#25-04Memristive Nanoswitches for Hardware-based Artificial Intelligence
Addressed topics: Micro- and Nanoelectronics, Nano technology
Master-Thesis Diploma Thesis
The growing importance of the artificial intelligence in almost all areas of life necessitates the development of new hardware concepts that are more closely based on the function of the biological brain. The science, which researches the replication of biological components, synapses and neurons with nanotechnological functional elements, is called as “Neuromorphic computing”. Artificial synapses are typically designed in the form of so-called crossbar arrays, Figure 1, which contain nanoionic memory cells in the individual nodes.
These memory cells are known as “memristors”, which are resistors that change in a time-dependent and non-volatile manner depending on the voltage applied. In this way, synaptic weights can be stored. In order to be able to train the network architectures created in this way, “on-chip training”, it is necessary to describe the single memory cells individually, which presupposes that all surrounding cells are inactive. Here, so-called nanoswitches, also known as “atomic switches”1,2, are used, which can be switched on and off with targeted voltage pulses. This allows the cells to be activated and deactivated individually and the memristors to be programmed in this way. Figure 2 shows a diagram of the structures to be set up and how they are controlled.
The subject of the announced master's thesis is the development of a nanoionic atomic layer switch and its implementation in memristive memory cells in order to be able to activate and deactivate them in a targeted manner. Such a switch consists of an inert and a soluble electrode, between which a very thin metal-oxide electrolyte is located. When a voltage pulse is applied, the material of the soluble electrode dissolves in the electrolyte and a conductive filament is formed. When this reaches the inert counter electrode, the cell becomes resistive and the switch is activated. When a negative voltage pulse is applied, the switch can be opened again. The aim of the work is the construction of such an atomic layer switch, the optimization of the layer system and the electrical characterization. The focus is on the relationship between the deposition conditions of the individual layers and the electrical behavior of the switch.
The Atomic Layer Deposition (ALD) is an important method for the formation of thin metallic, (semi)metal oxide or nitride layers from a precursor molecule under use of a second reactant. The reaction takes place thermally induced or enhanced by plasma. The growth rate is dependent on several parameters, like temperature and dosing time. The self-limiting character of the surface reaction is a special attribute of this method. The layer formation occurs during two separated process steps from a precursor and a second reactant. Due to the self-limiting character the deposited layers are typically highly conformal, homogeneous, without defects, and with a thickness of a few nanometers.
One suitable material for atomic switches is hafnium oxide which is deposited by ALD with an oxygen source as second reactant as exemplarily shown with water in Figure 3.
Also titania is mentioned as suitable material for memristive elements.4 Additionally, doping of the pristine oxides is described to purposeful enhance the layer properties.5 The material of memristive layers is as much relevant as the electrode materials and deposition methods.
Tasks:
- Literature review
- Fabrication of reference system of HfO2 and TiO2 (ALD) layers
- Variation of layer thicknesses under perpetuation of the materials
- Use of thermal and plasma-enhanced processes
- Variation of the ALD-deposition conditions and post-treatments
- Al-doping of hafnia and titania layers
- Deposition of top electrode layers, comparison of deposition method-dependent properties
- Implementation in switchable memristive memory cells
- Characterization: spectroscopic ellipsometry, electrical characterization, SEM (e.g., cross sectional analysis), memristive properties
- Electrical characterization of the finished switching elements
References
(1) Aono, M.; Hasegawa, T. The Atomic Switch. Proc. IEEE 2010, 98 (12), 2228–2236. https://doi.org/10.1109/JPROC.2010.2061830.
(2) Hino, T.; Hasegawa, T.; Terabe, K.; Tsuruoka, T.; Nayak, A.; Ohno, T.; Aono, M. Atomic Switches: Atomic-Movement-Controlled Nanodevices for New Types of Computing. Science and Technology of Advanced Materials 2011, 12 (1), 013003. https://doi.org/10.1088/1468-6996/12/1/013003.
(3) Li, J.; Guo, J.; Zhou, Z.; Xu, R.; Xu, L.; Ding, Y.; Xiao, H.; Li, X.; Li, A.; Fang, G. Atomic Layer Deposition Mechanism of Hafnium Dioxide Using Hafnium Precursor with Amino Ligands and Water. Surfaces and Interfaces 2024, 44, 103766. https://doi.org/10.1016/j.surfin.2023.103766.
(4) Ullah, F.; Tarkhan, M.; Fredj, Z.; Su, Y.; Wang, T.; Sawan, M. A Stable Undoped Low-Voltage Memristor Cell Based on Titania (TiOx). Nano Ex. 2024, 5 (1), 015003. https://doi.org/10.1088/2632-959X/ad1413.
(5) Park, W.; Park, Y.; Kim, S. Ferroelectric Properties of HfAlOx-Based Ferroelectric Memristor Devices for Neuromorphic Applications: Influence of Top Electrode Deposition Method. The Journal of Chemical Physics 2024, 161 (23), 234706. https://doi.org/10.1063/5.0239966.
The given references are just examples. The comprehensive literature is the base for this topic.
The topic is suitable for masters’ theses.
Estimated start of work: June 2025
We look for students with a solid background in technical understanding, physics, electrical engineering, thin film deposition methods or related fields.
If you are an interested and engaged student than apply with a short note of motivation and your CV. Describe your skills and give an overview of your recent marks.
Please contact us for more details.
Advisors
Falk Schaller
Fraunhofer Institute for Electronic Nano Systems ENAS
E-Mail: falk.schaller@enas.fraunhofer.de
Phone: +49 (0) 371 531 39048/ +49 371 45001 460
Shan Song
Fraunhofer Institute for Electronic Nano Systems ENAS
E-Mail: shan.song@enas-extern.fraunhofer.de
Phone: +49 (0) 371 531 33224/ +49 371 45001 461
And
Mathias Franz
Fraunhofer Institute for Electronic Nano Systems ENAS
E-Mail: mathias.franz@enas.fraunhofer.de
Phone: +49 (0) 371 531 33639 / +49 371 45001 612
Contact
Dr. Lysann Kaßner
lyg@…