Funding type: COFUND Marie Slodowska Curie Action

This project is under the Marie Skłodowska-Curie Actions (MSCA) program. There are no nationality conditions but the candidates must fulfill the MSCA mobility conditions, which means that she/he must not have stayed more than 1 year in France during the last 3 years immediately before the call deadline (31/05/2019)

Expected start date: 01/10/2019

Contacts:

Ass. Prof. Bertrand VILQUIN
This email address is being protected from spambots. You need JavaScript enabled to view it.
+33 4 72 18 62 54

Pr. Arnan Mitchell, RMIT
This email address is being protected from spambots. You need JavaScript enabled to view it.

Websites:

• https://www.rmit.edu.au/about/our-locations-and-facilities/facilities/research-facilities/micronano-research-facility
• http://inl.cnrs.fr/en/

Collaborations/External partners : STMicroelectronics

Domain and scientific context :

After more than 40 years of continuous evolution, our computing systems are reaching their limits. Indeed, the architecture of Von-Neumann, on which our computers are based, physically dissociates the hearts of calculations from the memory. The sequential processing of information is thus confronted with a bottleneck, more commonly known as "Memory Bottleneck". [1] One solution is to draw inspiration from the natural mathematical paradigms of the human brain, in which the data are massively parallel processed with high energy efficiency, realizing the hardware implementation of neuromorphic networks. The latter make it possible to bring the information storage sites (synapses) closer to the treatment sites (neurons). The major challenge of this bio-inspired approach is the realization of dense networks of artificial synapses to implement synaptic plasticity mechanisms. [2]

Figure 1. Overview of different technological options for performing artificial synapses and project positioning.

Keywords : biomimicry, memristors; articipal synapses; neuroporphic networks; functional oxide

Objectives : 

Theorized in 1971 by Leon Chua (Berkeley) then demonstrated by HP in 2008, memristors are the most promising approach to implement synaptic plasticity thanks to their numerous intermediate states of resistance (Figure 1). [3] Since then, various works have aimed to develop memristors of ReRAM (Resistive RAM) type. However, many debates remain today about the exact origin of their operation or their transferability in an industrial environment. [2] Very recently, Ferroelectric Tunnel Junction (FTJ) type memristors have also been shown to have synaptic learning capabilities. [4] In this context, Shanghai and Berkeley University have just demonstrated the feasibility of FTJ synapses based on zirconium-doped hafnium oxide (HfZrO2 or HZO, see Figure 2). [5] This material has been widely adopted in the microelectronics industry for more than 10 years, for the realization of transistor gate stacking, these results suggest that HfZrO2-based FTJ memristors are an ideal solution for implementing artificial synapses that are efficient and transferable in the medium term.

Figure 2. (a) Ferroelectric Tunnel Junction (FTJ) synaptic matrix. (b) TEM cut of the HZO stack and (c) demonstration of the synaptic plasticity of the FTJ synapses. According to [5]

As part of this approach, this work consists of making artificial synaptic arrays by producing FTJ type memristors based on doped hafnium oxide. The main objectives of this thesis concern (i) the development of ferroelectric tunnel junctions capable of emulating synaptic plasticity, (ii) the realization of synaptic matrices according to industrially transferable methods and (iii) the evaluation of learning capacities. of the synaptic network integrated into a prototype event circuit for emulating a network of impulse neurons.

Scientific challenges :

There are no man-made ferroelectric artificial synapses in the microelectronics industry to date. The aim of this thesis is to achieve the fabrication of such a device from transferable materials and processes in the semiconductor industry.

• The deposition of the ferroelectric material must respect the CMOS processes, in particular at low thermal budget (<450 ° C).
• Since HZO is a recent material, the influence of its thickness or stoichiometry on ferroelectric properties is still poorly known today. This knowledge is essential in order to optimize the operation of the targeted devices.

Expected original contributions :

• Development and demonstration of artificial synapses based on doped hafnium oxide (HZO) using CMOS compatible methods
• Identification of the optimal deposition conditions (i.e. thickness, stoichiometry, temperature, etc.) of the functional oxide for the targeted application
• Realization of integrated synaptic matrices of "Crossbar" type
• Prototyping and testing the learning capabilities of the synaptic network

Research program and methodology :
The thesis aims to develop and integrate very thin layers of functional oxides to develop a new class of nanodevices, transferable in an industrial environment. To do this, the thesis will take place between INL and RMIT. The work will be based on the following 3 tasks:

• T1 Ferroelectric synapses engineering: This task will be dedicated to the realization of ferroelectric artificial synapses. The layers of HZO will be deposited at the INL by PVD and ALD according to different technological variants (doping, thickness, annealing, ...). In articulation with T2, the objective is to lead to functional synapses.
• T2 Physical and electrical characterization: Stackings will be analyzed to evaluate their microstructural quality (TEM, EDX, XPS ...) to guide their optimization. The evaluation of synaptic learning abilities will be evaluated at the INL through different analysis techniques (measurements by Piezo-Force Microscopy, Conductive-AFM and under spikes). Through electrical and microstructural analyzes, the objectives concern the demonstration of synaptic learning and result in optimized FTJs.
• T3 Prototyping and testing of synaptic matrices: In order to achieve a synaptic matrix, the realized FTJ devices (T1 + T2) will be integrated within crossbar matrices. This development will be jointly carried out by RMIT and INL through the exchange of plates integrating lithographed electrode networks. The goal is to achieve functional FTJ matrices at the end of the thesis.

Scientific supervision:

Description of the supervision committee :

Integration inside the laboratories (percentage of working time inside these laboratories) : 67% at INL, 33% at RMIT

PhD funding : Co-Fund Marie Sladowska Curie Action (MSCA) ECL/RMIT (ECLAUsion program)

Profile of the candidate :

S/he should work towards his/her Masters/honours or Engineering degree in a field apposite to one of these areas: Physics ; Electronic Engineering; Material Science and Engineering. An experience in clean-room fabrication, material deposition or electronic characterization will be strongly appreciated.

Objectives for the valorization of the research work :

The results obtained can be valorized through communications in the field of materials science (Advanced Materials, ACS Nano ...), and nano-electronic devices (IEEE, AIP, Elsevier ...). Furthermore, dissemination of the work at international conferences dedicated to materials and devices for neuromorphic engineering will also be carried out (eMRS, IEEE-ESSDERC, IEEE-IMW, ...). The targeted valorisation objectives are from 2 to 5 peer-reviewed international journals and at least 3 papers in international conferences.

Skills that will be developed during the PhD :

The candidate will be trained in the fabrication of materials in the form of thin layers. He / She will also receive training in clean room steps for the realization of simple electrodes and devices. His working environment will allow him to acquire knowledge in electrical characterization of materials (current - voltage, capacitance - voltage measurements). He / she will be integrated into the international nanoelectronics community, the national functional oxide community (OXYFUN) and the international ferroelectric thin film scientific community.

Professional opportunities after the PhD:

The training and formation received during the thesis will allow the PhD student to apply for an academic position in all the laboratories involved in this field or to seek a private job in research and development departments of large companies that take advantage of this type of materials (Thales , STM, IBM ...).

Bibliographic references about the PhD topic :

[1] M. Khan et al., Proc. of IEEE W. Congr. On Comp. Int. (2008)
[2] G. Burr et al., Adv. In Phys :X, 2(1), p.89 (2017)
[3] D. Strukov et al., Nature, 453, pp.80-83 (2012)
[4] S. Boyn et al., Nat. Comm., 8, 14736 (2017)
[5] L. Chen et al., Nanoscale, 10(33), 15826 (2018)
[6] S. Martin et al., Rev. Sci. Instr. 88, 023901 (2017)
[7] STM to launch AI Chip in 2019, eeNews Europe, Nov. 2018