia, B. Sahelices, S. Dueñas, H. Castan, E. Perez, M.K. Mahadevaiah, Ch. Wenger
ECS Journal of Solid State Science and Technology 10(8), 083002 (2021)
DOI: 10.1149/2162-8777/ac175c
The use of thin layers of amorphous hafnium oxide has been shown to be suitable for the manufacture of Resistive Random-Access memories (RRAM). These memories are of great interest because of their simple structure and non-volatile character. They are particularly appealing as they are good candidates for substituting flash memories. In this work, the performance of the MIM structure that takes part of a 4 kbit memory array based on 1-transistor-1-resistance (1T1R) cells was studied in terms of control of intermediate states and cycle durability. DC and small signal experiments were carried out in order to fully characterize the devices, which presented excellent multilevel capabilities and resistive-switching behavior.

(39) Performance Assessment of Amorphous HfO₂-based RRAM Devices for Neuromorphic Applications
O.G. Ossorio, G. Vinuesa, H. Garcia, B. Sahelices, S. Duenas, H. Castan, E. Perez, M.K. Mahadevaiah, Ch. Wenger
ECS Transactions 102(2), 29 (2021)
DOI: 10.1149/10202.0029ecst, (Total Resilience)
The use of thin layers of amorphous hafnium oxide has been shown to be suitable for the manufacture of Resistive Random-Access memories (RRAM). These memories are of great interest because of their simple structure and non-volatile character. They are particularly appealing as they are good candidates for substituting flash memories. In this work, the performance of the MIM structure that takes part of a 4 kbit memory array based on 1-transistor-1-resistance (1T1R) cells was studied in terms of control of intermediate states and cycle durability. DC and small signal experiments were carried out in order to fully characterize the devices, which presented excellent multilevel capabilities and resistive-switching behavior.

(40) Performance Assessment of Amorphous HfO₂-based RRAM Devices for Neuromorphic Applications
O.G. Ossorio, G. Vinuesa, H. Garcia, B. Sahelices, S. Duenas, H. Castan, E. Perez, M.K. Mahadevaiah, Ch. Wenger
ECS Transactions 102(2), 29 (2021)
DOI: 10.1149/10202.0029ecst, (KI-PRO)
The use of thin layers of amorphous hafnium oxide has been shown to be suitable for the manufacture of Resistive Random-Access memories (RRAM). These memories are of great interest because of their simple structure and non-volatile character. They are particularly appealing as they are good candidates for substituting flash memories. In this work, the performance of the MIM structure that takes part of a 4 kbit memory array based on 1-transistor-1-resistance (1T1R) cells was studied in terms of control of intermediate states and cycle durability. DC and small signal experiments were carried out in order to fully characterize the devices, which presented excellent multilevel capabilities and resistive-switching behavior.

(41) Performance Assessment of Amorphous HfO₂-based RRAM Devices for Neuromorphic Applications
O.G. Ossorio, G. Vinuesa, H. Garcia, B. Sahelices, S. Duenas, H. Castan, E. Perez, M.K. Mahadevaiah, Ch. Wenger
ECS Transactions 102(2), 29 (2021)
DOI: 10.1149/10202.0029ecst, (NeuroMem)
The use of thin layers of amorphous hafnium oxide has been shown to be suitable for the manufacture of Resistive Random-Access memories (RRAM). These memories are of great interest because of their simple structure and non-volatile character. They are particularly appealing as they are good candidates for substituting flash memories. In this work, the performance of the MIM structure that takes part of a 4 kbit memory array based on 1-transistor-1-resistance (1T1R) cells was studied in terms of control of intermediate states and cycle durability. DC and small signal experiments were carried out in order to fully characterize the devices, which presented excellent multilevel capabilities and resistive-switching behavior.

(42) A Versatile, Voltage-Pulse Based Read and Programming Circuit for Multi-Level RRAM Cells
S. Pechmann, T. Mai, M. Völkel, M.K. Mahadevaiah, E. Perez, E. Perez-Bosch Quesada, M. Reichenbach, Ch. Wenger, A. Hagelauer
Electronics (MDPI) 10(5), 530 (2021)
DOI: 10.3390/electronics10050530, (KI-PRO)
In this work, we present an integrated read and programming circuit for Resistive Random Access Memory (RRAM) cells. Since there are a lot of different RRAM technologies in research and the process variations of this new memory technology often spread over a wide range of electrical properties, the proposed circuit focuses on versatility in order to be adaptable to different cell properties. The circuit is suitable for both read and programming operations based on voltage pulses of flexible length and height. The implemented read method is based on evaluating the voltage drop over a measurement resistor and can distinguish up to eight different states, which are coded in binary, thereby realizing a digitization of the analog memory value. The circuit was fabricated in the 130 nm CMOS process line of IHP. The simulations were done using a physics-based, multi-level RRAM model. The measurement results prove the functionality of the read circuit and the programming system and demonstrate that the read system can distinguish up to eight different states with an overall resistance ratio of 7.9.

(43) Graphene–Silicon Device for Visible and Infrared Photodetection
A. Pelella, A. Grillo, E. Faella, G. Luongo, M.B. Askari, A. Di Bartolomeo
ACS Applied Materials & Interfaces 13(40), 47895 (2021)
DOI: 10.1021/acsami.1c12050, (Graphen)
The fabrication of a graphene–silicon (Gr-Si) junction involves the formation of a parallel metal–insulator–semiconductor (MIS) structure, which is often disregarded but plays an important role in the optoelectronic properties of the device. In this work, the transfer of graphene onto a patterned n-type Si substrate, covered by Si₃N₄, produces a Gr-Si device, in which the parallel MIS consists of a Gr-Si₃N₄-Si structure surrounding the Gr-Si junction. The Gr-Si device exhibits rectifying behavior with a rectification ratio up to 104. The investigation of its temperature behavior is necessary to accurately estimate the Schottky barrier height (SBH) at zero bias, φb0 = 0.24 eV, the effective Richardson’s constant, A* = 7 × 10–10 AK–2 cm^–2, and the diode ideality factor n = 2.66 of the Gr-Si junction. The device is operated as a photodetector in both photocurrent and photovoltage mode in the visible and infrared (IR) spectral regions. A responsivity of up to 350 mA/W and an external quantum efficiency (EQE) of up to 75% are achieved in the 500–1200 nm wavelength range. Decreases in responsivity to 0.4 mA/W and EQE to 0.03% are observed above 1200 nm, which is in the IR region beyond the silicon optical band gap, in which photoexcitation is driven by graphene. Finally, a model based on two parallel and opposite diodes, one for the Gr-Si junction and the other for the Gr-Si₃N₄-Si MIS structure, is proposed to explain the electrical behavior of the Gr-Si device.

(44) Advanced Temperature Dependent Statistical Analysis of Forming Voltage Distributions for Three Different HfO₂-Based RRAM Technologies
E. Perez, D. Maldonado, C. Acal, J.E. Ruiz-Castro, A.M. Aguilera, F. Jimenez-Molinos, J.B. Roldan, Ch. Wenger
Solid State Electronics 176, 107961 (2021)
DOI: 10.1016/j.sse.2021.107961, (NeuroMem)
In this work, voltage distributions of forming operations are analyzed by using an advanced statistical approach based on phase-type distributions (PHD). The experimental data were collected from batches of 128 HfO₂-based RRAM devices integrated in 4-kbit arrays. Three di erent switching oxides, namely, polycrystalline HfO₂, amorphous HfO₂, and Al-doped HfO₂, were tested in the temperature range from -40 to 150 ^oC. The variability of forming voltages has been usually studied by using the Weibull distribution (WD). However, the performance of the PHD analysis demonstrated its ability to better model this crucial operation. The capacity of the PHD to reproduce the experimental data has been validated by means of the Kolmogorov-Smirnov test, while the WD failed in many of the cases studied. In addition, PHD allows to extract information about intermediate probabilistic states that occur in the forming process and the transition probabilities between them; in this manner, we can deepen on the conductive lament formation physics. In particular, the number of intermediate states can be related to the device variability.

(45) Advanced Temperature Dependent Statistical Analysis of Forming Voltage Distributions for Three Different HfO₂-Based RRAM Technologies
E. Perez, D. Maldonado, C. Acal, J.E. Ruiz-Castro, A.M. Aguilera, F. Jimenez-Molinos, J.B. Roldan, Ch. Wenger
Solid State Electronics 176, 107961 (2021)
DOI: 10.1016/j.sse.2021.107961, (Total Resilience)
In this work, voltage distributions of forming operations are analyzed by using an advanced statistical approach based on phase-type distributions (PHD). The experimental data were collected from batches of 128 HfO₂-based RRAM devices integrated in 4-kbit arrays. Three di erent switching oxides, namely, polycrystalline HfO₂, amorphous HfO₂, and Al-doped HfO₂, were tested in the temperature range from -40 to 150 ^oC. The variability of forming voltages has been usually studied by using the Weibull distribution (WD). However, the performance of the PHD analysis demonstrated its ability to better model this crucial operation. The capacity of the PHD to reproduce the experimental data has been validated by means of the Kolmogorov-Smirnov test, while the WD failed in many of the cases studied. In addition, PHD allows to extract information about intermediate probabilistic states that occur in the forming process and the transition probabilities between them; in this manner, we can deepen on the conductive lament formation physics. In particular, the number of intermediate states can be related to the device variability.

(46) Optimization of Multi-Level Operation in RRAM Arrays for In-Memory Computing
E. Perez, A.J. Perez-Avila, R. Romero-Zaliz, M.K. Mahadevaiah, E. Perez-Bosch Quesada, J.B. Roldan, F. Jimenez-Molinos, Ch. Wenger
Electronics (MDPI) 10(9), 1084 (2021)
DOI: 10.1016/j.mee.2019.05.004, (Neutronics)
Accomplishing multi-level programming in resistive random access memory (RRAM) arrays with truly discrete and linearly spaced conductive levels is crucial in order to implement synaptic weights in hardware-based neuromorphic systems. In this paper, we implemented this feature on 4-kbit 1T1R RRAM arrays by tuning the programming parameters of the multi-level incremental step pulse with verify algorithm (M-ISPVA). The optimized set of parameters was assessed by comparing its results with a non-optimized one. The optimized set of parameters proved to be an effective way to define non-overlapped conductive levels due to the strong reduction of the device-to-device variability as well as of the cycle-to-cycle variability, assessed by inter-levels switching tests and during 1k reset-set cycles. In order to evaluate this improvement in real scenarios the experimental characteristics of the RRAM devices were captured by means of a behavioral model, which was used to simulate two different neuromorphic systems: an 8x8 vector-matrix-multiplication (VMM) accelerator and a 4-layer feedforward neural network for MNIST database recognition. The results clearly showed that the optimization of the programming parameters improved both the precision of VMM results as well as the recognition accuracy of the neural network in about 6 % compared with the use of non-optimized parameters.

(47) Optimization of Multi-Level Operation in RRAM Arrays for In-Memory Computing
E. Perez, A.J. Perez-Avila, R. Romero-Zaliz, M.K. Mahadevaiah, E. Perez-Bosch Quesada, J.B. Roldan, F. Jimenez-Molinos, Ch. Wenger
Electronics (MDPI) 10(9), 1084 (2021)
DOI: 10.1016/j.mee.2019.05.004, (KI-PRO)
Accomplishing multi-level programming in resistive random access memory (RRAM) arrays with truly discrete and linearly spaced conductive levels is crucial in order to implement synaptic weights in hardware-based neuromorphic systems. In this paper, we implemented this feature on 4-kbit 1T1R RRAM arrays by tuning the programming parameters of the multi-level incremental step pulse with verify algorithm (M-ISPVA). The optimized set of parameters was assessed by comparing its results with a non-optimized one. The optimized set of parameters proved to be an effective way to define non-overlapped conductive levels due to the strong reduction of the device-to-device variability as well as of the cycle-to-cycle variability, assessed by inter-levels switching tests and during 1k reset-set cycles. In order to evaluate this improvement in real scenarios the experimental characteristics of the RRAM devices were captured by means of a behavioral model, which was used to simulate two different neuromorphic systems: an 8x8 vector-matrix-multiplication (VMM) accelerator and a 4-layer feedforward neural network for MNIST database recognition. The results clearly showed that the optimization of the programming parameters improved both the precision of VMM results as well as the recognition accuracy of the neural network in about 6 % compared with the use of non-optimized parameters.

(48) Optimization of Multi-Level Operation in RRAM Arrays for In-Memory Computing
E. Perez, A.J. Perez-Avila, R. Romero-Zaliz, M.K. Mahadevaiah, E. Perez-Bosch Quesada, J.B. Roldan, F. Jimenez-Molinos, Ch. Wenger
Electronics (MDPI) 10(9), 1084 (2021)
DOI: 10.1016/j.mee.2019.05.004, (Total Resilience)
Accomplishing multi-level programming in resistive random access memory (RRAM) arrays with truly discrete and linearly spaced conductive levels is crucial in order to implement synaptic weights in hardware-based neuromorphic systems. In this paper, we implemented this feature on 4-kbit 1T1R RRAM arrays by tuning the programming parameters of the multi-level incremental step pulse with verify algorithm (M-ISPVA). The optimized set of parameters was assessed by comparing its results with a non-optimized one. The optimized set of parameters proved to be an effective way to define non-overlapped conductive levels due to the strong reduction of the device-to-device variability as well as of the cycle-to-cycle variability, assessed by inter-levels switching tests and during 1k reset-set cycles. In order to evaluate this improvement in real scenarios the experimental characteristics of the RRAM devices were captured by means of a behavioral model, which was used to simulate two different neuromorphic systems: an 8x8 vector-matrix-multiplication (VMM) accelerator and a 4-layer feedforward neural network for MNIST database recognition. The results clearly showed that the optimization of the programming parameters improved both the precision of VMM results as well as the recognition accuracy of the neural network in about 6 % compared with the use of non-optimized parameters.

(49) Multilevel Memristor based Matrix-Vector Multiplication: Influence of the Discretization Method
A.J. Perez-Avila, E. Perez, J.B. Roldan, Ch. Wenger, F. Jimenez-Molinos
Proc. 13^th Spanish Conference on Electron Devices (CDE 2021), 66 (2021)
DOI: 10.1109/CDE52135.2021.9455724, (KI-PRO)

(50) Multilevel Memristor based Matrix-Vector Multiplication: Influence of the Discretization Method
A.J. Perez-Avila, E. Perez, J.B. Roldan, Ch. Wenger, F. Jimenez-Molinos
Proc. 13^th Spanish Conference on Electron Devices (CDE 2021), 66 (2021)
DOI: 10.1109/CDE52135.2021.9455724, (Total Resilience)

(51) Multilevel Memristor based Matrix-Vector Multiplication: Influence of the Discretization Method
A.J. Perez-Avila, E. Perez, J.B. Roldan, Ch. Wenger, F. Jimenez-Molinos
Proc. 13^th Spanish Conference on Electron Devices (CDE 2021), 66 (2021)
DOI: 10.1109/CDE52135.2021.9455724, (NeuroMem)

(52) Toward Reliable Compact Modeling of Multilevel 1T-1R RRAM Devices for Neuromorphic Systems
E. Perez-Bosch Quesada, R. Romero-Zaliz, E. Perez, M.K. Mahadevaiah, J. Reuben, M.A. Schubert, F. Jimenez-Molinos, J.B. Roldan, Ch. Wenger
Electronics (MDPI) 10(6), 645 (2021)
DOI: 10.3390/electronics10060645, (Total Resilience)
In this work, three different RRAM compact models implemented in Verilog-A are analyzed and evaluated in order to reproduce the multilevel approach based on the switching capability of experimental devices. These models are integrated in 1T-1R cells to control their analog behavior by means of the compliance current imposed by the NMOS select transistor. Four different resistance levels are simulated and assessed with experimental verification to account for their multilevel capability. Further, an Artificial Neural Network study is carried out to evaluate in a real scenario the viability of the multilevel approach under study.

(53) Toward Reliable Compact Modeling of Multilevel 1T-1R RRAM Devices for Neuromorphic Systems
E. Perez-Bosch Quesada, R. Romero-Zaliz, E. Perez, M.K. Mahadevaiah, J. Reuben, M.A. Schubert, F. Jimenez-Molinos, J.B. Roldan, Ch. Wenger
Electronics (MDPI) 10(6), 645 (2021)
DOI: 10.3390/electronics10060645, (NeuroMem)
In this work, three different RRAM compact models implemented in Verilog-A are analyzed and evaluated in order to reproduce the multilevel approach based on the switching capability of experimental devices. These models are integrated in 1T-1R cells to control their analog behavior by means of the compliance current imposed by the NMOS select transistor. Four different resistance levels are simulated and assessed with experimental verification to account for their multilevel capability. Further, an Artificial Neural Network study is carried out to evaluate in a real scenario the viability of the multilevel approach under study.

(54) Toward Reliable Compact Modeling of Multilevel 1T-1R RRAM Devices for Neuromorphic Systems
E. Perez-Bosch Quesada, R. Romero-Zaliz, E. Perez, M.K. Mahadevaiah, J. Reuben, M.A. Schubert, F. Jimenez-Molinos, J.B. Roldan, Ch. Wenger
Electronics (MDPI) 10(6), 645 (2021)
DOI: 10.3390/electronics10060645, (KI-PRO)
In this work, three different RRAM compact models implemented in Verilog-A are analyzed and evaluated in order to reproduce the multilevel approach based on the switching capability of experimental devices. These models are integrated in 1T-1R cells to control their analog behavior by means of the compliance current imposed by the NMOS select transistor. Four different resistance levels are simulated and assessed with experimental verification to account for their multilevel capability. Further, an Artificial Neural Network study is carried out to evaluate in a real scenario the viability of the multilevel approach under study.

(55) Memristive-Based In-Memory Computing: From Device to Large-Scale CMOS Integration
E. Perez-Bosch Quesada, E. Perez, M.K Mahadevaiah, Ch. Wenger
Neuromorphic Computing and Engineering 1(2), 024006 (2021)
DOI: 10.1088/2634-4386/ac2cd4
With the rapid emergence of in-memory computing systems based on memristive technology, the integration of such memory devices in large-scale architectures is one of the main aspects to tackle. In this work we present a study of HfO₂-based memristive devices for their integration in large-scale CMOS systems, namely 200 mm wafers. The DC characteristics of single metal-insulator-metal devices are analyzed taking under consideration device-to-device variabilities and switching properties. Furthermore, the distribution of the leakage current levels in the pristine state of the samples are analyzed and correlated to the amount of formingless memristors found among the measured devices. Finally, the obtained results are fitted into a physic-based compact model that enables their integration into larger-scale simulation environments.

(56) Variability and Energy Consumption Tradeoffs in Multilevel Programming of RRAM Arrays
E. Perez, M.K. Mahadevaiah, E. Perez-Bosch Quesada, Ch. Wenger
IEEE Transactions on Electron Devices 68(6), 2693 (2021)
DOI: 10.1109/TED.2021.3072868, (Neutronics)
Achieving a reliable multi-level programming operation in resistive random access memory (RRAM) arrays is still a challenging task. In this work, we assessed the impact of the voltage step value used by the programming algorithm on the device-to-device (DTD) variability of the current distributions of four conductive levels and on the energy consumption featured by programming 4-kbit HfO₂-based RRAM arrays. Two different write-verify algorithms were considered and compared, namely, the incremental gate voltage with verify algorithm (IGVVA) and the incremental step pulse with verify algorithm (ISPVA). By using the IGVVA, a main trade-off has to be taken into account since reducing the voltage step leads to a smaller DTD variability at the cost of a strong increase in the energy consumption. Although the ISPVA can not reduce the DTD variability as much as the IGVVA, its voltage step can be decreased in order to reduce the energy consumption with almost no impact on the DTD variability. Therefore, the final decision on which algorithm to employ should be based on the specific application targeted for the RRAM array.

(57) Variability and Energy Consumption Tradeoffs in Multilevel Programming of RRAM Arrays
E. Perez, M.K. Mahadevaiah, E. Perez-Bosch Quesada, Ch. Wenger
IEEE Transactions on Electron Devices 68(6), 2693 (2021)
DOI: 10.1109/TED.2021.3072868, (Total Resilience)
Achieving a reliable multi-level programming operation in resistive random access memory (RRAM) arrays is still a challenging task. In this work, we assessed the impact of the voltage step value used by the programming algorithm on the device-to-device (DTD) variability of the current distributions of four conductive levels and on the energy consumption featured by programming 4-kbit HfO₂-based RRAM arrays. Two different write-verify algorithms were considered and compared, namely, the incremental gate voltage with verify algorithm (IGVVA) and the incremental step pulse with verify algorithm (ISPVA). By using the IGVVA, a main trade-off has to be taken into account since reducing the voltage step leads to a smaller DTD variability at the cost of a strong increase in the energy consumption. Although the ISPVA can not reduce the DTD variability as much as the IGVVA, its voltage step can be decreased in order to reduce the energy consumption with almost no impact on the DTD variability. Therefore, the final decision on which algorithm to employ should be based on the specific application targeted for the RRAM array.

(58) Sensitivity of HfO₂-based RRAM Cells to Laser Irradiation
D. Petryk, Z. Dyka, E. Perez, I. Kabin, J. Katzer, J. Schäffner, P. Langendörfer
Microprocessors and Microsystems 87, 104376 (2021)
(RESCUE)

(59) Comparative Analysis and Optimization of the SystemC-AMS Analog Simulation Efficiency of Resistive Crossbar Arrays
T. Rizzi, E. Perez-Bosch Quesada, Ch. Wenger, C. Zambelli, D. Bertozzi
Proc. 36^th Conference on Design of Circuits and Integrated Systems (DCIS 2021), 196 (2021)

(60) Study of Quantized Hardware Deep Neural Networks Based on Resistive Switching Devices, Conventional versus Convolutional Approaches
R. Romero-Zaliz, E. Perez, F. Jimenez-Molinos, Ch. Wenger, J.B. Roldan
Electronics (MDPI) 10(3), 346 (2021)
DOI: 10.3390/electronics10030346, (KI-PRO)
A comprehensive analysis of two types of artificial neural networks (ANN) is performed to assess the influence of quantization on the synaptic weights. Conventional multilayer-perceptron (MLP) and convolutional neural networks (CNN) have been considered by changing their features in the training and inference contexts, such as number of levels in the quantization process, the number of hidden layers on the network topology, the number of neurons per hidden layer, the image databases, the number of convolutional layers, etc. A reference technology based on 1T1R structures with bipolar memristors including HfO₂ dielectrics was employed, accounting for different multilevel schemes and the corresponding conductance quantization algorithms. The accuracy of the image recognition processes was studied in depth. This type of studies are essential prior to hardware implementation of neural networks. The obtained results support the use of CNNs for image domains. This is linked to the role played by convolutional layers at extracting image features and reducing the data complexity. In this case, the number of synaptic weights can be reduced in comparison to conventional MLPs.

(61) Study of Quantized Hardware Deep Neural Networks Based on Resistive Switching Devices, Conventional versus Convolutional Approaches
R. Romero-Zaliz, E. Perez, F. Jimenez-Molinos, Ch. Wenger, J.B. Roldan
Electronics (MDPI) 10(3), 346 (2021)
DOI: 10.3390/electronics10030346, (NeuroMem)
A comprehensive analysis of two types of artificial neural networks (ANN) is performed to assess the influence of quantization on the synaptic weights. Conventional multilayer-perceptron (MLP) and convolutional neural networks (CNN) have been considered by changing their features in the training and inference contexts, such as number of levels in the quantization process, the number of hidden layers on the network topology, the number of neurons per hidden layer, the image databases, the number of convolutional layers, etc. A reference technology based on 1T1R structures with bipolar memristors including HfO₂ dielectrics was employed, accounting for different multilevel schemes and the corresponding conductance quantization algorithms. The accuracy of the image recognition processes was studied in depth. This type of studies are essential prior to hardware implementation of neural networks. The obtained results support the use of CNNs for image domains. This is linked to the role played by convolutional layers at extracting image features and reducing the data complexity. In this case, the number of synaptic weights can be reduced in comparison to conventional MLPs.

(62) Study of Quantized Hardware Deep Neural Networks Based on Resistive Switching Devices, Conventional versus Convolutional Approaches
R. Romero-Zaliz, E. Perez, F. Jimenez-Molinos, Ch. Wenger, J.B. Roldan
Electronics (MDPI) 10(3), 346 (2021)
DOI: 10.3390/electronics10030346, (Total Resilience)
A comprehensive analysis of two types of artificial neural networks (ANN) is performed to assess the influence of quantization on the synaptic weights. Conventional multilayer-perceptron (MLP) and convolutional neural networks (CNN) have been considered by changing their features in the training and inference contexts, such as number of levels in the quantization process, the number of hidden layers on the network topology, the number of neurons per hidden layer, the image databases, the number of convolutional layers, etc. A reference technology based on 1T1R structures with bipolar memristors including HfO₂ dielectrics was employed, accounting for different multilevel schemes and the corresponding conductance quantization algorithms. The accuracy of the image recognition processes was studied in depth. This type of studies are essential prior to hardware implementation of neural networks. The obtained results support the use of CNNs for image domains. This is linked to the role played by convolutional layers at extracting image features and reducing the data complexity. In this case, the number of synaptic weights can be reduced in comparison to conventional MLPs.

(63) An Analysis on the Architecture and the Size of Quantized Hardware Neural Networks based on Memristors
R. Romero-Zaliz, A. Cantudo, E. Perez, F. Jimenez-Molinos, Ch. Wenger, J.B. Roldan
Electronics (MDPI) 10(24), 03141 (2021)
DOI: 10.3390/electronics10030346, (KI-PRO)
We have performed different simulation experiments in relation to hardware neural networks (NN) to analyze the role of the number of synapses for different NN architectures in the network accuracy, considering different datasets. A technology that stands upon 4-kbit 1T1R ReRAM arrays, where resistive switching devices based on 𝐻𝑓𝑂₂ dielectrics are employed, is taken as a reference. In our study, fully dense (FdNN) and convolutional neural networks (CNN) were considered, where the NN size in terms of the number of synapses and of hidden layer neurons were varied. CNNs work better when the number of synapses to be used is limited. If quantized synaptic weights are included, we observed that NN accuracy decreases significantly as the number of synapses is reduced; in this respect, a trade-off between the number of synapses and the NN accuracy has to be achieved. Consequently, the CNN architecture must be carefully designed; in particular, it was noticed that different datasets need specific architectures according to their complexity to achieve good results. It was shown that due to the number of variables that can be changed in the optimization of a NN hardware implementation, a specific solution has to be worked in each case in terms of synaptic weight levels, NN architecture, etc.

(64) An Analysis on the Architecture and the Size of Quantized Hardware Neural Networks based on Memristors
R. Romero-Zaliz, A. Cantudo, E. Perez, F. Jimenez-Molinos, Ch. Wenger, J.B. Roldan
Electronics (MDPI) 10(24), 03141 (2021)
DOI: 10.3390/electronics10030346, (Total Resilience)
We have performed different simulation experiments in relation to hardware neural networks (NN) to analyze the role of the number of synapses for different NN architectures in the network accuracy, considering different datasets. A technology that stands upon 4-kbit 1T1R ReRAM arrays, where resistive switching devices based on 𝐻𝑓𝑂₂ dielectrics are employed, is taken as a reference. In our study, fully dense (FdNN) and convolutional neural networks (CNN) were considered, where the NN size in terms of the number of synapses and of hidden layer neurons were varied. CNNs work better when the number of synapses to be used is limited. If quantized synaptic weights are included, we observed that NN accuracy decreases significantly as the number of synapses is reduced; in this respect, a trade-off between the number of synapses and the NN accuracy has to be achieved. Consequently, the CNN architecture must be carefully designed; in particular, it was noticed that different datasets need specific architectures according to their complexity to achieve good results. It was shown that due to the number of variables that can be changed in the optimization of a NN hardware implementation, a specific solution has to be worked in each case in terms of synaptic weight levels, NN architecture, etc.

(65) An Analysis on the Architecture and the Size of Quantized Hardware Neural Networks based on Memristors
R. Romero-Zaliz, A. Cantudo, E. Perez, F. Jimenez-Molinos, Ch. Wenger, J.B. Roldan
Electronics (MDPI) 10(24), 03141 (2021)
DOI: 10.3390/electronics10030346, (NeuroMem)
We have performed different simulation experiments in relation to hardware neural networks (NN) to analyze the role of the number of synapses for different NN architectures in the network accuracy, considering different datasets. A technology that stands upon 4-kbit 1T1R ReRAM arrays, where resistive switching devices based on 𝐻𝑓𝑂₂ dielectrics are employed, is taken as a reference. In our study, fully dense (FdNN) and convolutional neural networks (CNN) were considered, where the NN size in terms of the number of synapses and of hidden layer neurons were varied. CNNs work better when the number of synapses to be used is limited. If quantized synaptic weights are included, we observed that NN accuracy decreases significantly as the number of synapses is reduced; in this respect, a trade-off between the number of synapses and the NN accuracy has to be achieved. Consequently, the CNN architecture must be carefully designed; in particular, it was noticed that different datasets need specific architectures according to their complexity to achieve good results. It was shown that due to the number of variables that can be changed in the optimization of a NN hardware implementation, a specific solution has to be worked in each case in terms of synaptic weight levels, NN architecture, etc.

(66) Influence of Variability on the Performance of HfO₂ Memristor-based Convolutional Neural Networks
R. Romero-Zaliz, E. Perez, F. Jimenez-Molinos, Ch. Wenger, J.B. Roldan
Solid State Electronics 185, 108064 (2021)
DOI: 10.1016/j.sse.2021.108064, (KI-PRO)
A study of convolutional neural networks (CNNs) was performed to analyze the influence of quantization and variability in the network synaptic weights. Different CNNs were considered accounting for the number of convolutional layers, size of the filters in the convolutional layer, number of neurons in the final network layers and different sets of quantization levels. The conductance levels of fabricated 1T1R structures based on HfO₂ memristors were considered as reference for four or eight level quantization processes at the inference stage of the CNNs, which were previous trained with the MNIST dataset. We also included the variability of the experimental conductance levels that was found to be Gaussian distributed and was correspondingly modeled for the synaptic weight implementation.

(67) Influence of Variability on the Performance of HfO₂ Memristor-based Convolutional Neural Networks
R. Romero-Zaliz, E. Perez, F. Jimenez-Molinos, Ch. Wenger, J.B. Roldan
Solid State Electronics 185, 108064 (2021)
DOI: 10.1016/j.sse.2021.108064, (NeuroMem)
A study of convolutional neural networks (CNNs) was performed to analyze the influence of quantization and variability in the network synaptic weights. Different CNNs were considered accounting for the number of convolutional layers, size of the filters in the convolutional layer, number of neurons in the final network layers and different sets of quantization levels. The conductance levels of fabricated 1T1R structures based on HfO₂ memristors were considered as reference for four or eight level quantization processes at the inference stage of the CNNs, which were previous trained with the MNIST dataset. We also included the variability of the experimental conductance levels that was found to be Gaussian distributed and was correspondingly modeled for the synaptic weight implementation.

(68) Influence of Variability on the Performance of HfO₂ Memristor-based Convolutional Neural Networks
R. Romero-Zaliz, E. Perez, F. Jimenez-Molinos, Ch. Wenger, J.B. Roldan
Solid State Electronics 185, 108064 (2021)
DOI: 10.1016/j.sse.2021.108064, (Total Resilience)
A study of convolutional neural networks (CNNs) was performed to analyze the influence of quantization and variability in the network synaptic weights. Different CNNs were considered accounting for the number of convolutional layers, size of the filters in the convolutional layer, number of neurons in the final network layers and different sets of quantization levels. The conductance levels of fabricated 1T1R structures based on HfO₂ memristors were considered as reference for four or eight level quantization processes at the inference stage of the CNNs, which were previous trained with the MNIST dataset. We also included the variability of the experimental conductance levels that was found to be Gaussian distributed and was correspondingly modeled for the synaptic weight implementation.

(69) Raman Shifts in MBE-Grown Si_xGe_1-x-ySn_y Alloys with Large Si Content
J. Schlipf, H. Tetzner, D. Spirito, C.L. Manganelli, G. Capellini, M.R.S. Huang, C.T. Koch, C.J. Clausen, A. Elsayed, M. Oehme, S. Chiussi, J. Schulze, I.A. Fischer
Journal of Raman Spectroscopy 52(6), 1167 (2021)
DOI: 10.1002/jrs.6098, (Dfg-QWIP)
We examine the Raman shift in Silicon-Germanium-Tin alloys with high Silicon content grown on a Germanium virtual substrate by molecular beam epitaxy. The Raman shifts of the three most prominent modes, Si-Si, Si-Ge and Ge-Ge, are measured and compared with results in previous literature. We analyze and fit the dependence of the three modes on the composition and strain of the semiconductor alloys. We also demonstrate the calculation of the composition and strain of Si_xGe_1-x-ySn_y from the Raman shifts alone, based on the fitted relationships. Our analysis extends previous results to samples lattice-matched on Ge and with higher Si content than in prior comprehensive Raman analyses, thus making Raman measurements as a fast and non-destructive characterization technique accessible for a wider compositional range of these ternary alloys.

(70) Modelling Photodetection at the Graphene/Ag₂S Interface
D. Spirito, B. Martín-García, V. Miseikis, C. Coletti, F. Bonaccorso, R. Krahne
Physica Status Solidi - Rapid Research Letters 15(6), 2100120 (2021)
DOI: 10.1002/pssr.202100120
Mixed‐dimensional systems host interesting phenomena that involve electron and ion transport along or across the interface, with promising applications in optoelectronic and electrochemical devices. Herein, a heterosystem consisting of a graphene monolayer with a colloidal Ag2S nanocrystal film atop, in which both ions and electrons are involved in photoelectrical effects, is studied. An investigation of the transport at the interface in different configurations by using a phototransistor configuration with graphene as a charge‐transport layer and semiconductor nanocrystals as a light‐sensitive layer is performed. The key feature of charge transfer is investigated as a function of gate voltage, frequency, and incident light power. A simple analytical model of the photoresponse is developed, to gain information on the device operation, revealing that the nanocrystals transfer electrons to graphene in the dark, but the opposite process occurs upon illumination. A frequency‐dependence analysis suggests a fractal interface between the two materials. This interface can be modified using solid‐state electrochemical reactions, leading to the formation of metallic Ag particles, which affect the graphene properties by additional doping, while keeping the photoresponse. Overall, these results provide analytical tools and guidelines for the evaluation of coupled electron/ion transport in hybrid systems.

(71) Thermoelectric Efficiency of Epitaxial GeSn Alloys for Integrated Si-based Applications: Assessing the Lattice Thermal Conductivity by Raman Thermometry
D. Spirito, N. von den Driesch, C.L. Manganelli, M.H. Zoellner, A.A. Corley-Wiciak, Z. Ikonic, T. Stoica, D. Grützmacher, D. Buca, G. Capellini
ACS Applied Energy Materials 4(7), 7385 (2021)
DOI: 10.1021/acsaem.1c01576
Energy harvesting for Internet of Things applications, comprising sensing, life sciences, wearables, and communications, requires efficient thermoelectric (TE) materials, ideally semiconductors compatible with Si technology. In this work, we investigate the potential of GeSn/Ge layers, a group IV material system, as TE material for low-grade heat conversion. We extract the lattice thermal conductivity, by developing an analytical model based on Raman thermometry and heat transport model, and use it to predict thermoelectric performances. The lattice thermal conductivity decreases from 56 W/(m·K) for Ge to 4 W/(m·K) by increasing the Sn atomic composition to 14%. The bulk cubic Ge_0.86Sn_0.14 alloy features a TE figure of merit of ZT ∼ 0.4 at 300 K and an impressive 1.04 at 600 K. These values are extremely promising in view of the use of GeSn/Ge layers operating in the typical on-chip temperature range.

(72) AC Field Assisted Deposition of Influenza Viruses on Nanoelectrodes
S. Stanke, Ch. Wenger, F.F. Bier, R. Hölzel
Proc. 3^rd European BioSensor Symposium (EBS 2021), abstr. (2021)
(exosurf)

(73) THz Intersubband Electroluminescence from N-Type Ge/SiGe Quantum Cascade Structures
D. Stark, M. Mirza, L. Persichetti, M. Montanari, S. Markmann, M. Beck, T. Grange, S. Birner, M. Virgilio, C. Ciano, M. Ortolani, C. Corley, G. Capellini, L. Di Gaspare, M. De Seta, D.J. Paul, J. Faist, G. Scalari
Applied Physics Letters 118(10), 101101 (2021)
DOI: 10.1063/5.0041327, (FLASH)
We report electroluminescence originating from L-valley transitions in n-type Ge/Si_0.15Ge_0.85 quantum cascade structures centered at 3.4 and 4.9 THz with a line broadening of Δf/f≈0.2 . Three strain-compensated heterostructures, grown on a Si substrate by ultrahigh vacuum chemical vapor deposition, have been investigated. The design is based on a single quantum well active region employing a vertical optical transition, and the observed spectral features are well described by non-equilibrium Green's function calculations. The presence of two peaks highlights a suboptimal injection in the upper state of the radiative transition. Comparison of the electroluminescence spectra with a similar GaAs/AlGaAs structure yields one order of magnitude lower emission efficiency.

(74) Terahertz Intersubband Electroluminescence from N-Type Germanium Quantum Wells
D. Stark, M. Mirza, L. Persichetti, M. Montanari, S. Markmann, M. Beck, T. Grange, S. Birner, M. Virgilio, C. Ciano, M. Ortolani, C. Corley-Wiciak, G. Capellini, L. Di Gaspare, M. De Seta, D.J. Paul, J. Faist, G. Scalari
Proc. Conference on Lasers & Electro-Optics / Europe and the European Quantum Electronics Conference (CLEO®/Europe-EQEC 2021), (2021)
DOI: 10.1109/CLEO/Europe-EQEC52157.2021.9541838, (FLASH)

(75) CMOS-Compatible Bias-Tunable Dual-Band Detector based on GeSn/Ge/Si Coupled Photodiodes
E. Talamas Simola, V. Kiyek, A. Ballabio, V. Schlykow, J. Frigerio, C. Zucchetti, A. De Iacovo, L. Colace, Y. Yamamoto, G. Capellini, D. Grützmacher, D. Buca, G. Isella
ACS Photonics 8(7), 2166 (2021)
DOI: 10.1021/acsphotonics.1c0061, (SiGeSn NanoFETs)
Infrared (IR) multispectral detection is attracting an increasing interest with the rising demand for high spectral sensitivity, room temperature operation, CMOS compatible devices. Here, we present a two-terminal dual-band detector, which provides a bias-switchable spectral response in two distinct IR bands. The device is obtained from a vertical GeSn/Ge/Si stack, forming a double junction n-i-p-i-n structure, epitaxially grown on a Si wafer. The photoresponse can be switched, by inverting the bias polarity, between the near and the short-wave IR bands, with specific detectivities of 1.9×10¹⁰ and 4.0×10⁹ cm·(Hz)^1/2/W, respectively. The possibility of detecting two spectral bands with the same pixel opens up interesting applications in the field of IR imaging and material recognition, as shown in a solvent detection test. The continuous voltage tuning, combined with the non-linear photoresponse of the detector, enables a novel approach to spectral analysis, demonstrated by identifying the wavelength of a monochromatic beam.

(76) Current Leakage Mechanisms Related to Threading Dislocations in Ge-Rich SiGe Heterostructures Grown on Si(001)
H. Tetzner, I.A. Fischer, O. Skibitzki, M.M. Mirza, C.L. Manganelli, G. Luongo, D. Spirito, D.J. Paul, M. De Seta, G. Capellini
Applied Physics Letters 119(15), 153504 (2021)
DOI: 10.1063/5.0064477, (FLASH)
This work investigates the role of threading dislocation densities (TDD) in the low density regime on the vertical transport in Si_0.06Ge_0.94 heterostructures integrated on Si(001). The use of unintentionally doped Si_0.06Ge_0.94 layers enables the study of the impact of grown-in threading dislocations (TD) without interaction with processing-induced defects originating, e.g., from dopant implantation. The studied heterolayers, while equal in composition, the degree of strain relaxation, and the thickness feature three different values for the TDD as 3x10⁶, 9x10⁶, and 2x10⁷ cm^-2 . Current–voltage measurements reveal that leakage currents do not scale linearly with TDD. The temperature dependence of the leakage currents suggests a strong contribution of field-enhanced carrier generation to the current transport with the trapassisted tunneling via TD-induced defect states identified as the dominant transport mechanism at room temperature. At lower temperatures and at high electric fields, direct band-to-band tunneling without direct interactions with defect levels becomes the dominating type of transport. Leakage currents related to emission from mid-gap traps by the Shockley–Read–Hall (SRH) generation are observed at higher temperatures (>100°C). Here, we see a reduced contribution coming from SRH in our material, featuring the minimal TDD (3x10⁶ cm^-2 ), which we attribute to a reduction in point defect clusters trapped in the TD strain fields.

(77) Behavioral Model of Dot-Product Engine Implemented with 1T1R Memristor Crossbar Including Assessment
J. Wen, M. Ulbricht, E. Perez, X. Fan, M. Krstic
Proc. 24^th International Symposium on Design and Diagnostics of Electronic Circuits and Systems (DDECS 2021), 29 (2021)
DOI: 10.1109/DDECS52668.2021.9417070, (KI-PRO)

(78) Threading Dislocation Reduction of Ge by Introducing a SiGe/Ge Superlattice
Y. Yamamoto, C. Corley, M.A. Schubert, M.H. Zoellner, B. Tillack
ECS Journal of Solid State Science and Technology 10(3), 034005 (2021)
DOI: 10.1149/2162-8777/abea5e
The influence of introducing a SiGe/Ge superlattice (SL) between Ge layers and Si substrate for the sake of the reduction of the threading dislocation density (TDD) without additional annealing is investigated. By introducing the SiGe/Ge SL and optimizing the layer stack, the TDD of the Ge layer becomes ~1/3. In the case of 2.8 μm thick Ge without introducing the SiGe/Ge SL, the TDD at the surface is 7.6 × 10⁸ cm⁻². A slight TDD reduction is observed by introducing a Si_0.2Ge_0.8/Ge SL between the Si substrate and the Ge layer. By inserting 5, 10 and 20 cycles of Si_0.2Ge_0.8/Ge, the TDD is reduced to 7.1 × 10⁸, 5.9 × 10⁸ and 5.3 × 10⁸ cm⁻², respectively. The lateral lattice parameters of these SLs are ~5.656 Å, which is a smaller value compared to that of bulk Ge, indicating plastic relaxation by misfit dislocation formation. Further TDD reduction is realized with increasing Si concentration in the SiGe/Ge SL without changing the cycle of the SL. However, surface roughening due to pit formation occurs if the Si concentration in the SL is higher than 50% because of increased strain at the interfaces between SiGe and Ge. With increasing SiGe and Ge thickness ratio in the SL layer and maintaining periodicity and cycles, the TDD is reduced to 2.8 × 10⁸ cm⁻² without degrading the surface roughness. This improvement is related to a relaxation of the SiGe/Ge SL by plastic deformation.

(79) A Nonlinear Resistive Switching Behaviors of Ni/HfO₂/TiN Memory Structures for Self-rectifying Resistive Switching Memory
M.J. Yun, D. Lee, S. Kim, Ch. Wenger, H.-D. Kim
Materials Characterization 182, 111578 (2021)
This work reports forming free/self-rectifying resistive switching characteristics and dependency of the top electrode (TE) of a crystalline HfO₂-based resistive switching memory device. In the memory cells, nonlinear bipolar resistive switching characteristics, i.e., an asymmetric current-voltage curve like the Schottky diode, was observed. In addition, the device exhibits resistive switching behaviors without forming process, which makes it possible to switch the resistance state under ultra-low current levels of <10 nA. In addition, compared to the resistive switching of the proposed resistive switching memory devices with different TEs, the V_SET was decreased when using TE with lower work function, and the height read margin was obtained in the sample with the Ni TE, covering over 56 × 56 arrays. Consequently, these results indicate that the interface control resistive switching properties in memory structures having the Schottky junction warrant the realization of selector-free resistive switching memory cells in a highdensity crossbar array.

(80) Vibration Analysis of a Wind Turbine Gearbox for Off-Cloud Health Monitoring through Neuromorphic-Computing
P.S. Zarrin, C. Martin, P. Langendörfer, Ch. Wenger, M. Diaz
Proc. 42^th International Conference on Information Systems (ICIS 2021), (2021)
DOI: 978-0-9966831-5-9, (RRAM (Resistive RAM))

(81) Vibration Analysis of a Wind Turbine Gearbox for Off-Cloud Health Monitoring through Neuromorphic-Computing
P.S. Zarrin, C. Martin, P. Langendörfer, Ch. Wenger, M. Diaz
Proc. 42^th International Conference on Information Systems (ICIS 2021), (2021)
DOI: 978-0-9966831-5-9, (NeuroMem)

(82) Vibration Analysis of a Wind Turbine Gearbox for Off-Cloud Health Monitoring through Neuromorphic-Computing
P.S. Zarrin, C. Martin, P. Langendörfer, Ch. Wenger, M. Diaz
Proc. 47^th Annual Conference of the IEEE Industrial Electronics Society (IECON 2021), (2021)
DOI: 10.1109/IECON48115.2021.9589879, (Total Resilience)

(83) Vibration Analysis of a Wind Turbine Gearbox for Off-Cloud Health Monitoring through Neuromorphic-Computing
P.S. Zarrin, C. Martin, P. Langendörfer, Ch. Wenger, M. Diaz
Proc. 47^th Annual Conference of the IEEE Industrial Electronics Society (IECON 2021), (2021)
DOI: 10.1109/IECON48115.2021.9589879, (RRAM (Resistive RAM))

(84) Vibration Analysis of a Wind Turbine Gearbox for Off-Cloud Health Monitoring through Neuromorphic-Computing
P.S. Zarrin, C. Martin, P. Langendörfer, Ch. Wenger, M. Diaz
Proc. 47^th Annual Conference of the IEEE Industrial Electronics Society (IECON 2021), (2021)
DOI: 10.1109/IECON48115.2021.9589879, (NeuroMem)

(85) Vibration Analysis of a Wind Turbine Gearbox for Off-Cloud Health Monitoring through Neuromorphic-Computing
P.S. Zarrin, C. Martin, P. Langendörfer, Ch. Wenger, M. Diaz
Proc. 42^th International Conference on Information Systems (ICIS 2021), (2021)
DOI: 978-0-9966831-5-9, (Total Resilience)