FINAL REPORT
ON
Progress on thin film based memristors
technology
BY
ANANYA BHARGAVA 2022A8PS1245P
Birla Institute of Technology and Sciences, Pilani
5th December, 2024
INSTRUCTOR’S NAME: MR. SATYENDRA KUMAR
MOURYA
1
ABSTRACT
This report offers a comprehensive overview of memristors, a novel
class of circuit elements that promises to revolutionize modern
computing. The report starts by introducing memristors and detailing
the different types of memristors, their structural and operational
characteristics, and the mechanisms that enable their unique
properties. A special emphasis is placed on the oxide materials that are
frequently used in the fabrication of memristors and the numerous
techniques used in the formation of these devices. The functionality of
the memristor can also be described by applying the pipe-water
analogy that provides a simple description of its behavior.
Further, the performance matrix of memristors is reviewed with
attention to some of its most impressive metrics: switching speed and
scalability, in addition to energy efficiency. The second part of the
report is to talk about logic family associated with memristors including
team model for an integrated systems comparison between TCAD and
CAD tools for the simulation behavior of memristors tools such as
Silvaco, Sentaurus, and the rest with TCAD.
Applications of memristors are examined, demonstrating their potential
in fields like artificial intelligence, neuromorphic computing, and real-
time data processing. Finally, the report concludes with a summary of
the promising future of memristors in next-generation technologies,
emphasizing their advantages in terms of speed, scalability, non-
volatility, and cost-effectiveness.
2
TABLE OF CONTENTS
Page No.
1. Introduction 4
2. Types of Memristors 5
3. Structure of Memristors 7
4. Working and Mechanism of Memristors 8
5. Oxide Materials and their Fabrication 10
6. Pipe-Water Analogy 13
7. Performance Matrix 14
8. Logic Family Implementation (TEAM Model) 16
9. TCAD vs CAD 17
10. Silvaco and Sentaurus 18
11. Implementation using TCAD tools 21
12. Other TCAD tools 22
13. Application 24
14. Conclusion 24
15. Reference and Bibliography 25
3
INTRODUCTION
• A memristor (memory resistor) is a semiconductor device which
regulates the flow of electrical current in a circuit and can retain
memory of past electrical states even when power is switched off.
• First postulated by Leon Chua in 1971, the memristor is
recognized as the fourth fundamental passive circuit element
alongside resistors, capacitors, and inductors
• Characteristics include Non-Volatility, Resistance variability, Low
Power consumption, Fast switching speed, Passive element, Non-
Linearity, No phase-shift introduced between current and voltage
at zero crossing.
• It is autodoped- removes some of the oxygen atoms from lattice
of TiO2, the remaining vacancies are positively charged donors to
the material and can remove upto 3 percent of oxygen atoms from
lattice without actually causing the basic structure to collapse
• It has a wide bandgap of 3ev to provides good insulation, because
it needs high voltage to move oxygen ions and for stability.
• The potential applications of memristors are vast, including high-
density memory storage, signal processing, analogue
computation, and (AI) applications and neuromorphic computing.
4
TYPES OF MEMRISTORS
2D Materials: Two-dimensional materials such as graphene,
molybdenum disulfide (MoS₂), and hexagonal boron nitride (h-BN) are
often used in fabricating 2D memristors. They are particularly suited for
flexible and transparent electronics, where traditional materials like
silicon may not perform well.
Advantages- High Density and Scalability, Flexibility and
Transparency, Fast Switching and Low Power Consumption.
Electronic Properties includes High Electron Mobility, Low Power
Consumption, High On/Off Ratios, Flexible and Transparent,
Strong Exciton Binding Energy
3D Materials: 3D memristors like Titanium Dioxide(TiO₂) often utilize
bulk materials and three-dimensional architectures to enhance their
performance and scalability.
Advantages- Increased Storage Capacity, Enhanced Performance,
Versatility, High Integration Density.
Electronic Properties includes High Storage Density, Enhanced
Scalability, Improved Switching Speed, High On/Off Ratio, Greater
Thermal Stability, Potential for Multi-Level Storage.
[Link] Switching Memristors (RSMs)
Materials: TiO₂, HfO₂, Phase-Change Materials (GeSbTe)
Mechanism: Resistance change due to ion migration or phase change
Applications: Non-volatile memory, neuromorphic computing
Efficiency: High scalability and reliability
5
2. Spintronic Memristors
Materials: Magnetic Tunnel Junctions (MTJs)
Mechanism: Resistance change based on electron spin alignment
Applications: High-speed, low-power memory
Efficiency: Faster switching, lower power consumption
3. Organic Memristors
Materials: Conductive polymers (e.g., polythiophene)
Mechanism: Charge carrier movement in organic materials
Applications: Flexible, low-cost electronics
Efficiency: Cost-effective but lower performance
4. Quantum Memristors
Materials: Quantum dots, superconducting materials
Mechanism: Quantum tunneling and entanglement
Applications: Quantum computing, ultra-fast processing
Efficiency: Potential for extreme speed and power efficiency
Most Efficient: Oxide-based RSMs (e.g., TiO₂) offer the best balance of
performance, scalability, and cost, while Spintronic Memristors and
Quantum Memristors excel in speed and power efficiency for advanced
applications.
6
STRUCTURE OF MEMRISTORS
Memristors consist of two distinct layers of TiO₂, each playing a critical
role in the device's functionality:
1. Stoichiometric TiO₂ Layer (Non-Oxygen-Deficient)
• Contains a balanced TiO₂ structure without significant oxygen
vacancies.
• Primarily serves as an insulating layer or high resistance layer that
provides structural and electronic stability.
• Ensures reliable operation by supporting the integrity of the
device.
2. Oxygen-Deficient TiO₂₋ₓ Layer (Doped with oxygen vacancies)
• Contains TiO₂ with intentional oxygen deficiencies, creating mobile
oxygen vacancies as memristors are autodoped.
• These vacancies act as charge carriers, enabling resistive switching
behavior.
• Under a high electric field, these vacancies migrate within the
layer, forming or disrupting conductive pathways.
7
WORKING OF MEMRISTORS
1. In the Oxygen-deficient TiO₂-x Layer, these oxygen vacancies are
mobile and when a high electric field is applied across the device
they can move within the layer, creating conductive pathways.
2. The tunneling gap refers to the energy gap between conductive
regions in a material, which electrons must "tunnel" through to
move from one region to another. In resistive switching materials
like TiO₂, this gap is influenced by the positioning of oxygen
vacancies.
3. When the gap is narrow, electrons can more easily tunnel through,
resulting in higher conductivity (low-resistance state). When the
gap widens, there is more barrier, tunneling becomes more
difficult, leading to lower conductivity (high-resistance state).
4. Positive Bias Voltage: Oxygen vacancies drift towards the left,
reducing the tunneling gap and creating a low-resistance state.
(ON state)
5. Negative Bias Voltage: Oxygen vacancies move back, widening the
tunneling gap and restoring a high-resistance state. (OFF state)
6. This effect is just moving the Oxygen vacancies back and forth and
changing the resistance. Once the voltage is removed, these
vacancies stay in place due to the energy barrier in the atomic
lattice, locking in the resistance state.
This stable configuration retains the information, allowing the
memristor to hold data even without power.
8
MECHANISM OF MEMRISTORS
Strukov realized a proof-of-concept memristor, which consists of a thin
nano layer (2 nm) of TiO2 and a second oxygen deficient nano layer of
TiO2−x (8 nm) sandwiched between two Pt nano wires (50nm)
By applying a positive voltage, to the top platinum nanowire, oxygen
vacancies drift from the TiO2−x layer to the TiO2 undoped layer, thus
changing the boundary between the TiO2−x and TiO2 layers. As a
consequence, the overall resistance of the layer is reduced
corresponding to an “ON” state. When enough charge passes through
the memristor that ions can no longer move, the device enters a
hysteresis region and keeps q at an upper bound with fixed
memristance, M (memristor resistance).
By reversing the process, the oxygen defects diffuse back into the
TiO2−x nano layer. The resistance returns to its original state, which
corresponds to an “OFF” state. The significant aspect to be noted here
is that only ionic charges, namely oxygen vacancies (O2−) through the
cell, change memristance. The resistance change is non-volatile hence
the cell acts as a memory element that remembers past history of ionic
charge flow through the cell.
Memristor is a switching device. And memristor has two switching
portions based on its state; one is high resistance state and low
resistance state. On the other hand, it has two regions which are
“doped” and “undoped” regions they are related to RON and ROFF
respectively.
9
OXIDE MATERIALS IN MEMRISTORS
Below is an overview of widely used materials and the criteria for their
selection:
1. Titanium Dioxide (TiO₂)
• Properties: Known for stable switching and ease of fabrication.
• Mechanism: Oxygen vacancies in TiO₂ layers lead to changes in
resistance.
• Applications: Commonly used in Resistive Random-Access
Memory (ReRAM) devices for reliable performance.
2. Hafnium Oxide (HfO₂)
• Advantages: Offers excellent endurance, scalability, and low
power consumption.
• Applications: Ideal for high-performance memory requiring
precise resistance control.
3. Zinc Oxide (ZnO)
• Properties: High mobility and transparency, making it suitable for
flexible and transparent electronics.
• Applications: Used in wearable and flexible memristors.
4. Aluminum Oxide (Al₂O₃)
• Role: Often serves as a dielectric layer, improving device stability
and reliability.
10
5. Nickel Oxide (NiO)
• Potential: High resistance contrast enhances data storage
capabilities.
• Applications: Suitable for advanced ReRAM with higher storage
density.
Material Selection Criteria for Memristors:
1. Switching Uniformity: Ensures consistent and reliable
performance, crucial for memory and computing applications.
2. Scalability: The material should maintain performance at smaller
sizes for integration into high-density memory arrays.
3. Fabrication Compatibility: The material must be compatible with
existing semiconductor manufacturing processes like CMOS for
cost-effective mass production.
4. Non-volatility: The material should retain data even without
power, making it suitable for long-term memory storage.
5. Switching Speed and Energy Efficiency: Fast switching and low
energy consumption are essential for high-performance
applications like real-time computing.
6. Long-term Stability: The material should endure many switching
cycles without degradation, ensuring reliability over time.
These materials and selection criteria are crucial for developing
memristors for applications such as non-volatile memory, neuromorphic
computing, and flexible electronics.
11
FABRICATION OF MEMRISTORS
Some of the Fabrication techniques are –
1. Sputtering: used to deposit thin oxide layers like TiO2 and HfO2,
enables fine control over film thickness and composition. Essential
for creating scalable and high-performance memristors.
2. Sol-Gel Processing: It involves the deposition of oxide layers
through solution-based methods. It is cost effective.
3. Atomic Layer Deposition (ALD): It provides high precision, ensures
excellent film consistency across large areas (Uniformity), used for
HfO2 deposition in advanced memristor designs.
4. Electrohydrodynamic Jet Printing: utilizes electric fields to pattern
oxide materials precisely. It enables the creation of flexible and
transparent memristors.
5. Thermal Oxidation: Involves oxidizing metal films at elevated
temperatures, offers a straightforward method for producing
uniform oxide layers (Simplicity), early memristor prototypes.
Morphologies refer to the physical arrangement of the oxide
materials, which includes thin films, nanowires, or layered structures
that enhance device performance. Nanostructures increase surface
area and facilitate ion mobility.
12
PIPE WATER ANALOGY
1. Pressure at the input of the pipe is similar to the applied voltage,
and rate of flow of water through pipe is the electrical current
2. An analogy for a memristor is an interesting kind of pipe that
expands or shrinks when water flows through it.
3. If the water flows in one direction, it causes the pipe’s wall to
increase, thus increasing the diameter of the pipe and enabling
the water to flow faster.
4. In the memristor, current flow causes the movement of ions in
such a way that it creates a larger conductive path for electrons,
reducing resistance. (ON STATE)
5. If the water flows in opposite direction, flexible pipe compresses,
thus the diameter of the pipe decreases, thus slowing down the
flow of water.
6. In a memristor, reverse current flow causes ions to shrink the
conductive path, increasing resistance and reducing current.
(OFF STATE)
7. If the water pressure is turned off, the pipe will retain its most
recent diameter until the water turned back on.
8. Thus, the pipe does not store the water like a bucket (like a
capacitor) but it remembers how much water flowed through it.
(like a memristor)
13
PERFORMANCE METRICS OF MEMRISTORS
Memristors are revolutionary devices offering immense potential for
modern computing applications. Their natural intelligence, exceptional
speed-energy efficiency, and advanced features make them ideal for
next-generation technologies such as artificial intelligence (AI),
neuromorphic computing, and real-time analytics.
Key Advantages of Memristors
1. Switching Speed: Memristors can switch states in just 85
picoseconds (ps), which enables quick data processing and memory
access and essential for real-time applications.
2 Scalability: Memristors are highly scalable, functioning effectively at a
2nm feature size. This allows for their integration into dense memory
arrays, meeting the growing need for compact and efficient systems.
3 Multilevel Storage: Memristors can store over 256 levels of data,
allowing them to hold more than 8 bits per cell. This multilevel
capability boosts memory density and ensures energy-efficient
performance compared to traditional binary memory.
4 Stackability: Memristors can be stacked up to 8 layers, enabling
higher storage capacities within a smaller space.
5 Non-Volatility: Memristors retain stored data even when power is
turned off, making them a reliable choice for non-volatile memory
applications.
6 Cost Efficiency: Memristors are relatively inexpensive to fabricate,
offering a practical solution for widespread use in both consumer and
industrial applications.
14
15
LOGIC FAMILY AND TEAM MODEL
Memristors, often called the "fourth fundamental circuit element," are
gaining attention in logic design and digital circuits due to their unique
ability to store information and alter resistance based on past
voltage/current. In logic design, Memristor Ratioed Logic (MRL) is an
emerging logic family that uses voltage levels to represent logic states,
making it potentially compatible with standard CMOS logic. It is inspired
by Diode Logic, sharing characteristics like non-inverting and non-
restoring properties. However, since memristors are passive two-
terminal devices, CMOS inverters are required to complete the logic
basis. Other memristor-based logic families include Material Implication
Logic, Memristor Aided Logic (MAGIC), and Programmable Logic
Memristor Arrays.
The TEAM Model offers a mathematical framework to simulate the
dynamic resistance changes in memristors, based on voltage thresholds.
By fitting the TEAM model to memristors, basic logic gates like AND, OR,
XOR, and more complex circuits like full adders can be simulated to
evaluate their performance. This model captures crucial memristor
characteristics, such as threshold-dependent switching, which allows for
a realistic simulation of their impact on gate performance, speed,
power consumption, and output accuracy. The TEAM model bridges
theoretical memristor properties with practical circuit simulations,
providing valuable insights for their integration into modern electronics.
16
TCAD vs CAD TOOLS
TCAD (Technology Computer-Aided Design) Tools:
• TCAD tools like Silvaco, Sentaurus focus on the physical simulation
of semiconductor devices.
• These tools simulate the underlying physics of devices, including
charge transport, doping profiles, electrical behavior, thermal
effects, and fabrication processes.
• Used for device-level design and analysis, such as transistor
performance, memristor modeling, and material property
investigations.
CAD (Computer-Aided Design) Tools:
• CAD tools are primarily used for circuit-level design and layout.
• Tools like PSpice, LTSpice, and HSPICE simulate electrical circuits
and their components without focusing on the microscopic
physical details.
• Ideal for designing and simulating circuits like logic gates,
amplifiers, or system-level architectures.
17
Prominent TCAD tools are:
1. Silvaco TCAD -It offers a comprehensive suite for semiconductor
simulation, featuring:
• Device Simulation: allows 2D and 3D modeling of devices like
MOSFETs and memristors, analyzing their electrical characteristics,
Process Simulation, Compact Modeling: SmartSpice enables
circuit simulation, integrating device models for performance
predictions.
• User-Friendly Interface: Its intuitive GUI makes complex
simulations accessible for various users.
Main Libraries are-
• ATLAS: For device simulations, including semiconductor physics.
• SmartSpice: For circuit simulation.
• Victory: For process and device simulations, including layout and
optimization.
• S-PISCES: For 2D/3D simulation of semiconductor devices.
• TMA: For thermal simulations.
Silvaco is widely used in developing advanced semiconductor devices
and integrated circuits.
18
Sentaurus TCAD- It provides sophisticated simulation capabilities,
including:
• Device Simulation: allows detailed 2D and 3D modeling for
various semiconductor technologies, including FinFETs and SOI
devices.
• Process Simulation models and optimizes manufacturing steps
like doping and etching.
• Circuit Simulation: Integrates with HSPICE for accurate circuit
simulations, linking device models to circuit designs.
• Advanced Physical Models: Incorporates detailed models for
carrier transport and quantum effects, ideal for nanoscale devices.
Main Libraries are-
• Sentaurus Device: For semiconductor device modeling
• Sentaurus Process: For semiconductor fabrication and process
simulations
• Sentaurus TCAD: For integrated device and process design.
• Sentaurus Workbench: For simulation workflow management.
• Sentaurus Quantum: For quantum transport simulation
19
IMPLEMENTATION OF MEMRISTORS USING
TCAD Tools
TCAD tools play a crucial role in the implementation and simulation of
memristors by providing a platform to analyze their behavior at the
device level. These tools allow researchers and engineers to model the
physical, electrical, and material properties of memristors, enabling a
deeper understanding of their operation and optimization for various
applications.
Implementation Steps Using TCAD Tools
1. Device Structure Definition: TCAD tools like Silvaco, Sentaurus or
Synopsys TCAD help define the physical structure of the
memristor, including layers of materials (e.g., TiO₂ or HfO₂ for the
resistive switching layer). Parameters such as thickness, doping
concentrations, and material properties are input.
2. Material Modeling: Memristor behavior depends on material
properties like ion mobility, defect density, and conductivity. TCAD
tools simulate material interactions, incorporating models for ionic
transport and electric-field-induced switching.
3. Simulation of Resistive Switching: TCAD tools enable modeling of
the memristive effect by simulating the change in resistance due
to ionic migration and redox reactions. Models such as drift-
diffusion and Poisson's equation are used to simulate charge
carrier dynamics.
4. Electrical Characterization: Simulations involve applying voltage
or current signals to study memristor characteristics such as I-V
curves, hysteresis behavior, and switching dynamics. These
simulations validate the memristor’s performance under varying
operational conditions.
20
5. Optimization and Scaling: TCAD tools allow adjustments to
material properties, device dimensions, and operational
conditions to optimize memristor performance. Scalability analysis
ensures the device meets design requirements for integration into
circuits.
OTHER TCAD TOOLS
1. COMSOL Multiphysics- : COMSOL’s multiphysics simulation capability
is unique, enabling coupled simulations across different physical
phenomena. It’s ideal for modeling devices like memristors where
thermal effects and coupled behaviors influence device performance.
Best For: Multiphysics simulations, integrating electrical, mechanical,
and thermal effects.
2. Medici (by Dassault Systèmes)- Medici specializes in high-accuracy
3D device modeling, particularly for nanoscale devices. It’s well-suited
for simulating advanced semiconductor structures and is used
extensively for next-gen technologies like FinFETs.
21
Best For: 3D simulations of advanced semiconductor devices.
3. Wise (by Synopsys)-Wise focuses on advanced process steps and
device optimization, particularly for next-gen technologies like FinFETs.
It’s ideal for cutting-edge semiconductor simulation.
Best For: Next-generation technology simulations, especially for FinFETs
4. MST (by Mitsubishi Electric)- MST is specialized for simulating
material properties and performance of resistive devices. It’s especially
valuable for memristor modeling, focusing on electrical characteristics
and resistive switching.
Best For: Material property modeling for resistive devices like
memristors.
5. Simulink (by MathWorks)- Simulink is used for system-level modeling
and integration with other TCAD tools. It’s particularly useful for
simulating neuromorphic systems involving memristors in large-scale
computing architectures.
Best For: System-level simulations and integration with other TCAD
tools.
6. CoventorWare (by Coventor)- CoventorWare excels in simulating
MEMS devices and semiconductor process integration, making it ideal
for applications where electromechanical behavior interacts with
semiconductor properties. It’s a niche tool for MEMS-based
semiconductors. Best For: MEMS and semiconductor device simulation.
Which is better and why:
• Sentaurus is likely the best overall if you need a comprehensive,
multi-physics approach for complex simulations, especially for
high-accuracy modeling of semiconductor devices like
memristors.
22
• Silvaco is a strong contender if you need user-friendliness
combined with advanced process simulation tools.
• COMSOL is ideal if you want multiphysics modeling, especially
when the device’s behavior is influenced by multiple physical
phenomena (such as thermal effects in memristors).
• Medici and Wise are better suited for advanced nanoscale
simulations and next-generation technology (e.g., FinFETs), while
MST shines for material-based simulations of memristors.
23
CONCLUSION
Memristors are transformative technology that will change the
computing world. Their unique features, such as ultra-fast switching
speeds, scalability, multilevel storage, and non-volatility, place them at
the top for next-generation applications in the realms of artificial
intelligence, neuromorphic computing, and real-time data analytics. The
use of oxide materials in their fabrication improves the performance of
these, and advanced techniques and tools like TCAD, Silvaco, and
Sentaurus facilitate the design and implementation of memristor-based
systems.
Versatility of memristors offers the advantages that these would work in
high-density memory arrays and sustain data for long periods even
when their power is cut off, outperforming most of the technologies
related to traditional memory systems. The cost-effectiveness promotes
practical applicability of the products within consumers as well as
industrialized segments. Further development in this research is going
to shape the highly efficient, scalable, and reliable computing systems
wherein memristors will lead from the front.
This promise will unlock new opportunities in computing, and the
future seems bright with more efficient and intelligent technologies.
Overall, memristors hold a lot of promise for revolutionizing the
electronics landscape and making way for the next wave of computing
innovations.
24
References and Bibliography
1. Chua, L. O. (1971). Memristor: The 4th circuit element. IEEE
Transactions on Circuit Theory, 18(5), 507-519.
[Link]
OR_THE_4_TH_CIRCUIT_ELEMENT/citations
2. Kumar, Abhishek. (2012). MEMRISTOR: THE 4 TH CIRCUIT
ELEMENT. 3. 176-180.
3. Islam, Md Mydul & Al Mahmud, Faysal. (2018). Design and
Implementation of Memristor.
[Link]
d_Implementation_of_Memristor/citation/download
4. Prakash S , Vijay Kuruba , Yempati Venkata Rajesh , Marineni
Vamsi, 2021, Design and Implementing an Memristor in Basic
Analog Circuits, INTERNATIONAL JOURNAL OF ENGINEERING
RESEARCH & TECHNOLOGY (IJERT) Volume 10, Issue 11
(November 2021),
5. Teja Nibhanupudi, S.S., Roy, A., Veksler, D. et al. Ultra-fast
switching memristors based on two-dimensional materials. Nat
Commun 15, 2334 (2024). [Link]
46372-y
6. Youtube Links :
• [Link]
• [Link]
• [Link]
• [Link]
• [Link]
25