Superconducting Magnetic RAM for Next Generation of Supercomputers

Informazioni relative al progetto

EuSuper

ID dell’accordo di sovvenzione: 796603

DOI

10.3030/796603

Progetto chiuso

Data della firma CE 21 Marzo 2018

Data di avvio 15 Novembre 2018

Data di completamento 14 Novembre 2020

Finanziato da

• EXCELLENT SCIENCE - Marie Skłodowska-Curie Actions

Costo totale

€ 164 203,80

Contributo UE

€ 164 203,80

Coordinato da

CONSIGLIO NAZIONALE DELLE RICERCHE
Italy

Periodic Reporting for period 2 - EuSuper (Superconducting Magnetic RAM for Next Generation of Supercomputers)

Periodo di rendicontazione: 2019-11-15 al 2020-11-14

Sintesi del contesto e degli obiettivi generali del progetto

The continuous search for computing-power and data-storage is quickly approaching the physical limits of conventional silicon-based electronics. To overcome these limits, different approaches beyond conventional CMOS technology are presently under investigation. On one hand, quantum computers, based on non-classical superposition of logic units (bit), offer bright perspectives. On the other hand, energy-efficient superconducting circuits based on Josephson junctions have already demonstrated a computational speed 500 times larger than conventional CMOS-based ones. The complete implementation of a supercomputer based on this technology is limited by the lack of memories operating at cryogenic temperatures.
In this context, EuSuper aims at developing a new generation of nano-sized superconducting non-volatile magnetic memories, with improved efficiency and enhanced functionalities. This is accomplished by : 1) exploiting a wise hybridization between ferromagnetic insulators (FI) and conventional superconducting (S) metals; 2) controlling/tuning the superconducting condensate by an external electrostatic field.
The peculiar behavior of FI/S systems is determined by interfacial quantum phenomena arising at the boundary between the ferromagnetic and superconducting materials. Within a distance from the interface of the order of the superconducting coherence length, the exchange interaction of the FI induces a spin split of the density of states into the S, as per an effective Zeeman splitting generated by an external magnetic field of up to few Tesla (magnetic proximity effect). Besides, the recently demonstrated gate-induced tuning of the superconducting order parameter of fully metallic superconductors may play a key role and introduce new paradigms that ultimately add novel unconventional functionalities and physical insights to S and FI-based nanodevices.
The achievements of this project pave the way for innovative superconducting spintronic applications, i.e. in classical large-scale supercomputing, suitable in all fields of science (solid state physics, artificial intelligence, cryptography, etc.) where increasing speed of calculation and storage are exponentially increasing on demand. Also, from a fundamental physics point of view, the results obtained within the project help in clarifying the interplay between the electrostatic fields and the superconducting condensate as well as the ferromagnetism and superconductivity in FI/S mesoscopic devices and spin-valve. From the technological side, EuSuper counts on small size and scalability of systems to develop an innovative class of superconducting memories ready for the market of incoming next-generation cryogenics supercomputers.
EuSuper is a synergic MSCA that combines the capabilities and expertise of two of the most outstanding and globally recognized research institutions: Massachusetts Institute of Technology (USA) and Consiglio Nazionale delle Ricerche (Italy).

Lavoro eseguito dall’inizio del progetto fino alla fine del periodo coperto dalla relazione e principali risultati finora ottenuti

EuSuper is a powerful playground in which state-of-the-art FI/S physics merges with the experienced researcher proven background in ferromagnetic/superconductor physics and expertise in state-of-the-art nanofabrication, aiming at the investigation and exploitation of novel physics in EuS/Al and GdN/NbN heterostructures.
The research objectives (ROs) of the project were the following: i) Growth and characterization of state-of-the-art FI/S-based thin-film heterostructures, (RO1); ii) Miniaturization of FI/S-based hybrid heterostructures, (RO2). iii) Nanoscale engineering of novel FI/S-based ground-breaking superconducting cryogenic RAM prototype (Fig. 1), (RO3).
I successfully achieved the main points of the ROs by focusing all the efforts on the growth and characterization of EuS/Al and GdN/NbN bilayers and more complex stacks (exchange-coupled Josephson junctions). I was also able to develop two specific reliable top-down nanofabrication approaches, especially designed for S and FI/S multilayers. These achievements represent the basis of the miniaturization of S bridges and FI/S mesoscopic heterojunctions, starting from the simplest well-established configurations towards more complex structures. All the above-mentioned accomplishments paved the way towards the miniaturization of the FI/S-based superconducting spin-valves. This will ultimately lead to the fabrication and commercialization of the first reliable nanometric superconducting non-volatile RAM cell, which feasibility was widely demonstrated and published in high impact factor peer-reviewed articles during the return phase at NEST-CNR.
Besides the experimental work, I gave informal/internal seminars and mini-courses (for students and postdocs), which allowed me to transfer my knowledge to the outgoing/return hosts, and enabled the host groups and I to engineer/optimize the integration of their FI/S structure growth recipes to the nanofabrication processes.

Progressi oltre lo stato dell’arte e potenziale impatto previsto (incluso l’impatto socioeconomico e le implicazioni sociali più ampie del progetto fino ad ora)

The achievements of the overall project provide a strategic step forward towards the development of the novel physics emerging from the precise control of the nanoscale interaction between the ferromagnetism/superconductivity and electric field/superconductivity. The outcomes of EuSuper are able to revolutionize the standard approaches in superconducting spintronics by applying the ground-breaking nanofabrication and experimental methods developed within the project, thus opening new opportunities for developing and commercializing novel magnetic/electric field-driven superconducting nanodevice concepts that represent the essential ingredients for large scale applications in high-performance superconducting computing technology as a real alternative to existing power-hungry computers based on conventional silicon semiconductor technology. During its development, EuSuper has produced results that can be widely considered as beyond state-of-the-art literature and technology. In particular, the achievements reached in EuS/Al miniaturization may be considered at the basis of novel device concepts for fundamental research and applications within the project development (i.e. field-effect superconducting magnetoelectric devices). EuSuper also exploited its own potential even in terms of side results. A significant discovery may revolutionize the technology based on superconducting NbN nanocircuits. In particular, a large superconducting critical current enhancement (~ 30%) and hysteretic infinite electro-resistance induced by an electrostatic field were successfully demonstrated in NbN nanobridges. This first-generation of superconducting field effect transistor (SuFET) is of fundamental interest and become highly attractive for the all-metallic superconducting cryogenic-nanoelectronics. Similar results were also obtained in proximity coupled NbN/ferromagnetic-insulator nano-bridges, which could easily lead to an additional novel superconducting spintronic device: a triplet paired SuFETs. Our gating experiments were also performed in fully suspended superconducting Ti nanobridges. Such devices showed a reliable full suppression of the critical current under applying a large enough external electrostatic field. This latter evidence itself represent a ground-breaking achievement for the overall physics of superconductors. All abovementioned results, made the final goal of EuSuper more intriguing: the superconducting RAM may become more versatile, tunable, and performant than hypothesized in the project proposal.

Sketch of the prototype of the superconducting non-volatile RAM cell.