Phase-Change Materials and Switches for Enabling Beyond-CMOS Energy Efficient Applications

Projektinformationen

PHASE-CHANGE SWITCH

ID Finanzhilfevereinbarung: 737109

DOI

10.3030/737109

Projekt abgeschlossen

EK-Unterschriftsdatum 29 November 2016

Startdatum 1 Januar 2017

Enddatum 30 November 2020

Finanziert unter

• EXCELLENT SCIENCE - Future and Emerging Technologies (FET)

Gesamtkosten

€ 3 883 412,50

EU-Beitrag

€ 3 883 412,00

Koordiniert durch

ECOLE POLYTECHNIQUE FEDERALE DE LAUSANNE
Switzerland

Periodic Reporting for period 2 - PHASE-CHANGE SWITCH (Phase-Change Materials and Switches for Enabling Beyond-CMOS Energy Efficient Applications)

Berichtszeitraum: 2019-01-01 bis 2020-11-30

Zusammenfassung vom Kontext und den Gesamtzielen des Projekts

The project PHASE‐CHANGE SWITCH addresses the need for combined energy efficiency and extended functionality with the engineering of new classes of solid‐state Beyond CMOS switches exploiting the abrupt phase‐change (Metal‐Insulator‐Transition ‐ MIT) in materials and at
temperatures that make them interesting for electronic circuits and systems by their performance and scalability. The project includes disruptive research contributions on the whole value chain, from novel phase‐change materials to new device and
circuit architectures together with their scaling and integration on silicon and GaN platforms. Materials alloying and straining techniques in phase‐change systems are used for the engineering of the transition temperature and the ON and OFF band gaps (conductivity) of VO2.

The project focuses on smart design and exploitation of the unique properties of the phasechange VO2 beyond CMOS switches, by targeting with the same technology platform the following objectives:
(i) von‐ Neumann steep‐slope logic devices and circuits, to extend CMOS with novel functionality and energy efficiency,
(ii) uniquely reconfigurable energy efficient radio‐frequency (RF) circuit functions from 1 to 100GHz,
(iii) unconventional scalable neuristors exploiting the hysteretic RC switching behaviour for neuromorphic computation, and,
(iv) disruptive classes of solid‐state devices for neuromorphic computation, exploiting non‐volatile memory effects.

The project is expected to create new applications and markets and reinforce the leadership of European industrial players in the field of energy efficient Internet‐of‐Things and high frequency communications.

Arbeit, die ab Beginn des Projekts bis zum Ende des durch den Bericht erfassten Berichtszeitraums geleistet wurde, und die wichtigsten bis dahin erzielten Ergebnisse

WP1: The PHASE-CHANGE SWITCH project is based on the use of Vanadium dioxide (VO2) phase change material layers deposited by three different technologies: i) Reactive magnetron sputtering deposition; ii) Pulsed laser deposition; iii) Atomic layer deposition. As each of this deposition methods has distinctive advantages and drawbacks all three are explored in detail in WP1 to identify deposition processes suiTable 1. to fulfil the requirements posed by the three applications targeted in WP2 (Steep-slope switches for beyond-CMOS logic), WP3 (Reconfigurable RF applications) and WP4 (Neuromorphic circuits) while materials modelling is carried out in WP5. FoMs have been derived and target values have defined. Deposition techniques to fullfill these targets have been successfully developed.
WP2: two-terminal [2T] VO2 switches were optimized for steep-slope switching and for oscillating neural networks. The 2T VO2switches are the core elements of the oscillators for neural networks. Oscillators have been fabricated and their operational characteristics were analysed. A maximum frequency of 105 Hz was achieved, limited by the parasitic capacitance of the test setup. 3T hybrid VO2 oscillators, consisting of a MOS transistor and a 2T VO2 switch, were fabricated and tested. The 3T oscillators offer the advantage that the oscillation properties can be controlled more effective. The device shows oscillation below 0.5 V with a power supply of 1 V, and power consumption of 20 μW.
WP3: CMOS compatible HR Si reconfigurable phase shifters, defected ground plane and multiple split ring filters, rectangular and Peano inductors were fabricated and proposed. The footprint of the proposed devices and their reconfigurability range improved the state the art.
GaN VO2 switches with handling power of 40 W were demonstrated. Switching times of 35 ns were measured and a reliability of 109 cycles was proven.
WP4: The main achievements of the WP were:
• the experimental demonstration of the pattern recognition
• Characterization of materials for ionic liquid gating
• The development of a Recipe for synthesis of ionic gels
WP5: It was shown that: Local Density Approx (LDA)+U gives surprisingly good description of VO2’s electronic structure, total energy, and alloying dependence, giving:
• The band gap for M1 phase
• Needs correct description of spins in R, M1
• gives Latent heat – R – M1 energy difference
• Calculates TC
• Gives alloying/doping dependence.

Fortschritte, die über den aktuellen Stand der Technik hinausgehen und voraussichtliche potenzielle Auswirkungen (einschließlich der bis dato erzielten sozioökonomischen Auswirkungen und weiter gefassten gesellschaftlichen Auswirkungen des Projekts)

(1) In VO2 material processing, optimization and integration:
• Material (VO2 and Ge-VO2) Figures of Merit (FoM) defined and updated in multiple iterations for all target applicationsby partner AMO
• The final VO2 film quality on Si and GaN platforms was finally below the level achievable on TiO2 substrates.
o High quality functional layers within the specifications defined by the FoM by EPFL and UCAM
o ALD deposition has been developed to match sputtered growth, and understanding of them will match other methods by UCAM
o first of their kind VO2 layers with <20 nm grain size deposited by ALD. Here, flash anneal process provides control of roughness and grain size below 20nm by partners UCAM and IBM
o First ever reported MIT temperature of >93°C by Ge doping and by preserving high on/off conductivity ratio by partner EPFL
(2) in Steep slope MIT logic solid-state switch and logic applications:
• Steep slope switch with sub-10mV/decade slope fully demonstrated by partner EPFL.
• Experimental demo of 4 coupled oscillators for neuromorphic computing has been successfully by partner IBM.
• Experimental demo of pattern recognition by partner IBM
• 10x increase of oscillation frequency: 3 MHz by partner IBM
(3) In Phase change RF switch and reconfigurable RF functions:
• Switching times of 35 ns were obtained by partner TRT for off/on transitions and power handlings of 40 W, both on GaN.
• VO2 was integrated by partner EPFL on HR Si substrates for a variety of filtering functions phase shifters, resonators, inductors, beyond the state of the art.
• Additionally, frequency dependent scalar-vectorial modelling/ characterization instruments were proposed.
(4) in Energy efficient neuromorphic circuits:
• Experimental demonstration of coupled VO2 oscillators following the goals set at the beginning of the project.
• Non-volatile memory devices based on ionic-liquid gating of VO2 were experimentally fabricated and characterized by partner MPG. Material choice, device scaling and architecture limited energy and speed consideration.
(5) in Modelling and simulation:
• Advance in understanding pf electronic structure of semiconducting M1 state by partner UCAM
• Ability to predict effect on Tc and M1 band gap versus alloying composition was achieved by partner UCAM
• Strain was not finally investigating as the project set a high priority on the Ge-doping that was in the centre of the technological platform.
(6) In socio-economic impact and exploitation:
The project produced 25 high level publications in high level journals and conferences
The project achieved excellent technical results that have been considered by the partners for specific exploitation actions, as it follows:
• IBM started new neuromorphic EC projects targeting circuit level developments.
•Thales is considering further industrial exploitation of VO2 GaN switches for applications in airborne system design.
• EPFL plans to create a spinoff in 2021.
vAMO’s experience is very successfully exploited in successfully applying and being funded for the cluster “NeuroSys – Neuromorphic Hardware for Autonomous Artificial Intelligence Systems” that will receive €45 million funding from the Federal Ministry of Education and Research.
https://www.amo.de/blog/2021/02/03/neurosys-succeeds-in-the-first-clusters4future-ideas-competition/

https://actu.epfl.ch/news/a-revolutionary-material-for-aerospace-and-neuro-8/