preview all symposia

2022 Fall Meeting

Functional materials

R

Integration of advanced materials on silicon: from classical to neuromorphic and quantum applications

The symposium aims at gathering scientists working on monolithic and heterogeneous integrations of new materials, to enable additional functionalities on silicon-based platforms. Its originality lies in the fact that it considers both classical approaches and emerging topics linked to neuromorphic and quantum applications. The various research fields covered in the symposium pave the way towards highly functionalized Si-based technologies which address current and future challenges in our society.

Scope:

The microelectronics industry has delivered faster and more efficient computing devices at a remarkably consistent pace for several decades. This has mostly been achieved by downscaling classical MOS transistors, which continuously provided improved performance and lower energy consumption for every new technology node. These advances have led to the development of highly performing personal computers and low-power mobile devices, which are nowadays affordable for mass population.

More recently, the demand for high performance devices and mass data transfer has soared, driven by new societal needs linked to the “Internet-of-things” and the growing demand for ultra-fast data communication and data transfer, cognitive systems and new computing paradigms, such as neuromorphic and quantum information processing. However, transistors cannot scale down indefinitely. Industrials are therefore looking beyond classic architectures and concepts to secure future generations of devices. Still, the best contenders are likely to be those that can be integrated with conventional silicon chip platform.

Neuromorphic networks for example require dense arrays of interconnected devices, patterned on silicon using the processing know-how generated by the conventional industry. For quantum information science, silicon is also emerging as a promising route. Elementary silicon qubit devices have been demonstrated with high-fidelity operation, highlighting the potential of silicon-based quantum devices in terms of scalability and manufacturability. Programmable quantum circuits based on silicon photonics chips are currently extensively investigated. Even for emerging materials that are not yet widely used in the industry, like topological insulators, quantum-dots structures, magnetic or superconductor materials, silicon could be a platform of choice for device integration.

The symposium aims at highlighting novel and innovative approaches that allow for monolithic and heterogeneous integration on silicon technology, targeting CMOS, application-specific integrated solutions (based on integrated photonics, neural networks, spintronic devices...) or quantum systems.

The scope includes the fundamental understanding of new material properties, the implementation of novel integration schemes, the modeling techniques and new application fields. The focus will be on the fabrication, characterization and simulation of materials considered as non-standard for Si technology. Contributions related to innovative hetero-integration techniques will be encouraged. Finally, a particular attention will be given to devices and applications beyond current computation technologies that aim at addressing new computing paradigms such as quantum and neuromorphic computation. The productive interaction across disciplines will help materials scientists drive the exciting transition towards higher-value, highly functionalized Si-based microelectronics.

Hot topics to be covered by the symposium:

Material growth, characterization and simulation:

  • Group IV and compound semiconductors:
    Group IV materials and alloys (SiGe, GeSn SiGeSn), III-V and II-VI compound semiconductors, grown or transferred on monocrystalline substrates or insulators. Group IV and III-V quantum dots and nanowires integrated on Si.
  • Oxides and nitrides:
    Functional perovskites, ZnO, GaN and heterostructures, oxides with resistive or metal insulator transition, topological insulators, piezoelectric materials, materials for the implementation of neuromorphic devices.
  • 2 dimensional materials:
    Growth and transfer of Graphene, Transition Metal Dichalcogenides and Boron Nitride on semiconductors, hybrid 2D/semiconductor devices.
  • Novel materials for Quantum applications
    Semiconductor/Superconductor Interfaces, Topological insulators, Semiconductor Quantum Dot qubit Materials, purified 28Si, Spin qubit, Si/SiGe Heterostructures

Integration techniques:

  • Advanced heteroepitaxy:
    Selective growth on patterned substrates, epitaxial lateral overgrowth, self-assembly techniques, remote epitaxy.
  • Layer Transfer:
    2.5D & 3D integration (monolithic & heterogeneous)
    Innovative synthesis & integration methods of materials and devices used for quantum systems

Applications:

  • Data processing and communication:
    Advanced CMOS scaling, single electron & single photon devices, neuromorphic architectures, IOT, spintronics, ultra-low power & RF electronics, Integrated photonics, IR and THz lasers.
  • Neuromorphic systems:
    Bioinspired nano electronics or photonics, neural networks on chips, with possible use in artificial intelligence and machine learning.
  • Quantum information science and emerging applications of quantum materials:
    Quantum communication, quantum computing, quantum sensing.
  • Life-Sciences application and environmental sensors:
    Semiconductor plasmonics, mid-infrared and THz sensing, gas sensors, integration with piezo-materials for MEMS-like sensors and opto-mechanics.
Start atSubject View AllNum.
 
III-V Materials : Cezar Zota
14:00
Authors : G Patriarche
Affiliations : Université Paris-Saclay, CNRS, Centre de Nanosciences et de Nanotechnologies, 91120, Palaiseau, France.

Resume : Molecular beams of Ga and As4 are implemented in an aberration-corrected transmission electron microscope. GaAs nanowires are grown in situ from Au catalyst particles. Real-time observation gives access to their morphological and structural parameters while growing and the formation of atomic planes at the catalyst-nanowire interface can be examined. We use various conditions which can result in solid or liquid catalyst particle. For liquid catalyst, the contact angle of the droplet evolves rapidly with the V/III vapor flux ratio. At contact angles around 120°, the atomic plane stacking switches from hexagonal to cubic, with a transition region of mixed crystal phases. In agreement with recently reported results, but using a different growth technique and higher growth rates, we observe that the formation mechanisms of the two crystal phases differ singularly. Namely, hexagonal monolayers grow by slow and continuous step flow on a flat nanowire top facet ; cubic monolayers appear incrementally and concomitantly with a truncation of the nanowire top facet. By changing the wetting angle of the catalyst, and completely stopping the growth between two different crystal structures, we were able to realize a crystal phase superlattice GaAs heterostructure controlled at the monolayer between the cubic (zinc blende) and the hexagonal (wurzite) phase structures.

R.1.1
14:30
Authors : Katarzyna E. Hnida-Gut (1,2), Marilyne Sousa (2), Preksha Tiwari (2,3), Heinz Schmid (2)
Affiliations : (1) IHP Leibniz Institute for High Performance Microelektronics – Frankfurt (Oder), Germany; (2) IBM Research Europe – Zurich, Zurich, Switzerland; (3) Polariton Technologies, Zurich, Switzerland

Resume : High-performance (opto-) electronics would greatly benefit from a versatile III-V co-integration process on silicon-based substrates. However, it requires to overcome technological challenges such as the large lattice mismatch between the III-V and the silicon, their different thermal expansion and polarity. Indeed, those would lead to a large density of defects, which would be detrimental to their optimal operation. To surpass this limitation, the so-called “Template-Assisted Selective Epitaxy” (TASE) technique was developed [1]: by directing the growth from an oxide template and growing III-V in pre-defined shapes using metal-organic chemical vapor deposition (MOCVD), the successful integration of materials like InGaAs, InP or GaSb among others was achieved. However, the MOCVD growth present few limitations: it requires the use of toxic precursors, has low growth rates and struggles to fill high aspect ratio structures. Moreover, it is not suitable to every III-V semiconductor. Thus, it is not adapted to Sb-based layers due to the small process window to suppress III-element droplet formation. In this work we address difficult In-based semiconductors integration on silicon by introducing an alternative electrochemical approach [2]. Traditionally, electrodeposition (ED) is associated with creation of metal coatings on conductive surfaces including protecting/functional layers of metals, their alloys, colloids, conductive polymers, magnetic materials, oxides, and nanostructures of precisely controlled composition, dimensions, and ordering on the coated surface. As the field develops more and more, effort is put into increasing the applicability of electrodeposition in areas where physical deposition techniques predominate, i.e. microelectronics, and energy conversion devices. Here, we show that prefabricated hollow oxide template structures of micro-and sub-micron dimensions, each containing a local embedded electrode can be successfully filled with electrodeposited In. Pre-defined structures for V-element saturation and formation of crystalline III-V semiconductor on Si could thus be achieved. The proposed integration path combines the advantages of fast, uniform templates filling, which are characteristic for wet techniques and with the high-quality semiconductor crystal formation, which are characteristic for vapor phase growth methods. Acknowledgement The research was carried out thanks to the financial support of the Marie Skłodowska-Curie Action H2020 EU TECNO (894326). References: 1. Schmid, H., Borg, M., Moselund, K., Gignac, L., Breslin, C. M., Bruley, J., et al. (2015). Template-assisted Selective Epitaxy of III-V Nanoscale Devices for Coplanar Heterogeneous Integration with Si. Appl. Phys. Lett. 106, 233101, doi:10.1063/1.4921962 2. Hnida-Gut, K.E., Sousa, M., Hopstaken, M., Reidt, S., Moselund, K., Schmid, H. (2022). Electrodeposition as an Alternative Approach for Monolithic Integration of InSb on Silicon. Front. Chem. 9:810256, doi: 10.3389/fchem.2021.810256

R.1.2
15:00
Authors : Tobias Schreitmüller, Patrick Jong, Daniel Ruhstorfer, Akhil Ajay, Andreas Thurn, Jonathan J. Finley, and Gregor Koblmüller
Affiliations : Walter Schottky Institute, Technical University of Munich, 85748 Garching, Germany

Resume : The ability to integrate III-V semiconductor nanowires (NW) on the silicon (Si) platform opens many perspectives for advanced nanoelectronic and optoelectronic device applications on-chip. However, for energy-efficient device performance, as in III-V NW-solar cells or light emitting diodes (LEDs), the design of axial or radial heterostructures, the control of accurate doping properties and the formation of low-resistance ohmic contacts are crucial. In this contribution, we present ongoing developments of radial n-i-p core-multishell NW heterostructures monolithically integrated on the n-Si (111) platform for potential 3D-nanowire light emitting devices in the near-infrared spectral range. The NW structure is designed to host n-type doped GaAs(Sb) cores to take advantage of the higher electron (than hole) mobility in the system for efficient carrier injection via the Si substrate, while the shell is composed of either GaAs-homojunctions or (In,Al)GaAs(Sb)-based heterojunctions that define intrinsic and p-type doped regions [1]. Hereby, one of the most challenging aspects is to realize n-type conduction with controllable charge carrier density in the Si-doped GaAs(Sb) NW cores due to the amphoteric behavior of Si dopants [2]. We show that indeed n-type behavior is feasible by pioneering a novel catalyst-free, vapor-solid growth process of Si-doped GaAs NWs using molecular beam epitaxy (MBE) [3]. Through prototype device structures we probe the access resistance to the n-type NW-core using array and single-NW I-V characterization. The results show an increasing current density for larger contact area at the NW/Si interface, indicating a low access resistance for more efficient carrier injection. The n-doped NW cores were then implemented into radial n-i-p NW homo-junction devices to establish electrical contact formation and perform first electroluminescence (EL) experiments. The EL measurements illustrate successful diode characteristics, with luminescence features that are typical for the underlying material properties. Furthermore, we show the influence of various passivation methods on the electrical properties of an outer p-GaAs shell and an optimization of the corresponding contact resistance. [1] K. Tomioka, et al., Nano Lett. 10, 1639 (2010) [2] H. Hijazi et al., Nano Lett. 19, 4498 (2019) [3] D. Ruhstorfer et al., Appl. Phys. Lett. 116, 52101 (2020)

R.1.3
15:15
Authors : Chen Wei, Laurent Travers, Julien Chaste, Cléophanie Brochard, Andrea Cattoni, David Bouville, Etienne Herth, Jean-Christophe Harmand, Federico Panciera
Affiliations : Université-Paris-Saclay, CNRS, Centre de Nanosciences et de Nanotechnologies, 91120, Palaiseau, France.

Resume : In order to take advantage of the unique physical properties of NWs, it is crucial to control their geometry, crystal structure, and doping level. This goal will ultimately be achieved by a deep understanding of the growth mechanisms. A layer-by-layer mechanism has been recognized during the later stable growth of NWs. Nevertheless, the first stages of the growth process are the least understood and have been investigated almost exclusively by various ex-situ techniques. Here, we present real-time observations of the nucleation and growth of self-catalyzed GaAs nanowires using a transmission electron microscope (TEM) equipped with molecular-beam-epitaxy (MBE) sources. Custom-made substrates, in the form of suspended electron-transparent <111>-oriented Si membranes, were designed and fabricated using MEMS technology. A combination of finite-element simulations and Raman spectroscopy was used to calibrate the sample temperature and optimize its design. Nanowires were grown directly on the membrane substrate inside the microscope via vapor-liquid-solid(VLS), and the process was monitored in situ and in real-time with high spatial and temporal resolution. Based on our direct observations, we will discuss the main steps of nanowire growth, from Ga droplet deposition to crystal nucleation and growth. Additionally, regarding the beginning of epitaxial growth, the interface condition between GaAs NWs/Si substrate and potential defects triggered by lattice mismatch will be revealed by the moiré pattern.

R.1.4
15:30 Coffee break    
 
Materials and devices for quantum computing : Abderraouf Boucherif
16:00
Authors : Karim Cherkaoui 1, Luca Larcher 2, Andrea Padovani 2, Enrico Caruso 3, Erik Lind 4, Lars-Erik Wernersson 4, Johannes Svensson 4, Jun Lin 1, Scott Monaghan 1 and Paul K Hurley 1
Affiliations : 1 Tyndall National Institute, University College Cork, Ireland; 2 Applied Materials Italia Srl, Via Ruini, 74/L 42122 Reggio Emilia (RE) Italy; 3 Infineon Technologies, Villach, Austria; 4 Lund University, Lund, Sweden

Resume : Quantum logic gates can be formed by performing operations on coupled spin states of single electrons with are in neighbouring quantum dots (QDs). The use of silicon is of particular interest, as the planar Si MOS technology is a very well-developed process over the last 50 years [1]. There is a growing appreciation that in such systems the limiting factor for the spin coherence lifetime is determined by interaction with defects at the Si/SiO2 interface, in the oxide, or at the gate/SiO2 interface [2]. In addition, practical systems will require the integration of the readout and control electronics close to the quantum bit (Qubit) stage. In this case, the use of high frequency III-V RF devices in conjunction with silicon CMOS through 3D monolithic integration provides a promising route to achieve high performance with high integration density [3]. It is clear that the ability to characterise, understand and control the interaction of electrons with electrically active defects states in MOS structures will be central to the development of a practical quantum computer based on charge based qubits. In this contribution, we will focus on the case of an InGaAs MOS structure, characterised using multi-frequency capacitance voltage (CV) and conductance voltage (GV), showing how analysis of the capacitance and conductance variations with frequency can differentiate between defects at the oxide/semiconductor interface and defects distributed through the insulating oxide.

R.2.1
16:30
Authors : Philippe Ferrandis (1), Thomas Bédécarrats (2), Mikael Cassé (2)
Affiliations : (1) Université de Toulon, Univ. Grenoble Alpes, CNRS, Institut Néel, 38000 Grenoble, France (2) CEA, LETI, Univ. Grenoble Alpes, 38000 Grenoble, France

Resume : Among promising materials for the realization of future quantum processors, silicon has several assets. A silicon quantum bit (qubit) device can be performed with an industry-standard fabrication process and allows a co-integration with classical control hardware [1]. However, an excellent control of carrier transfer in the channel of the transistor is required, ruling out any electrical active defects, which could act as recombination centers. Such an optimization of the device necessitates a fine characterization of the channel to evaluate the purity of the material. In this work, we investigate the presence of defects in the channel of a fully depleted silicon-on-insulator (FDSOI) transistor designed for qubit applications. A buried oxide of 145 nm formed on a silicon substrate is covered by an 11 nm thick unintentionally doped silicon channel. A SiO2/HfSiON bilayer and a TiN metallic electrode compose the gate. P-doped source and drain contact regions were realized by implantation. A Si3N4 layer is used to passivate uncovered silicon areas. To enlarge the capacitance of the gate and allow measurements, 111 cells were connected in parallel. The ON state of the transistor is achieved by applying a negative voltage to the gate. Source and drain electrodes were connected together and voltage pulses were applied to the gate contact to perform capacitance deep level transient spectroscopy (DLTS) measurements between 77 K and 350 K. A depleted section of the channel extends beneath the gate and towards the source and drain regions when applying a positive voltage to the gate contact. By tuning a positive reverse bias and negative voltage pulses, we were able to probe the channel and localize the electrical active defects responsible for the DLTS signal. Three hole traps, labelled H1, H2 and H3, were detected by DLTS at respectively 0.54 eV, 0.57 eV and 0.65 eV above the valence band edge. The apparent capture cross sections of these levels are respectively 7.9x10-16 cm2, 2.4x10-14 cm2 and 3.3x10-12 cm2. H1 and H2 are donor-like states whereas H3 is an acceptor-like state. Electrical simulations using Synopsys demonstrated that the DLTS probed zone is nearby gate edges and extends towards the source and drain contacts when the gate voltage changes from 0 V to 1 V. According to this feature, all traps have been localized very close to the source and drain implanted zones. The origin of these traps is likely related to damages produced during the formation of the p-doped regions and are assigned to bulk [2] and Si/SiO2 interface defects [3]. In the temperature range 77 K – 350 K, no traps have been detected in the active region, i.e. below the gate contact. Finally, this investigation highlights the good quality of the channel material and the ability of the studied FDSOI transistor to work as a qubit device. [1] R. Maurand et al., Nat. Commun. 7, 13575 (2016) [2] E. Fretwurst et al., Nucl. Instr. and Meth. in Phys. Res. A 388, 356 (1997) [3] E. Simoen et al., Phys. Status Solidi C 13, 718 (2016)

R.2.2
16:45
Authors : Stefano Calcaterra, Daniel Chrastina, Giovanni Isella, Andrea Ballabio, Giulio Tavani, Daniel Jirovec, Jaime Saez, Juan Aguilera, Georgios Katsaros
Affiliations : L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy; L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy; L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy; L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy; L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy; Institute of Science and Technology Austria, Am Campus 1, 3400 Klosterneuburg, Austria; Institute of Science and Technology Austria, Am Campus 1, 3400 Klosterneuburg, Austria; Institute of Science and Technology Austria, Am Campus 1, 3400 Klosterneuburg, Austria; Institute of Science and Technology Austria, Am Campus 1, 3400 Klosterneuburg, Austria;

Resume : Over the past few years, interest in quantum computing has increased exponentially, with the spin degree of freedom of electrons or holes in semiconductor quantum dots (QDs) representing one (among many) possible qubit realization schemes. Hole spin qubits are created by electrostatically gating a 2DHG to define QDs with discrete energy levels, and the spin states are Zeeman-split by a magnetic field. Germanium is the mainstream semiconductor with the highest p-type mobility: this is leading to a renewed interest in Ge as a material for applications beyond the dominance of silicon. Ge quantum wells (QWs), are attracting interest due to possible applications which benefit not only from the properties of the 2-dimensional hole gas (2DHG) formed in the Ge QWs, but from the compatibility of SiGe with Si-based growth and fabrication. Holes in Germanium are particularly favoured for qubit realization, because the strong spin-orbit coupling (SOC) allows spin-flip transitions to be induced applying an oscillating lateral electric field, rather than a localized time-varying magnetic field. Since the most common isotopes of Si and Ge are both spin-0, a major source of electron or hole spin dephasing is reduced by orders of magnitude compared to III-V based systems Ge-based hole QD systems have already been used to demonstrate systems of two qubits, four qubits, and a singlet-triplet qubit based on a double QD. A Ge QW was grown by a plasma-activated variant of chemical vapor deposition, LEPECVD (low-energy plasma-enhanced CVD, LEPECVD) on a Si0.3Ge0.7 substrate on virtual substrate with a linearly graded concentration profile from pure Si to SiGe 70% as a buffer layer. Electrical characterization measurements were performed at low temperature (1.6-10 K) on Hall bar devices, both gated and ungated, in order to understand the effect of the Al2O3 oxide deposition and of the metallic gate on the carrier mobility as a function of sheet density, reaching a value higher than 100000 cm^(2)/Vs. This allows analysis of scattering mechanisms and the calculation of the percolation density, one of the most relevant figures of merit for qubit stability, with a value of around 10^11 cm^(-2). Shubnikov–de Haas (SdH) oscillations and quantum Hall effect were observed. The last was observed in an anomalous form at high bias, lacking extended zero resistivity regions and corresponding plateaus in the transverse resistivity, while the peaks featured an elliptical shape instead of the theoretical Lorentzian shape. Finally, the Landau level lifetime, which corresponds to the quantum scattering lifetime, was extracted from SdH oscillations. This proved to be relatively high as compared to the momentum lifetime coming from mobility, which may help to explain the excellent results obtained by collaborating research groups which fabricated qubits on this material.

R.2.3
17:00
Authors : Linnea Bendrot, Eunjung Cha, Cezar Zota
Affiliations : IBM Research Europe - Ruschlikon, Switzerland

Resume : Cryogenic low-noise amplifiers (LNAs) based on InP high electron mobility transistors (InP HEMTs) have shown superior noise performance among all semiconductor circuits. This makes the cryogenic HEMT LNA a crucial component for the readout of microwave qubits in quantum computing systems. To improve the qubit readout, further advances of the HEMT noise properties are crucial. The minimum noise temperature of the InP HEMT according to the Pospieszalski model is in part dependent on the source resistance of the transistor. The noise properties of the HEMT will thus be improved by realizing Ohmic contacts with very low resistivity. In addition, the cryogenic environment enables the use of superconducting metals, such as Nb, to reduce the losses of transmission lines and matching networks. To this end, in this work, we study the properties of cryogenic Nb contacts on HEMT heterostructures. Several different metal stacks are studied, along with different annealing conditions. Our results indicate that Nb-based contacts can achieve very low contact resistivity, matching the state of the art of standard contact stacks, indicating the promising use of this material in superconducting LNA circuits.

R.2.4
17:15
Authors : C. Corley-Wiciak [1], C. Richter [2], M.H. Zoellner [1], C. Hsun[2], K. Anand [1], I. Zaitsev [1], C. Manganelli [1], A. A. Corley-Wiciak [1], Y. Yamamoto [1], E. Zatterin [3] , T. Schuelli [3], F. Reichmann [1], L. Schreiber [4] ,W. Langheinrich [5], W. M. Klesse [1], M. Virgilio [6], G. Capellini [1,7]
Affiliations : [1] IHP - Leibniz-Institut für innovative Mikroelektronik, Frankfurt(Oder), Germany [2] IKZ - Leibniz-Institut für Kristallzüchtung, Berlin, Germany [3] ESRF, Grenoble, France [4] JARA Institute for Quantum Information, RWTH Aachen, Germany [5] Infineon Technologies Dresden GmbH&Co.KG, Dresden, Germany [6] Università di Pisa, Pisa, Italy [7] Dipartimento di Scienze, Universita Roma Tre, Roma, Italy

Resume : Recently, electron spin qubits housed in electrostatic quantum dots in epitaxial Si/SiGe heterostructures have evolved by the demonstration of high coherence times and multi-qubit algorithms. Moreover, they offer a prospect for the large scale integration of solid state qubits since they are compatible with industrial fabrication processes, benefitting from the maturity of silicon-based CMOS technology. A qubit device of this type consists of gate electrodes fabricated lithographically on top of an epitaxial layer stack comprised of a thin (< 10 nm) thin Si quantum well (QW) layer sandwiched between plastically relaxed SiGe layers on a Si substrate. One key requirement for realizing large arrays of qubits with shared gate control is a high degree of spatial homogeneity of the lattice strains in the thin Si layer. The local strain environment around each qubit affects the energy level of the conduction band, and, likely, the qubit performance. Investigating this effect requires a technique combining high strain sensitivity with nanoscale spatial resolution. Our method of choice for this study is Scanning Xray Diffraction Microscopy (SXDM), performed at the nanodiffraction beamline ID01/ESRF, to map non-destructively the lattice strain tensor around several fully CMOS compatible electron shuttling devices for qubit application (“QuBus)”. We recorded diffraction maps for multiple asymmetric Bragg reflections in the same region of the device, allowing us to extract the full strain tensor at each spot. Our experiment benefits from the recent extremely bright source upgrade at ESRF, enabling a mapping the local lattice constants in the 10 nm Si layer with a lateral resolution of approximately 50 nm. Combined with the ability to selectively study the Si QW layer independent from the buffer, we achieve a three dimensional tomographic description of the lattice strain. We observe gradients larger than > 0.05 % over ranges > 1 µm for the three symmetric components of strain tensor in the QW layer. These are attributed to misfit dislocation networks in the plastically relaxed buffer layers. Simultaneously, we determine that the stress induced by the electrodes causes short-ranged (< 200 nm) variations of all strain tensor components on a similar magnitude. Based on the experimental data, we perform Finite Element Method (FEM) thermomechanical simulations to calculate the strain distribution around the electrodes at qubit operation temperature T = 20 mK. Moreover, this strain profile is translated into spatially resolved variations of the Si conduction band energy level by bandstructure simulation. This energy level variation is found to be of the magnitude as the charging energy of an electrostatic quantum dot of approx. 1 meV. Thus, we demonstrate that strain inhomogeneities occurring in a quantum device must be taken into account in the optimisation and design of material quality and qubit architecture for scaled CMOS-compatible quantum computing.

R.2.5
 
Poster session : -
17:30
Authors : Janguk Han, Yoon Ho Jang, Jihun Kim, Woohyun Kim, Cheol Seong Hwang†
Affiliations : Department of Materials Science and Engineering and Inter-university Semiconductor Research Center, College of Engineering, Seoul National University, Seoul, 08826, Republic of Korea

Resume : Graph data structures representing the relationship between two objects belonging to the non-Euclidean system cannot be described by vectors. Therefore, additional preprocessing steps are required to handle graph data with the conventional computing method, which leads to an increase in computing cost and loss of information. In addition, extracting hidden information from complex real-world networks requires plenty of computations in an unsupervised manner. The hidden information of graph data has conventionally been achieved by finding the effective similarity. This study demonstrates a novel structure, "metal cell at diagonal memristive crossbar array" (mCBA) for encoding and computing graph data. The mCBA hardware was fabricated by shorting the diagonal components of a passive crossbar array composed of Pt/Al2O3/HfO2/TiN self-rectifying memristors. The connected edges of the graph data are mapped to the low resistance state of the memristors. In contrast, the unconnected edges are represented by the high resistance state of the memristors. The node itself is represented by the shorted circuit at diagonal positions in mCBA. Graph data mapped to mCBA was analyzed through two grounding methods of the signal wires (bit line). The multi-ground method (MGM) could be used as a graph search algorithm by grounding two or more bit-lines of mCBA to suppress sneak current. The single-ground method (SGM) could be used as a similarity finding algorithm to find the hidden information of graph data by grounding to a single bit-line of mCBA to occur sneak current. The sneak current in mCBA, which consists of self-rectifying memristors, is generated through diagonal metal cells. SGM made it possible to effectively calculate the distance between nodes, the connection probability, and the similarity with a single voltage bias. We experimentally demonstrated pathfinding, link prediction, and community detection algorithms using mCBA to verify the feasibility and performance of the proposed MGM and SGM. The three algorithms using both SGM and MGM were successfully performed on mCBA to which non-Euclidean graph data was mapped without any preprocessing. The demonstration results confirmed that SGM could efficiently reproduce the similarity calculated by complex algorithms in conventional computing methods as the output current. In addition, the results of the variation effect analysis of unit cells confirmed the robustness of the mCBA-based graph algorithms.

R.P.1
17:30
Authors : Yoon Ho Jang, Sung Keun Shim, Janguk Han, Jihun Kim, Woohyun Kim, Cheol Seong Hwang
Affiliations : Department of Materials Science and Engineering and Inter-university Semiconductor Research Center, College of Engineering, Seoul National University, Seoul, 08826, Republic of Korea

Resume : In recent years, Reservoir Computing (RC), which is a temporal kernel-based computing method that processes input at the reservoir (a fixed recurrent network) and reads it from the readout layer, has been studied due to its cost-efficient learning in the machine learning field. However, in the memristor-based RC, the time scale is fixed with the material properties of the memristor, which makes it difficult to adjust the kernel properties. In this study, W/HfO2/TiN (WHT) memristor (M), resistor (R), and capacitor (C)-based temporal kernels with a new structure (1M1R1C and 2M1C) are proposed to solve the aforementioned issues. The R-C time constant of the circuit can be varied, and the kernel system with tunable dynamics can be demonstrated by combining the memristor with a capacitor and a normal resistor (or memristor). To verify the performance of the 1M1R1C kernel, the task of recognizing digit images in the MNIST database was conducted. With a compressed single readout network (196x10), the 1M1R1C kernel system achieved an accuracy of 90.1%. This kernel machine took 200 ns of time and ~25 pJ of energy to process one input pulse, which is 103 –104 times shorter and 100–400 times lower than the previous studies [1-3]. In addition, 1M1R1C kernel was applied to automatic medical diagnosis, which consists of breast cancer diagnosis using ultrasound images and arrhythmia diagnosis based on electrocardiogram. These two applications have vastly different operating signal frequencies (MHz to Hz). This study implemented a system for efficient medical diagnosis by optimizing the kernel system for each task. On the other hand, the 2M1C kernel system maps the processing results to two memristors (M1 and M2), which project the input signals into a higher-dimensional space. High dimensionality characteristics of the 2M1C kernel significantly improve the separability of the 2M1C kernel compared to 1M1R1C. In this study, 2M1C kernel was applied to Mackey-glass time series prediction and showed excellent performance compared with previous studies. The 2M1C kernel was implemented in hardware by integrating two WHT memristors and ZrO2/Al2O3/ZrO2 capacitor. 1M1R1C kernel system shows that the kernel can adapt to various tasks, and real-time data processing is possible by changing R and C in the corresponding structure. 2M1C kernel has proven that it can be applied to more complex tasks (time-series prediction) by further improving the processing power of the kernel through M1-M2 dual mapping. Both kernels implemented an energy-efficient system that can process sequential data sets in real-time. They can also be used to mimic sensory adaptation characteristics and to develop artificial sensory systems. [REFERENCE] [1] Midya, R. et al. Adv. Intell.Syst. 1, 1900084 (2019). [2] Du, C. et al. Nat. Commun. 8, 1–10 (2017). [3] Moon, J. et al. Nat. Electron. 2, 480–487 (2019).

R.P.2
17:30
Authors : Park Jinuk, Park Jin-Hong
Affiliations : Semiconductor and Display Engineering Dept;Department of Electronics and Computer Engineering

Resume : Because of oxygen vacancy inside of IGZO, metal materials such as Cu and Au easily diffuse into the thin film of IGZO. IGZO is essential material for non-volatile resistive switching for memory. In order to use it as a material for semiconductors, the properties of IGZO need to be studied in detail. Here we study IGZO I-V Curve depend on its composition. We used IGZO solution. High mobility and stability can be acquired.

R.P.3
17:30
Authors : Gyeongdo Baek, Jingi Gim, Sanghyeon Lim, Hyunjee Jung, Mohammad M. Afandi, Jehong Park, Jongsu Kim
Affiliations : Department of Display Science and Engineering, Pukyong National University

Resume : The Ga2O3 activated with various rare-earths or transition metals has been studied as a luminescent material for photoluminescence (PL) and electroluminescence (EL) devices. Herein, the alternating-current (AC)-driven metal-oxide-semiconductor structure-based electroluminescence (MOS-EL) device with Ga2O3:Cr3+/SiOx oxide layers on a silicon (Si) substrate has been demonstrated. The Ga2O3:Cr3+ oxide films were optimized in the Cr3+ concentration and the annealing temperature, which revealed that the monoclinic β-Ga2O3 with a thickness of less than 100 nm was formed on the Si substrate accompanied by the amorphous SiOx layer with a thickness of less than 10 nm. The MOS-EL device started to operate below 10 V (corresponding to less than 1 MV/cm in an electric field) under the sinusoidal AC waveform with the near-infrared emission attributed to the d-d transitions of Cr3+ ions. The MOS EL device showed the long-term reliability as well as a large opto-electric hysteresis.

R.P.4
17:30
Authors : Sanghyeon Lim, Jingi Gim, Gyeongdo Baek, Hyunjee Jung, Mohammad M. Afandi, Jehong Park, Jongsu Kim
Affiliations : Department of Display Science and Engineering, Pukyong National University

Resume : The alpha-beta mixed phased Zn2SiO4:Mn2+-based metal–oxide–semiconductor electroluminescence (MOS EL) device on silicon wafer is demonstrated through a rapid thermal annealing process. The MOS EL device consists of two sublayers: alpha and beta double-phased crystalline-like Zn2SiO4:Mn2+ for an emitting layer and amorphous SiOx interlayer as a carrier-injecting–accelerating layer on a silicon wafer. It exhibits the overlapped spectrum of the green from the crystalline alpha phase and the yellow from the amorphous-like beta phase, and its relative EL intensities are strongly dependent on the annealing temperature; the lower the temperature, the more dominant the yellow peak. Also it shows asymmetric electrical and optical properties; the EL intensity is higher under a positive applied voltage than a negative polarity. Furthermore. the EL performance is improved by codoping with Mn2+ ions; their EL decay increases in the case of As ions, and the EL intensity increases in the case of Al ions. Finally, it exhibits the best luminous efficiency of more than 1  lm W−1 at a lower threshold voltage (the electric field of less than 1 MV cm−1) and the higher long-term reliability.

R.P.5
17:30
Authors : Krunoslav Juraić (1), Davor Gracin (1), Marija Ivezić (1), Ivan Vadla (2)
Affiliations : (1) Ruđer Bošković Institute, Bijenička cesta 54, 10000 Zagreb, Croatia (2) Solvis Ltd. Ulica Vesne Parun 15, Varaždin, Croatia

Resume : Ethylene vinyl acetate copolymer (EVA) is the most commonly used encapsulation material in silicon photovoltaic modules. Its main function is to bond the module components together, to protect the solar cells from mechanical and environmental influences and to insulate them electrically. It is well known that the performance of solar modules (conversion efficiency) decreases when they are operated outdoors. One of the reasons for this is the degradation of the encapsulation material by UV light. To prevent photodegradation, various stabilisers are added to the EVA encapsulation material. However, the effect cannot be completely eliminated and some recent studies suggest that the added stabiliser is involved in the EVA photodegradation (yellowing process). Several new approaches and concepts are proposed as a solution for EVA photodegradation and improving conversion efficiency. One of them is the use of downshifting materials as an additive to EVA encapsulant. Downshifting materials convert photons with higher energy (UV range) into photons with lower energy (in the visible part of the spectrum), where the quantum efficiency of the solar cell is higher. In this way, they protect the c-Si solar cell by blocking the UV light and improving the conversion efficiency. One of the materials with such a function is organic luminescent dyes such as perylene and naphthalimide-based dyes, which have improved photostability under sunlight irradiation. However, several studies have shown that they degrade under UV irradiation below 345 nm. In addition to protecting and improving the efficiency of solar cells, EVA foil modified with organic luminescent dyes can also have an aesthetic function. In recent years, interest in coloured modules has increased. The aim of this work was to investigate the influence of UV irradiation on the performance of silicon solar cells encapsulated with a EVA foil modified with organic luminescent dyes. c-Si mini solar modules (20x20mm) were encapsulated with 4 different encapsulation materials: standard transparent EVA foil, EVA foil with UV cutoff additive, EVA foil modified with red and yellow organic luminescent dyes: Lumogn Red 305 and Lumogen Yellow 087. The degradation of the solar cell parameters was monitored by checking the IV characteristics and the spectral response of the solar cells during longer period of exposure to UV light. We also monitored the transmittance in the UV-Vis region of the EVA encapsulant. c-Si solar modules encapsulated with organic luminescent dyes were also tested under real outdoor conditions on a large scale. We compared and discussed the results for all encapsulation samples used.

R.P.6
Start atSubject View AllNum.
 
Applied materials : Gilles Patriarche
09:30
Authors : Yvon CORDIER
Affiliations : Université Côte d’Azur, CNRS, CRHEA

Resume : Within the two last decades, group III-Nitrides on Silicon and especially GaN based structures have impressively developed, rising the hopes to rapidly integrate such materials in low cost CMOS compatible process lines. The story mainly started with high-frequency GaN high electron mobility transistors (HEMTs) to be developed on Silicon for reducing the cost of telecommunication systems such as for 5G, followed by light emitting diodes (LEDs) for solid state lighting. In-spite of serious difficulties to grow such materials on large diameter substrates, some succeeded and a new industry has emerged. But today, the GaN-on-Si industry experiences a new rise, much impressive than the original one, where in the one hand GaN HEMTs are developed to replace Silicon or SiC based electron devices in power switching systems and in the other hand, micro-LEDs are developed for displays. Even when there was some success in the development of alternative epitaxy strategies like nanowire-based LEDs, the great majority of III-Nitride materials grown today on Silicon relies on the use of planar growth on an Aluminum Nitride nucleation layer [1] able to protect the Silicon substrate from chemical reactions with metals like Gallium [2]. Furthermore, thanks to a lattice mismatch of 2.4% with GaN, the use of AlN induces a compressive strain in the GaN layers grown on it, which is the key to compensate the thermoelastic strain responsible for the generation of wafer bowing and cracks in the films. Such AlN nucleation layer can be used to integrate others materials like Graphene or cubic Silicon Carbide. Also, other devices can benefit the GaN-on-Si platform technology, such as micro-electro-mechanical systems on membranes easy to fabricate thanks to the selectivity of the etching of Silicon with regards to III-Nitrides, or surface or bulk acoustic wave devices for sensors or high-frequency filters. The preferred substrate orientations for III-Nitrides, Si(111) or Si(110) are not in the mainstream for the monolithic co-integration with Silicon CMOS (Si(100)). Even when such integration has been demonstrated via the use of substrates with a noticeable miscut angle or via hybrid SOI substrates (Si(111)/Si(100)) the noticeable progresses in the hetero-epitaxy of III-Nitrides on large diameter substrates and in the layer transfer on CMOS wafers make the approach interesting for heterogeneous co-integration. [1] A. Watanabe, T. Takeuchi, K. Hirosawa, H. Amano, K. Hiramatsu, I. Akasaki, The growth of single crystalline GaN on a Si substrate using AIN as an intermediate layer, J. Crystal Growth 128 (1993) 391. [2] A. Dadgar, A. Strittmatter, J. Bläsing, M. Poschenrieder, O. Contreras, P. Veit, T. Riemann, F. Bertram, A. Reiher, A. Krtschil, A. Diez, T. Hempel, T. Finger, A. Kasic, M. Schubert, D. Bimberg, F.A. Ponce, J. Christen, A. Krost, Metalorganic chemical vapor phase epitaxy of gallium-nitride on silicon, Phys. Status Solidi C 6 (2003) 1583.

R.3.1
10:00
Authors : Roman Krahne,1,* Vincenzo Caligiuri,2 Aniket Patra,2 Renuka Pothuraju,1 and Antonio De Luca2
Affiliations : 1 Istituto Italiano di Tecnologia, Via Morego 30, 16163 Genova, Italy 2 Dipartimento di Fisica, Università della Calabria, 87036 Rende, and CNR-Nanotec, Italy

Resume : Layered Metal-Dielectric-Metal (MDM) structures provide a highly versatile platform for light manipulation. We have demonstrated that this system can act as a photonic cavity in the visible sustaining ENZ resonances, and that the similarity of the Helmholtz to the Schroedinger equation enables to apply the formalisms of quantum mechanics, providing deep physical insights.[1-3] This analysis can be extended to multiple cavity layers that form ENZ bands, which can be described by the Kronig-Penney model.[4] The possibility to tune the splitting of the two low-energy cavity modes provides additional degrees of freedom for tuning the resonances to the emission and absorption bands of emitting dyes, which can be used to enhance their radiative rate. [5] Therefore, MDM cavities constitute ultrathin optical cavities with resonances in the visible range that can be implemented on large areas by cost-efficient thermal deposition methods. This platform allows to fabricate superabsorbers, optical cavities for light emission enhancement, modulation and photodetection, and can provide active elements in ultrafast signal processing.[6] Here, I will present our most recent progress on light-matter interaction [7,8] and coupled resonances in such cavities, and their implementation in devices for light emission enhancement. References [1] V. Caligiuri, M. Palei, G. Biffi, S. Artyukhin, R. Krahne, A Semi-Classical View on Epsilon-Near-Zero Resonant Tunneling Modes in Metal/Insulator/Metal Nanocavities, Nano Lett. 2019, 19, 3151-3160. [2] V. Caligiuri, M. Palei, G. Biffi, R. Krahne, Hybridization of epsilon-near-zero modes via resonant tunneling in layered metal-insulator double nanocavities, Nanophoton. 2019, 8, 1505. [3] B. Zappone, V. Caligiuri, A. Patra, R. Krahne, A. De Luca, Understanding and Controlling Mode Hybridization in Multicavity Optical Resonators Using Quantum Theory and the Surface Forces Apparatus, ACS Photonics 2021, 8, 3517-3525. [4] V. Caligiuri, G. Biffi, A. Patra, R. D. Pothuraju, A. De Luca, R. Krahne, One-Dimensional Epsilon-Near-Zero Crystals, Adv. Photon. Res. 2021, 2 , 2100053. [5] V. Caligiuri, M. Palei, M. Imran, L. Manna, R. Krahne, Planar Double-Epsilon-Near-Zero Cavities for Spontaneous Emission and Purcell Effect Enhancement, ACS Photonics 2018, 5, 2287-2294. [6] J. Kuttruff, D. Garoli, J. Allerbeck, R. Krahne, A. De Luca, D. Brida, V. Caligiuri, N. Maccaferri, Ultrafast all-optical switching enabled by epsilon-near-zero-tailored absorption in metal-insulator nanocavities, Commun. Phys. 2020, 3, 114. [7] V. Caligiuri, G. Biffi, M. Palei, B. Martín-García, R. D. Pothuraju, Y. Bretonnière, R. Krahne, Angle and Polarization Selective Spontaneous Emission in Dye-Doped Metal/Insulator/Metal Nanocavities, Adv. Opt. Mater. 2020, 8, 1901215. [8] A. Patra, V. Caligiuri, R. Krahne, A. De Luca, Strong Light–Matter Interaction and Spontaneous Emission Reshaping via Pseudo-Cavity Modes, Adv. Opt. Mater. 2021, 9, 2101076.

R.3.3
10:15
Authors : Elena Hardt(1), Carlos Alvarado Chavarin(1), Soenke Gruessing(2), Julia Flesch(3), Oliver Skibitzki(1), Davide Spirito(1), Gian Marco Vita(4), Giovanna De Simone(4), Alessandra di Masi(4), Changjiang You(3), Bernd Witzigmann(5), Jacob Piehler(3), and Giovanni Capellini(1,4)
Affiliations : (1)IHP - Leibniz Institut für innovative Mikroelektronik, Im Technologiepark 25, 15236 Frankfurt (Oder),Germany (2)University of Kassel, Wilhelmshoeher Allee 71, 34121 Kassel, Germany (3)University of Osnabrueck, Barbarastrasse 11, 49076 Osnabrueck, Germany (4)Department of Sciences, Università Roma Tre, Viale G. Marconi 446, 00156, Roma, Italy (5)Friedrich-Alexander Universität Erlangen-Nürnberg, Konrad-Zuse Str. 3/5, 91052 Erlangen, Germany

Resume : Recent improvements in THz sources and detectors have enabled novel technological platforms with promising new perspectives in several fields, such as medicine, pharmacy and security screening. Here, the use of the electromagnetic spectrum in the THz range (0.1-3 THz) allows label-free, reliable measurements of biomolecules and protein conformation changes [1, 2]. One of the major hurdles in this technology is that the interaction of bioanalytes with THz radiation is physically limited due to the huge size mismatch between the analytes (few nm) and the wavelengths of the electromagnetic field (several tens of µm) [1]. To overcome this problem, exploiting plasmonic effects has been suggested to enhance the spectroscopy signal with several order of magnitute thanks to the loal field enhancement [3]. In this work, we present the use of a highly n-doped Ge-based bow-tie-shaped plasmonic antennas operating in the THz range. The antennas have been fabricated using CMOS-compatible materials in an industrial-grade pilot line. According to finite element simulations, the antennas are able to induce sub-wavelength THz field enhancement within the hotspots, resulting in reliable semi-quantitative detection of adsorbed BSA proteins in the mg/mL range, featuring a layer thickness down to 30-50 nm. We futhermore study and discuss the effect of antenna design, size, and arrangement on its sensitivity. The results show CMOS-compatible Si/Ge-based antennas are a good choice due to their potential for mass fabrication of low-cost biosensors. We believe, the initial results could pave the way to a CMOS-integrated, low-cost, on-chip microfluidic biosensor platform. [1] T. Mancini, R. Mosetti, A. Marcelli, M. Petrarca, S. Lupi and A. D'Arco, "Terahertz Spectroscopic Analysis in Protein Dynamics: Current Status," Radiation, vol. 2, pp. 100-123, 7 February 2022. [2] N. L. Henry and D. F. Hayes, "Cancer biomarkers," Molecular Oncology, vol. 6, pp. 140-146, 2012. [3] S. Law, R. Liu and D. Wasserman, "Doped semiconductors with band-edge plasma frequencies," Journal of Vaccum Science & Technology B, vol. 32, 31 July 2014.

R.3.4
10:30 Coffee break    
 
Advanced heterogenous integration : Shimura Yosuke
11:00
Authors : Jeehwan Kim
Affiliations : MiT

Resume : For future of electronics such as bioelectronics, 3D integrated electronics, and bendable electronics, needs for flexibility and stackability of electronic products have substantially grown up. However, conventional wafer-based single-crystalline semiconductors cannot catch up with such trends because they are bound to the thick rigid wafers such that they are neither flexible nor stackable. Although polymer- based organic electronic materials are more compatible as they are mechanically complaint and less costly than inorganic counterparts, their electronic/photonic performance is substantially inferior to that of single- crystalline inorganic materials. For the past half a decade, my research group at MIT has focused on mitigating such performance-mechanical compliance dilemma by developing methods to obtain cheap, flexible, stackable, single-crystalline inorganic systems. In today’s talk, I will discuss about our strategies to realize such a dream electronic system and how these strategies unlock new ways of manufacturing advanced electronic systems. I will highlight our remote epitaxy technique that can produce single-crystalline freestanding membranes from any compound materials with their excellent semiconducting performance. In addition, I will present unprecedented artificial heterostructures enabled by stacking of those freestanding 3D material membranes, e.g., world’s smallest vertically-stacked full color micro-LEDs, world’s best multiferroic devices, battery-less wireless e-skin, and reconfigurable hetero-integrated chips with AI accelerators.

R.4.1
11:30
Authors : Charlotte Van Dijck, Florian Maudet, Sourish Banerjee, Veeresh Deshpande, Catherine Dubourdieu
Affiliations : Helmholtz-Zentrum-Berlin für Materialen und Energy, Freie Universität Berlin; Helmholtz-Zentrum-Berlin für Materialen und Energy; Helmholtz-Zentrum-Berlin für Materialen und Energy; Helmholtz-Zentrum-Berlin für Materialen und Energy; Helmholtz-Zentrum-Berlin für Materialen und Energy, Freie Universität Berlin

Resume : Amorphous metal oxide semiconductors have been investigated for thin film transistor (TFT) applications because they offer higher mobility at low processing temperatures. While amorphous Indium Gallium Zinc Oxide (IGZO) and Indium Zinc Oxide (IZO) have been extensively studied as channel materials for TFTs, amorphous gallium oxide (a-GaOx) is an interesting transparent conducting oxide with wide bandgap (~4.9 eV) and ability to have n-type carriers. Additionally, it also has potential applications in UV detectors, solar cells, humidity sensors etc. This makes it suited for back-end-of-line (BEOL) integration of high voltage transistors and sensing devices. However, a detailed study of transistors based on an ALD grown a-GaOx channel has not yet been reported. In this work the amorphous GaOx films are deposited using plasma-enhanced atomic layer deposition (PE-ALD). The use of ALD allows for low deposition temperatures (250°C) and uniform films that are required for BEOL integration of TFTs. The plasma-enhanced ALD process developed recently by our group [1], allows tuning the conductivity of the a-GaOx layer through oxygen plasma exposure time during deposition. Utilizing this we show a-GaOx conductive channel back-gated transistors featuring 20 nm Al2O3 as the gate oxide. Three different channel thicknesses (22, 50 and 75 nm) were investigated. Detailed investigation of transistor characteristics such as subthreshold slope (SS), threshold voltage, ON current and their dependency on channel thickness and channel length is performed. Transistors with SS < 150 mV/dec and an ON/OFF ratio of 105 have been shown for a channel length of 6 µm. The devices show a hysteresis in the drain current with gate voltage sweep, which could be associated with charge traps in the channel surface exposed to atmosphere. Therefore, in order to minimize this effect, different encapsulations of the a-GaOx channel with in-situ ALD grown Al2O3 and ex-situ PECVD grown SiO2 were studied and the comparison of the two encapsulations will be presented. Based on these investigations, pathways to optimize the device characteristics (improvement in ON current, reduction in drain current hysteresis) will be discussed. [1] H. Kröncke et al., “Effect of O2 plasma exposure time during atomic layer deposition of amorphous gallium oxide.” Journal of Vacuum Science & Technology A 39, 052408 (2021)

R.4.2
11:45
Authors : C. Zucchetti(1), T. Guillet(2), A. Marchionni(1), A. Hallal(2), P. Biagioni(1), C. Vergnaud(2), A. Marty(2), H. Okuno(3), A. Masseboeuf(2), M. Finazzi(1), F. Ciccacci(1), M. Chshiev(2), F. Bottegoni(1), M. Jamet(2)
Affiliations : (1) LNESS-Dipartimento di Fisica, Politecnico di Milano, Piazza Leonardo da Vinci 32, 20133, Milano, Italy (2) Université Grenoble Alpes, CEA, CNRS, Grenoble INP, IRIG-Spintec, 38000 Grenoble, France (3) Université Grenoble Alpes, CEA, IRIG-MEM, 38000 Grenoble, France

Resume : The discovery of topological insulators (TIs) hold great promise for the field of spintronics thanks to the peculiar spin texture of their surface states. Indeed, the surface states of a TI possess spin-momentum locking which is expected to convert with high-efficiency a charge current into a spin current (Rashba-Edelstein effect) and viceversa (inverse Rashba-Edelstein effect). However, most of the experimental techniques employed in spintronics investigations require ferromagnetic films and TIs are known to chemically react when they are in contact with them [1]. This is detrimental for the spin-related properties of the TIs which are hosted by surface states. For this reason, a versatile platform allowing for the exploitation of the assets of TI is still lacking. Here, we circumvent the problem by exploiting germanium as a substrate for the growth of Bi2Se3, a prototypical TI. As probed by cross-sectional transmission electron microscopy, we obtain a very sharp Bi2Se3/Ge interface. Germanium adds two key features: first, it allows for efficient spin generation via optical orientation, thus avoiding any ferromagnetic material to generate the spin current; then, the 4% of lattice mismatch between Ge and Si allows for the integration of this system with the mainstream Si-based technology. To probe the spin properties of TIs we thus generate a spin population by optical spin orientation in Ge [2]. This photogenerated spins diffuse as a spin current toward the Bi2Se3, which acts as the spin detector. In this way, we probe the spin properties of the Bi2Se3/Ge pristine interface by investigating the spin-to-charge conversion taking place in the interface states. We compare the spin-to-charge conversion in Bi2Se3/Ge with the one taking place in Pt (a standard spin detector) in the same experimental conditions. Notably, the sign of the spin-to-charge conversion given by the TI detector is reversed compared to the Pt one, while the efficiency is comparable. By exploiting first-principles calculations, we ascribe the sign reversal to the hybridization of the topological surface states of Bi2Se3 with the Ge bands. [3] Hence, semiconductors not only preserve the elemental sharpness of TI and substrates at the interface, but the hybridization of electronics states allows engineering a device where, by gating the heterostructure, the spin-to-charge conversion could be tuned in magnitude and sign. These results pave the way for the implementation of highly efficient spin detection in TI-based architectures compatible with semiconductor-based platforms. [1] K. Ferfolja et al., J. Phys. Chem. C 122, 9980 (2018) [2] C. Zucchetti et al., Phys. Rev. B 96, 014403 (2017) [3] T. Guillet, C. Zucchetti et al., Phys. Rev. B 101, 184406 (2020)

R.4.4
12:00
Authors : Heintz, A.*(1,2), Ilahi, B. (1,2), Pofelski, A. (4), Botton, G. A. (4,5), Patriarche, G. (6), Barzaghi, A. (3), Fafard, S. (1,2), Arès, R. (1,2), Isella, G. (3), Boucherif, A. (1,2)
Affiliations : 1 Institut Interdisciplinaire d’Innovation Technologique (3IT), Université de Sherbrooke, 3000 Boulevard Université, Sherbrooke J1K OA5 QC, Canada 2 Laboratoire Nanotechnologies Nanosystèmes (LN2) —CNRS UMI-3463, Institut Interdisciplinaire d’Innovation Technologique (3IT), Université de Sherbrooke, 3000 Boulevard Université, Sherbrooke J1K OA5 QC, Canada 3 L-NESS and Dipartimento di Fisica, Politecnico di Milano, Via Anzani 42, I-22100 Como, Italy 4 Department of Materials Science and Engineering, McMaster University, Hamilton, ON L8S 4M1, Canada 5 Canadian Light Source, 44 Innovation Boulevard, Saskatoon, SK S7N 2V3 Canada 6 Centre de Nanosciences et de Nanotechnologies, CNRS, Univ. Paris-Sud, Université Paris-Saclay, C2N – Marcoussis, 91460 Marcoussis, France *Corresponding author. Email: alexandre.heintz@usherbrooke.ca; abderraouf.boucherif@usherbrooke.ca

Resume : A perfectly compliant substrate would allow the monolithic integration of high-quality semiconductor materials such as Ge and III-V on Silicon (Si) substrate, enabling novel functionalities on the well-established low-cost Si technology platform. Here, we demonstrate a compliant Si substrate allowing defect-free epitaxial growth of lattice mismatched materials. The method is based on the deep patterning of the Si substrate to form micrometer-scale pillars and subsequent electrochemical porosification. The investigation of the epitaxial Ge crystalline quality by X-ray diffraction, transmission electron microscopy and etch-pits counting demonstrates the full elastic relaxation of defect-free microcrystals. The achievement of dislocation free heteroepitaxy relies on the interplay between elastic deformation of the porous micropillars, set under stress by the lattice mismatch between Ge and Si, and on the diffusion of Ge into the mesoporous patterned substrate attenuating the mismatch strain at the Ge/Si interface.

R.4.5
12:15 Lunch break    
 
Group IV optoelectronics : Andriy Hikavyy
14:00
Authors : Andrea Barzaghi, Virginia Falcone, Stefano Calcaterra, Raffaele Giani, Andrea Ballabio, Giovanni Isella, Daniel Chrastina, Michele Ortolani, Michele Virgilio, Jacopo Frigerio.
Affiliations : L-NESS Dipartimento di Fisica, Politecnico di Milano, Via Francesco Anzani 42, Como, I-22100, Italy; L-NESS Dipartimento di Fisica, Politecnico di Milano, Via Francesco Anzani 42, Como, I-22100, Italy;L-NESS Dipartimento di Fisica, Politecnico di Milano, Via Francesco Anzani 42, Como, I-22100, Italy;L-NESS Dipartimento di Fisica, Politecnico di Milano, Via Francesco Anzani 42, Como, I-22100, Italy;L-NESS Dipartimento di Fisica, Politecnico di Milano, Via Francesco Anzani 42, Como, I-22100, Italy;L-NESS Dipartimento di Fisica, Politecnico di Milano, Via Francesco Anzani 42, Como, I-22100, Italy;L-NESS Dipartimento di Fisica, Politecnico di Milano, Via Francesco Anzani 42, Como, I-22100, Italy; Dipartimento di Fisica, Sapienza Università di Roma, Piazzale Aldo Moro 5, I-00185 Rome, Italy; 3Dipartimento di Fisica “E. Fermi”, Università di Pisa, Largo Pontecorvo 3, I-56127 Pisa, Italy; L-NESS Dipartimento di Fisica, Politecnico di Milano, Via Francesco Anzani 42, Como, I-22100, Italy

Resume : In the last decade, mid-infrared integrated photonics has raised a great interest due to the envisioned applications in molecular sensing, security, environmental monitoring and free-space communications. The current technology, based on the mature SOI platform has already reached a significant technology readyness level, but it cannot be exploited at wavelegths > 3.2 μm. Extending the operational wavelengths toward the long-wave infrared region (LWIR) ( 8 μm – 12 μm) presents many challenges and many material platforms incuding III-V semiconductors, halides, chalchogenides are currently under investigation. In this framework, the SiGe-on-Si material platform is considered very promising. First of all, by taking advantage of the wide transparency range of Ge, low-loss waveguides operating up to λ =11 μm have been recently demonstrated, as well as many passive photonic components including Mach-Zehnder interferometers, resonators, spectrometers and modulators. Nevertheless, key functionalities such as wavelength conversion, photodetection and high speed optical modulation are still missing. In this framework, Ge/SiGe quantum wells could be exploited to fill this gap. Intersubband optical transitions in the valence band of such heterstructures can be used for light detection, for high-speed modulation through the quantum confined Stark effect, and for wavelength conversion through second harmonic generation. Ge/SiGe QWs can be easily grown on top of SiGe buffers, making them fully compatible with the existing SiGe-on-Si material platform. In this work, we show the experimental demonstration of second harmonic generation at mid-infrared frequencies and we propose a concept for waveguide integration of Ge quantum wells. We will also show recent experimental results on intersubband absorption and a feasibility study on high speed optical modulation through the quantum confined Stark effect.

R.5.1
14:30
Authors : Luciana Di Gaspare 1*, Enrico Talamas Simola 1 , Michele Montanari 1, Tommaso Venanzi 2, Marina Cagnon Trouche 2, Leonetta Baldassarre 2, Luca Persichetti 1, Cedric Corley 3, Marvin Zöllner 3, Giovanni Capellini 1 3, Michele Virgilio 4, Michele Ortolani 2, and Monica De Seta 1
Affiliations : 1 Dipartimento di Scienze, Università Roma Tre, Roma 00146, Italy; 2 Dipartimento di Fisica, Sapienza Università di Roma, Roma 00185, Italy; 3 IHP- Leibniz Institut für innovative Mikroelektronic, Frankfurt (Oder) 15236, Germany; 4 Dipartimento di Fisica “E. Fermi”, Università di Pisa, Pisa 56127, Italy; *contact email: luciana.digaspare@uniroma3.it

Resume : The widening of quantum cascade laser (QCL) active materials to Group IV systems promises to create a breakthrough into semiconductor laser science and technology, enabling compact THz sources spanning the whole THz range and operating at room temperature. The recent observation of THz electroluminescence originating from intersubband transitions (ISBT) in the conduction band of n-type Ge/SiGe multi-quantum wells [1] suggests that the quality of this material system is approaching the standard required for QCL applications. Several criticalities still remain in the epitaxy of such complex structures, related to the high Ge/Si lattice mismatch and to the tricky control over the spatial localization of dopants in the Ge layers. We discuss here the advances achieved in the growth of n-type Ge/SiGe quantum cascade (QC) active regions by ultra-high-vacuum chemical vapor deposition (UHV-CVD). In particular, we report Fourier Transform Infrared absorption experiments performed at 10K to study the effect of the doping profile on the absorption spectra. Strain-compensated QC structures having 20 repetitions of the SW design used in Ref. 1 were grown on a Si(001) substrate. The central 3nm of different wells of the QC structure has a nominal 3x1017 cm-3 doping. The optical spectra were simulated using a multivalley self-consistent Schrödinger-Poisson solver. Since the expected spectra are strongly dependent on the dopant atom positions this was meant to explore the spatial control of dopants and the effect of field distortion produced by localized donors on the absorption spectra. When the dopant atoms are located in the second well of the structures as envisaged in the QCL design, a single absorption peak at 16 meV related to the lasing ISBT has been observed, in good agreement with the emission peak observed in [1]. In the samples doped in the fourth well a second more intense peak at 32 meV is instead present. The very good agreement between the measured and simulated spectra, confirms the good control achieved in the dopant position. Moreover, aiming at the realization of a Ge/SiGe QCL, we present the progress made in the deposition of very high-quality QC structures deposited on silicon on insulator (SOI) substrates, which facilitates the double metal waveguides fabrication process with respect to a standard Si substrate. A first series of samples, featuring 100 repetitions of the cascade structure (overall thickness of about 5 m), has been deposited. XRD and STEM data demonstrate the high quality of the material. Furthermore, a very low threading dislocation density of 1.5x106 cm -2 has been measured. This research has been supported by Regione Lazio, program POR FESR 2014-2020, project n. A0375-2020-36579 “Teralaser”. [1] D. Stark, et al. "THz intersubband electroluminescence from n-type Ge/SiGe quantum cascade structures", App. Phys. Lett. 118, 101101 (2021)

R.5.2
14:45
Authors : Andrea Barzaghi, Virginia Falcone, Stefano Calcaterra, Raffaele Giani, Francesco Rusconi, Andrea Ballabio, Daniel Chrastina, Paolo Biagioni, Giovanni Isella and Jacopo Frigerio
Affiliations : L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy; L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy; L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy; L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy; L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy; L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy; L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy; L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy; L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy; L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy

Resume : Detection in the mid-IR wavelength range is of great interest to enable gas sensing and biological spectroscopic detectors, which require an absorbing material in the 3-5 and 8-13 µm windows. Commercial detectors employ either MCT or InSb as sensing material, but they’re fragile, difficult to process and integrate on silicon, resulting in a very high cost. Cost effective MIR detectors may be fabricated by exploiting intersubband transitions (ISTs) in high-quality Ge multiple quantum wells nanostructures, thanks to the possibility of integration in standard CMOS processes. Moreover, the absorption of such quantum well infrared photodetectors (QWIP) can be tuned by varying the quantum well width and strain level, therefore shifting the transition energy to the desired value. n- and p-type SiGe QWIPs, which present a weaker absorption compared to pure Ge MQW structures, have already been demonstrated in the past, and germanium QWIPs have only been demonstrated by exploiting ISTs in the conduction band. Here p-type Ge MQWs structures are investigated as a platform for the fabrication of QWIPs which, thanks to the non-parabolicity and band-mixing effects in the valence band, present both TE and TM absorption and can therefore be employed in both vertical illuminated and waveguide geometries, while having a larger absorption coefficient compared to SiGe designs. To exploit this effect, three different Ge MQW designs, each with a different quantum well width, were grown by LEPECVD on high-resistivity Si wafers. Moreover, a second set of samples was grown with the same quantum well thickness to investigate the effect of different doping levels. A virtual substrate with a linearly graded concentration profile from pure Si to SiGe 80% was used as a buffer layer to achieve high quality QW superlattices with a low density of defects. High resolution x-ray diffraction (XRD) reciprocal space maps were then acquired and the relevant parameters (i.e. quantum well thickness, superlattice period, average Ge content of the MQW stack) and used as the starting point of the k·p simulations. Room-temperature absorption spectra of the MQW stacks were obtained by measuring the transmittance and reflectance spectra at normal incidence with a FTIR spectrometer and a LN2-cooled MCT detector, and the resulting spectra clearly show an absorption peak which depends on the QW thickness and can be attributed to the LH1-LH2 intersubband transition by 8-band k·p simulations. Moreover, the same measurement has been performed on samples with different doping levels, showing that in undoped samples the Fermi level is in the energy gap and no transitions can be observed, while at a large doping level free-carrier absorption is strongly increased and the absorption peak is less definite. Finally, transmission spectra have been measured in an optical cryostat as a function of temperature, clearly showing that the absorption peak does not shift with temperature, a strong signature of ISTs.

R.5.3
15:00
Authors : Raffaele Giani, Stefano Calcaterra, Andrea Barzaghi, Andrea Ballabio, Jacopo Frigerio, Giovanni Isella
Affiliations : L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy; L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy; L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy; L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy; L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy; L-NESS, Dipartimento di Fisica, Politecnico di Milano, P.zza Leonardo da Vinci, 32 20133 Milano, Italy

Resume : Germanium on silicon photodiodes have been studied for more than twenty years, mainly for integrated photonics in a waveguide configuration. [1] These devices tolerate fairly high dark current densities since the small volume and high optical power results in a sufficiently large signal to noise ratio. Vertically illuminated photodetectors are instead required for imaging applications of interest in the automotive and biomedical areas. In this case, a sufficiently low dark current density is required for the fabrication of multipixel devices. [2] [3] In recent years the reduction of threading dislocations [4] and the implementation of surface passivation [5] strategies have been investigated in order to reduce the dark current density. Instead, the effect of doping in the silicon-germanium heterostructure on the dark current has been not fully analyzed. In order to evaluate the impact of silicon substrate doping on dark current and photoresponse, a set of germanium photodiodes were grown by LEPECVD and microfabricated by optical lithography. All the investigated photodiodes feature a 1500 nm thick nominally intrinsic germanium layer and heavily doped germanium top contact layer with a thickness of 100 nm. The silicon substrates doping has been varied from 10^(14) cm^(-3) to 10^(19) cm^(-3). Current/voltage measurements have been performed on devices featuring different diameters to obtain the bulk and surface contribution to the total dark current density. Temperature dependent current/voltage and capacitance/voltage measurements have been performed to highlight the relative weight of the different physical mechanisms giving rise to the dark current such as diffusion, generation-recombination and trap-assisted tunnelling. To characterize the photoresponse of the photodetectors, the responsivity and the specific detectivity were measured for photodiodes at the interesting wavelengths (1000-1700 µm) and for different negative biases. We have observed a monotonic dependence of photodiodes performance as a function of substrate doping, highlighted by the relevant role of substrate resistivity in Ge on Si devices. [1] J. Michel, J. Liu, and L. C. Kimerling, Nat. Photonics 4, 527 (2010). [2] R. Kaufmann, G. Isella, A. Sanchez-Amores, S. Neukom, A. Neels, L. Neumann, A. Brenzikofer, A. Dommann, C. Urban, and H. von Känel, J. Appl. Phys. 110, 023107 (2011) [3] G. Xu et al., IEEE Photonics Technology Letters, vol. 34, no. 10, pp. 517-520, 15 May15, 2022, doi: 10.1109/LPT.2022.3168308. [4] H. Tetzner, I. A. Fischer, O. Skibitzki, M. M. Mirza, C. L. Magnanelli, G. Luongo, D. Spirito, D. J. Paul, M. De Seta, and G. Capellini, Appl. Phys. Lett. 119,153504 (2021) doi: 10.1063.5.0064477 [5] Joonas Isometsa, Tsun Hang Fung, toni P. Pasnen, Hanchen Liu, Marko Yli-koski, Ville Vahanissi, and Hele Savin, APL Materials 9, 1111113 (2021) doi: 10.1063/5.0071552

R.5.4
15:15
Authors : Agnieszka Anna Corley-Wiciak1, Davide Spirito1, Omar Concepción2, Marvin Hartwig Zoellner1, Diana Ryzhak1, Detlev Grützmacher2, Dan Buca2, Giovanni Capellini1&3
Affiliations : 1IHP – Leibniz-Institut für innovative Mikroelektronik, Im Technologiepark 25, 15236 Frankfurt (Oder), Germany; 2Peter Grünberg Institute 9 (PGI-9) and JARA-Fundamentals of Future Information Technologies (JARA- FIT), Forschungszentrum Jülich, 52425 Jülich, Germany; 3Dipartimento di Scienze, Università Roma Tre, V.le G. Marconi 446, 00146 Roma, Italy;

Resume : GeSn is a promising group IV material with a direct bandgap that could be most interesting for applications in photonics, microelectronics, or thermoelectrics. Furthermore, it could be potentially integrated into the complementary metal–oxide semiconductor (CMOS) technology. A key feature of this material system is that upon a suitable choice of strain and composition of the epitaxial layer, the band-gap structure of the GeSn can be largely engineered to obtain a direct band gap group IV semiconductor. To this aim, is of paramount importance to correctly evaluate the effect of composition, strain, and the deposition process on the crystal quality of GeSn layers. To this purpose, Raman spectroscopy is an effective experimental technique to determine these properties, as it is non-destructive, contactless, fast and locally resolve technique. In this work, we analyze the polarization dependence of Raman scattering from Ge1−xSnx (0.05 ≤ x ≤ 0.14) alloys grown by chemical vapor deposition (CVD) technique on Ge/Si virtual substrates. Polarized Raman scattering measurements were performed in backscattering geometry along the [001] and [110] directions with parallel and cross polarizations. The analysis of polarization dependence of spectra allowed us to separate unambiguously the different contributions to the Raman spectra, i.e. the main LO Ge-Ge, Ge-Sn, and Sn-Sn modes, as well as disorder-activated secondary multiphonons Ge-Ge modes. Our conclusions were supported by thorough structural characterization of the samples carried out by RBS, XRD, and TEM technics. Furthermore, Raman scattering measurements were performed on the same GeSn films at temperatures ranging from 83 to 325 K to investigate how the thermal-expansion and anharmonic phonon-coupling impact the temperature dependence of Raman modes. Our results help to understand the fundamental properties of GeSn alloys for their applications as light sources, optoelectronic or thermoelectric materials.

R.5.5
15:30 Coffee break    
 
Group IV materials : Monica De Seta
16:00
Authors : Yosuke Shimura1,2,* Masaki Okado1, Junya Utsumi1, Kako Iwamoto1, Ryo Yokogawa3,4, Motohiro Tomita5, Atsushi Ogura3,4, Hiroshi Uchiyama6, Hirokazu Tatsuoka1
Affiliations : 1 Shizuoka University (Japan); 2 Research Institute of Electronics Shizuoka University (Japan); 3 Meiji University (Japan); 4 Meiji Renewable Energy laboratory (Japan); 5 Waseda University (Japan); 6 JASRI (Japan) *Currently at imec (Belgium)

Resume : Formation of Sn-containing group-IV alloys, such as GeSn and SiGeSn which are the attractive candidates for thermoelectric generator, in the form of polycrystal with high Si and Sn contents is challenging due to low Sn solubility although high temperature is required for Si crystallization. Crystalline Sn nanodots were found to mediate the Ge or Si crystallization at low temperature as nuclei, besides, Sn atoms are introduced to the polycrystalline materials to be alloyed with them during the growth. Some results of phonon dispersion measurements at synchrotron will be also introduced if time allows.

R.6.1
16:30
Authors : Andriy Hikavyy, Sydney Eronmhonse, Clement Porret, Roger Loo
Affiliations : Imec, Kapeldreef 75, 3000 Leuven, Belgium: Technological University Dublin, Park House, 191, Co. Dublin, Ireland

Resume : Alternative to Si channel materials (SiGe or Ge) and new device concepts such as 3D transistor stacking, complementary FET (CFET) and gate-all-around (GAA) nanosheet devices, all considered for the upcoming technological nodes of 3 nm and below often require process temperatures < 500oC. Currently, the epitaxial growth of group IV semiconductor materials is used for the Source/Drain (S/D) engineering of both p- and n-MOS high performance transistors, allowing the deposition of SiGe:B and Si:P layers with active carrier concentrations ~ 1x1021 cm-3, ensuring low contact resistivities. Nevertheless, with reducing device dimensions, the growing importance of contact resistance in devices remains a major concern. To meet these temperature requirements, epitaxial processes based on the combination of high order Si and Ge precursors are explored. The strength of this approach has already been demonstrated by the successful implementation of Ge/SiGe stacks and Ge S/D deposited at low temperature for the production of pMOS Ge GAA devices exhibiting excellent electrostatic control at sub-30 nm gate lengths. In this contribution, we use advanced low temperature epitaxial processes to enable highly B-doped SiGe layers grown at temperatures down to 300°C with an active doping concentration exceeding 1x1021 cm-3. We also show that chemical concentration of B can be increased even higher up to few percents of B without deterioration of SiGe layers crystallinity. We will discuss the main properties of such layers and a possibility of their application as a S/D material in the advanced p-MOS transistors.

R.6.2
16:45
Authors : Felix Reichmann(1), Andreas P. Becker(1), Emily V. S. Hofmann(1)(2)(3), Neil J. Curson(2)(3), Wolfgang M. Klesse(1) and Giovanni Capellini(1)(4)
Affiliations : (1)IHP – Leibniz-Institut für innovative Mikroelektronik, Im Technologiepark 25, 15236 Frankfurt (Oder), Germany (2)London Centre for Nanotechnology, University College London, 17-19 Gordon Street, London, WC1H 0AH, UK (3)Department of Electronic and Electrical Engineering, University College London, London, WC1E 7JE, UK (4)Dipartimento di Scienze, Università Roma Tre, V.le G. Marconi 446, I-00146 Rome, Italy

Resume : The growing effort to fabricate low-dimensional GeSn structures for (opto-)electronic applications increases the importance to understand the fundamental surface and interface properties of the material system. Accordingly, we investigate in this work the modification of the electronic structure of the Ge(001) surface after the adsorption and incorporation Sn. We extend a growth-model for the adsorption of Sn on Ge(001), previously established by scanning tunneling microscopy, by utilizing photoemission techniques. We show the k-space resolved valence band structure and reveal Sn induced changes to the Ge valence states, such as the removal of surface states, the modification of effective masses and the creation of a new Sn-related surface state. Post-deposition annealing leads to full incorporation of Sn and, consequently, to the disappearance of valence band state attributable to Sn ad-atoms. Independent of the adsorption and/or incorporation of Sn, we observe that Fermi-level remains pinned close to the Ge valence band maximum, indicating the initial stages of a Schottky barrier formation and allowing us to identify the origin of the strong Fermi-level pinning. Our results provide new fundamental insights into the electronic structure of the system, crucial for the development of SnGe electronics devices, and more generally of use for understanding the controlled alloying of isoelectronic layered materials.

R.6.3
17:00
Authors : A. Salomon, J. Aberl, L. Vuku¨ić, M. Hauser, J. Schuster, T. Fromherz, M. Brehm
Affiliations : Institute of Semiconductor and Solid-State Physics, Johannes Kepler University Linz, Altenbergerstr. 69, A-4040 Linz, Austria

Resume : Current research advances for classical [1] and quantum [2] Si(Ge)-based light emitters raise expectations to finally overcome the restrictions imposed by the indirect bandgap nature of group-IV semiconductors. However, emitter integration in efficient electrically-pumped devices is limited by the absence of lattice-matched compounds that allow confining injected carriers to the emitter-site via type-I double-heterostructures (DHS), as it is state-of-the-art for the III-V (Al)GaAs system. Common knowledge suggests that the type-II band alignment and the limited critical thickness of strained, pseudomorphic, and Ge-rich Si1 xGex on Si(001) layers impede DHS-formation. We present an extensive study of Si1-xGex epilayers with variable Ge contents of 0.35

R.6.4
17:15
Authors : Sebastian Reiter (1), Weijia Han (1), Christian Mai (2), Davide Spirito (2), Josmy Jose (2), Christian Wenger (2), Inga A. Fischer (1)
Affiliations : 1 Experimental Physics and Functional Materials, Brandenburgische Technische Universität Cottbus-Senftenberg, Erich-Weinert-Straße 1, 03046 Cottbus, Germany; 2 IHP—Leibniz-Institut für innovative Mikroelektronik, Im Technologiepark 25, 15236 Frankfurt (Oder), Germany

Resume : Plasmonic excitations in metallic nanostructures have potential applications ranging from enhancing the functionality of optoelectronic devices to integrated biosensors. The optical properties of structures such as plasmonic nanohole arrays are highly sensitive to changes in the refractive index in the vicinity of the structures. The combination of plasmonic nanohole arrays with group-IV optoelectronic devices can be a strategy for the large scale fabrication of miniaturized and cost-effective biosensors on the Si platform. However, complementary metal-oxide-semiconductor (CMOS) fabrication processes place restrictions in particular on the metals that can be used for the fabrication of the structures. Here, we investigate the optical properties of Titanium Nitride (TiN) nanohole arrays fabricated using high-precision industrial fabrication processes for possible applications in integrated, plasmonic biosensors. Reflectance measurements show pronounced Fano-shaped resonances that can be attributed to extraordinary optical transmission through the nanohole arrays. Using the measured material permittivity as an input, the measured spectra are reproduced by simulations to a very large degree of accuracy. Our results show that the material, despite featuring higher losses compared to metals such as Ag or Au, is very promising for applications in on-chip plasmonic refractive index sensors.

R.6.5
17:30
Authors : Roger Loo 1, Nicolas Gosset 2, Megumi Isaji 2, Yumi Ikeda Kawamura 2, Andriy Hikavyy 1, Erik Rosseel 1, Clement Porret 1, Ankit Nalin Mehta 1, and Jean-Marc Girard 3
Affiliations : 1 Imec, Kapeldreef 75, 3000 Leuven, Belgium ; 2 Air Liquide Laboratories, Innovation Campus Tokyo, 2-2 Hikarinooka, Yokosuka, Kanagawa 239-0847 Japan ; 3 Air Liquide Advanced Materials, 75 quai d'Orsay, 75007 Paris (France)

Resume : Forksheet transistors are lateral nanosheet devices with a forked gate structure. The physical separation of N- and PFETs by a dielectric wall enables N-P space scaling and consequently sheet width maximization within the limited footprint of low-track-height standard cells. Bottom dielectric isolation has been proposed to circumvent the junction isolation trade-off between punch-through suppression on the one hand and junction leakage and capacitance on the other hand. A typical fabrication scheme includes the epitaxial growth of Si/Si1-yGey/multi-{Si1-xGex/Si} epi stacks (y>x) where the bottom Si1-yGey layer is later replaced by a SiN/SiCO isolation. A similar approach is followed for complementary FET (CFET) devices where pFET and nFET are stacked on top of each other with a dielectric isolation in between. Both fabrication schemes rely on the selective etching of the sacrificial Si1-yGey layer with respect to the {Si1-xGex/Si} multi-stack. Owing to the very small dimensions (e.g. sub-10 nm nanowire channel diameter), high etch selectivity towards both Si1-xGex and Si, and excellent process controls are mandatory. This sets stringent requirements on the epitaxial stacks (thicknesses and composition control, sharpness of interfaces, and absence of strain relaxation) as well as on the etch process itself (high selectivity, limited Si1-xGex and Si consumption). The selective Si1-yGey removal requires a great precision in its adjustment, with the risk of experiencing process variabilities and yield issues. Selective SiGe etching is typically done in advanced wet or dry chemistries and is sensitive to both strain in and oxidation of individual layers. Also, HCl-based vapor etching has been reported for SiGe removal, with high selectivity towards Si. The HCl vapor etching requires a relative high process temperature (typically ≥ 600C). This makes it less attractive for fabrication schemes with bottom or middle isolations. The presence of the Ge-rich Si1-yGey layer in the as-grown epi stack results in an enhanced risk for unwanted layer relaxation. A low temperature Br2-based vapor etching process is proposed as an alternative for the selective Si1-yGey removal in the isolation fabrication. Using Br2 as etching gas for Si1-xGex etching, in-stead of HCl or HBr, allows to reduce the processing temperature by ~300C. The difference in etching rate as a function of layer composition is the highest for Br2. After initial process screening on blanket epi layers to compare etching behavior for different process gases as function of material composition and crystallinity, it is demonstrated on fin patterned test structures that Br2 etching enables high etch selectivity of 10 nm Si0.5Ge0.5 towards 3x {9 nm Si1-xGex / 9 nm Si} multi-layers (x=0.2, 0.23, and 0.3).

R.6.6
Start atSubject View AllNum.
 
Neuromorphic devices and materials : N/A
14:00
Authors : Qing-Tai Zhao, Fengben Xi, Andreas Grenmyr, Jiayuan Zhang, Yi Han, Jin Hee Bae, Detlev Grützmacher
Affiliations : Peter Grünberg Institute (PGI 9) and JARA-Fundamentals of Future Information Technologies, Forschungszentrum Jülich GmbH, 52428 Jülich, Germany

Resume : A neuromorphic computing system employs a great number of artificial synapses which transfer information between neurons. In this paper we present CMOS compatible artificial synapses based on a ferroelectric Schottky barrier MOSFET on silicon. The ferroelectric polarization switching dynamics gradually modulate the Schottky barriers, thus programming the device conductance by applying positive and negative pulses to imitate the excitation and inhibition behaviors of biological synapse. The three and four terminal configurations of the device enable both homo- and hetero-synaptic plasticity with multi-functionalities, high endurance, low power consumption and high speed. Logic in-memory units, like AND, NAND, XOR, NOR, which are highly demanded in neuromorphic computing, are also demonstrated with the device.

R.7.1
14:30
Authors : Kamila Janzakova, Ismael Balafrej, Ankush Kumar, Nikhil Garg, Sebastien Pecqueur, Yannick Coffinier, Fabien Alibart
Affiliations : Institut Of Electronic, Microelectronic and Nanotechnology (IEMN), UMR 8520-CNRS, Université de Lille, Villeneuve d’Ascq, France Laboratoire Nanotechnologies Nanosystemes (LN2), IRL 3463-CNRS, Université de Sherbrooke, Sherbrooke QC, Canada

Resume : The last decades have been marked by a shift of paradigm in computing. From generic hardware and specialized algorithms, current approaches are relying more and more on specialized hardware with much more generic algorithms. This is particularly true in the context of Artificial Intelligence where implementing generic algorithms of Artificial Neural Networks requires a specialized hardware in order to keep the energy consumption low. Nevertheless, standard fabrication approaches are largely top-down and rely on deterministic hardware where all components and their organization needs to be known a priori. At the opposite, biology is relying on a bottom-up strategy where elementary components assemble and evolve toward the objective function during development resulting on optimal resources utilization. In this talk, we will present a strategy that can reproduce the bottom-up approach of biology for implementing meaningful computing functions. We propose to use bipolar AC electropolymerization to engineer dendritic-like fibers of PEDOT:PSS. We will show how such unconventional structures can implement various neuromorphic concept such as structural plasticity and synaptic plasticity of biological neural networks. Finally, we will show how this approach can be used to find optimal topologies of neural networks for tasks such as classification and signal reconstruction.

R.7.2
15:00
Authors : Swayam Prakash Sahoo[a,b,c], Alexandre Juneau‑Fecteau[b,c], Anne Lamirand[a], Victor Pierron[d], Laurence Méchin[d], Luc G. Fréchette[b,c], Bertrand Vilquin[a,b]
Affiliations : [a]. Univ Lyon, Ecole Centrale Lyon, INSA Lyon, UCBL, CPE Lyon, CNRS, Institut des Nanotechnologies de Lyon, UMR5270, 69130 Ecully, France [b]. Institut Interdisciplinaire d’Innovation Technologique (3IT), Université de Sherbrooke, Sherbrooke, QC J1K 0A5, Canada [c]. Laboratoire Nanotechnologies Nanosystèmes (LN2) - CNRS UMI-3463, Université de Sherbrooke, Sherbrooke, QC J1K 0A5, Canada [d]. Normandie Univ, UNICAEN, ENSICAEN, CNRS, GREYC (UMR 6072), 14000 Caen, France

Resume : Neuromorphic computing is being seen as a solution to address the memory bottleneck persistent with the present computing paradigm. Artificial synapses and neurons need to be built to realize such an architecture. One way to emulate a bio-synapse requires a material with a metal-insulator phase transition (MIT). VO2 undergoes a structural phase transformation (SPT) from a monoclinic structure at room temperature to tetragonal at approximately 70°C. The SPT is accompanied by an MIT leading to a large variation in its electrical (about 4 orders of magnitude of its resistivity) and optical properties, in particular, in its complex refractive index in the mid-IR frequency range. To keep with the current trends of the microelectronic industry, it is imperative to integrate VO2 on silicon. However, the higher lattice mismatch and formation of oxides and silicates at the interface between VO2 and crystalline Si degrade the quality and functionality of VO2 film. Additionally, VO2(M1) is a challenging material to integrate into patterned heterostructures because it can exist not only as multiple polymorphs (A, B, M1) but the high-temperature depositions can lead to the formation of various oxidation states phases that are present in the V-O system (VnO2n-1, VnO2n 1). This work was conducted to study the growth of VO2 on silicon with oxide buffer layers using RF magnetron sputtering of a V2O5 ceramic target in an argon atmosphere. We studied the structure-property relationships, specifically electrical and optical properties as a function of temperature across the Tc. Structural and compositional characterization are carried out using x-ray diffraction, atomic force microscopy, and x-ray photoemission spectroscopy respectively, optical responses are studied under spectroscopic ellipsometry and electrical characterizations are performed using the four-point probe method. With the use of a very thin metal oxide buffer layer between the silicon substrate and VO2 film, we demonstrate a high resistivity ratio (of the order 3 between the two phases) and investigate the scope of improvement. The results show the influence of substrate temperature, VO2 grain size, and strain on it as well as the crystal structure of the buffer layer on the structural and physical properties of interfaces and film morphology which subsequently affect the electrical bistability of VO2. The preliminary findings mentioned here are being utilized to improve the electrical bistability, thus allowing us to improve the reproducibility in operational modes (switching, memory, logical operations, etc.) of neuromorphic devices.

R.7.3
15:15
Authors : H.R.J. Cox1*, M. Buckwell2, W.H. Ng1, D.J. Mannion1, A. Mehonic1, P. R. Shearing2, S. Fearn3 & A.J. Kenyon1
Affiliations : 1: Department of Electronic and Electrical Engineering, University College London, Torrington Place, London WC1E 7JE, UK; 2: Department of Chemical Engineering, University College London, Torrington Place, London WC1E 7JE, UK; 3: Department of Materials, Imperial College London, South Kensington Campus, London SW7 2AZ, UK

Resume : A promising route to neuromorphic systems is using resistance switching in oxide-based, intrinsic ReRAM devices. During operation, under an applied field, oxygen is driven across the metal-oxide-metal (MIM) stacks modulating the resistance between the electrodes. The electrode is thought to act as an oxygen reservoir enabling the reversible exchange of oxygen with the high resistance oxide. How effective the electrodes are as a reservoir is a significant factor in determining the cycling endurance and stability of devices. Here we report the measurement of oxygen movement in silicon oxide-based ReRAM devices using a novel Secondary Ion Mass Spectrometry (SIMS) normalisation technique that allows the measurement of ionic movement with unparalleled sensitivity. Our technique allows us for the first time to observe the movement of 16O across electrically biased silicon oxide (SiOx) ReRAM stacks, measuring bulk concentration changes in a continuous profile with unprecedented sensitivity. Applying this to three devices, using different electrode materials, we systematically examine the exchange of oxygen across the oxide-metal interface. A clear link is revealed between the thermodynamics and microstructure of the electrode material with the scale and nature of this diffusion across the oxide-metal interface. Where the electrode is unable to accept oxygen, delamination and breakdown of the device is observed. Where the electrode acts as a reservoir we see a stable and reversible

R.7.4
15:30 Coffee break    
16:00
Authors : Saverio Ricci (1), David Kappel (2), Christian Tetzlaff (2), Matteo Farronato (1), Alessandro Milozzi (1), Stefan Slesazeck (3), Thomas Mikolajick (3), Erika Covi (3), Daniele Ielmini (1)
Affiliations : (1) Politecnico di Milano, Milano, Italy. (2) University of Göttingen, Göttingen, Germany. (3) NaMLab gGmbH, Dresden, Germany.

Resume : Due to the rapid development of artificial intelligence and internet of things, neuromorphic computing and brain-inspired architectures that mimic the cognitive functions of the brain in hardware have become a primary challenge. In this scenario, it is crucial to find suitable elementary units for both performing neuromorphic functions and fabricating devices in large scale. The volatile resistive switching memories (RRAMs), which feature spontaneous change of device conductance, own the distinct combination of high similarity to biological neurons and unique physical mechanisms. They rely on the formation and disruption of a metallic conductive filament (CF) with a lifetime ranging from few microseconds to several seconds, thus by controlling and predicting the CF lifetime, devices can be engineered for a wide range of applications, such as non-volatile memory for data storage, tunable short/long term memory for synaptic neuromorphic computing, and online-learning for edge-computing. Here we present Ag/HfO2/C RRAMs with 1-transistor and 1-resistor (1T1R) structure and 10 nm of oxide layer. The devices start from a high resistive state (HRS) higher than 1 TΩ and switch to a low resistive state in the range of 1 KΩ. The retention time of the CF can be easily tuned by changing the transistor gate voltage and thus the maximum current that can flow inside, showing that the retention time increases with the maximum current, going from few milliseconds to seconds. The set event is dependent on the voltage applied across the devices, revealing that the higher the voltage the bigger the switching probability. The combination of the stochastic behavior of these devices and the tunability of such processes make them perfect candidates to implement short-term memory functions. Here we present a simple full-memristive architecture able to store and distinguish ten patterns, associated to different items, with an overall retention capability in the order of several hundred of milliseconds. Different experimental conditions have been tested, in terms of stimulation voltage (linked to the switching probability) and spike rate (related to the retention time stochasticity of the RRAMs), to deeply analyze the network response and compare it with biological systems. By decreasing the spike rate from 50 to 10 Hz the system starts forgetting the stored item, since the retention time is comparable or smaller than the spike distance. On the other hand, a high switching probability (20% probability to have a set with a certain voltage) makes the system saturated and unable to store and distinguish any more the items. The best condition was reached working with small switching probability (5%) and fast spike rate (50Hz), in perfect agreement with the biological mechanisms that happen in the brain.

R.7.5
16:15
Authors : Pradheebha Surendiran1, Christoph Robert Meinecke2, Aseem Salhotra3, Georg Heldt2, Jingyuan Zhu1, Alf Månsson3, Stefan Diez5, Danny Reuter2, Hillel Kugler4, Heiner Linke1, Till Korten5
Affiliations : 1NanoLund and Solid State Physics, Lund University, Lund, Sweden; 2 Department of Nano Device Technologies, Fraunhofer ENAS, 09126 Chemnitz, Germany; 3 Department of Chemistry and Biomedical Sciences, Linnaeus university, Kalmar, Sweden; 4 Faculty of Engineering, Bar‐Ilan University, Ramat Gan, Israel; 5 B CUBE - Center for Molecular Bioengineering, Technische Universität Dresden, D-01307 Dresden, Germany

Resume : Computational problems of combinatorial nature including non-deterministic polynomial-time (NP) complete problems currently require exponential time to explore the solution space as the problem size increases, making conventional serial computation intractable, and parallel computation a necessity. Network-based Biocomputation (NBC) was demonstrated to solve the subset sum problem (SSP) [1], by encoding it into a graphical network of channels in a nanofabricated device, which is then explored by cytoskeletal filaments propelled by molecular motors to find all possible solutions. This approach of parallel computation promises to require orders of magnitude less energy than conventional computers due to high energy efficiency of the molecular motors. The current study focusses on using NBC to solve another problem, namely Exact Cover (ExCov). For an ExCov problem with a collection X of subsets with elements from a set Y, an exact cover is a subcollection X* such that each element in Y is contained in exactly one subset in X*. We demonstrate an algorithm [2] to translate an ExCov problem with 32 potential solutions into an SSP network which is then solved by using a molecular motor system (actin-myosin II). This work demonstrates that the NBC approach can be used for directly encoding more than one combinatorial problem, and upscaling of the solution space that can be solved by NBC by a factor of four compared to previous work [3]. We have recently detailed the challenges and requirements for further upscaling of NBC [4]. In future, our approach could be applicable in solving real world ExCov problems like airplane fleeting, designing electric circuits and solving puzzles like Sudoku and Rubik’s cubes. References [1] Nicolau, Dan V., et al. "Parallel computation with molecular-motor-propelled agents in nanofabricated networks." PNAS 113.10 (2016): 2591-2596. [2] Korten, Till, et al. "Design of network-based biocomputation circuits for the exact cover problem." New Journal of Physics 23.8 (2021): 085004. [3] Surendiran, Pradheebha, et al. “Solving Exact Cover Instances with Molecular-Motor-Powered Network-Based Biocomputation.” ACS Nanoscience Au (2022) (Accepted manuscript). [4] Zhu, Jingyuan, et al. "Physical requirements for scaling up network-based biocomputation." New Journal of Physics 23.10 (2021): 105004. Keywords: parallel computation, molecular motors, biocomputation, NP-complete Funding: This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 732482 (Bio4Comp).

R.7.6
16:30
Authors : Upanya Khandelwal, Rama Satya Sandilya, Sushobhan Avasthi, Saurabh Chandorkar, Pavan Nukala
Affiliations : Indian Institute of Science, Bengaluru, Karnataka, India

Resume : Strongly Correlated Materials such as NbO2, Vanadium Dioxide(VO2) exhibit insulator-to-metal transition (IMT) with increasing temperature. This transition can be electrically achieved via Joule heating, leading to threshold switching and Negative Differential Resistance (NDR) regions in the I-V characteristics of nano-and micro-scale devices. Devices operating within the NDR region can self-oscillate. Since VO2 has an IMT of 340K which is just above room temperature, the power expended in generating the self-oscillations is relatively low. The nature of these oscillations were shown to be periodic, stochastic, or chaotic by various authors. Here we demonstrate a transition between periodic oscillations to clustered oscillations in defective VO2 (100 nm) thin-films grown epitaxially on sapphire. The voltage-controlled I-Vs show a volatile hysteresis, but current-controlled I-V profiles exhibit a rugged and complex NDR profile (unlike the smooth S-shape NDRs observed by others), that can be correlated to the defects in the VO2 thin-film. With a load resistance in series, at low pulse voltages, the devices showed a simple capacitive charging and discharging. Between a threshold voltage (Vth) and a transition voltage (Vt), we observe periodic current oscillations (~MHz), whose frequency increases with increasing voltage. These properties of oscillations are similar to the simple integrate-and-fire model in neurons. At voltages beyond Vt, the device characteristics change, showing clusters of oscillations. These clustered oscillations closely resemble tonic bursting in neurons, and generating such oscillations from low-power devices is significant for neuromorphic computation. High amplitude oscillations disappeared on further increasing voltage, and the device settled to a particular state. Along with the electrical oscillations, we simultaneously observed mechanical oscillations. Interestingly, the frequency of electrical and mechanical oscillations is precisely the same, which predicts the phase transition is the cause of mechanical and electrical oscillations. This work also provides material design guidelines for generating complex NDR regions and thus clustered oscillations.

R.7.7

No abstract for this day


Symposium organizers
Abderraouf BOUCHERIFUniversity of Sherbrooke

2500 Bd de l'Université - Sherbrooke, QC J1K 2R1, Canada

Abderraouf.Boucherif@usherbrooke.ca
Andriy HIKAVYYIMEC

Kapeldreef 75, 3001 Leuven, Belgium

Andriy.Hikavyy@imec.be
Cezar ZOTAIBM Research GmbH

Saumerstrasse 4, Ruschlikon, Switzerland

zot@zurich.ibm.com
Monica DE SETADept. of Science University Roma Tre

Via della Vasca Navale 79, 00146 Roma, Italy

monica.deseta@uniroma3.it