Basic Introduction to Memristor

Catalog

Ⅰ Basic introduction

A memristor is a circuit device that expresses the relationship between magnetic flux and electric charge. The memristor has the dimension of resistance, but unlike resistance, the resistance of the memristor is determined by the charge flowing through it. Therefore, by measuring the resistance of the memristive, you can know the amount of charge flowing through it, and thus have the function of memory charge. The emergence of nano-memristive devices is expected to realize non-volatile random access memory. In addition, the integration, power consumption, and reading and writing speed of the memristive-based random access memory are superior to those of the traditional random access memory. Memristive is the best way to implement artificial neural network synapses in hardware. Due to the non-linear nature of memristance, chaotic circuits can be generated, which also have many applications in secure communication.

The person who first proposed the concept of memristor was Cai Shaotang, who taught at the University of California, Berkeley in the United States at that time. In 1971, when studying the relationship between charge, current, voltage, and magnetic flux, Professor Cai concluded that there should be another component besides resistor, capacitor, and inductor, representing the relationship between charge and magnetic flux. The effect of this component is that its resistance will change with the amount of current passed, and even if the current stops, its resistance will still stay at the previous value, and it will not be pushed back until it receives the reverse current.

memristor

Simply put, the memristor is a non-linear resistor with a memory function. The resistance value can be changed by controlling the change of the current. If the high resistance value is defined as "1" and the low resistance value is defined as "0", this kind of resistance can realize the function of storing data. In fact, it is a non-linear resistor with a memory function.

To use the analogy of a common water pipe, the current is the amount of water passing through, and the resistance is the thickness of the water pipe. When water flows in one direction, the water pipe will become thicker and thicker with the flow of water. At this time, if the water flow is turned off, the thickness of the water pipe will remain the same; on the contrary, when the water flows in the opposite direction, the water pipe will become thinner and thinner. Because such a component "remembers" the previous amount of current, it is called a memristor.

The relationship between the four basic elements

Because the memristor is small in size and low in energy consumption, it can store and process information well. The workload of a memristor is equivalent to the utility of a dozen transistors in a CPU chip.

Ⅱ Development process

1.Put forward ideas

Professor Cai proposed the memristor only because it should exist in the mathematical model. In order to prove the feasibility, he used a bunch of resistors, capacitors, inductors, and amplifiers to make a circuit that simulates the memristor effect. At that time, he didn't find any material that has an obvious memristor effect, and no one was looking for it. There are at least 15 years away from the popularization of home computers, HP will be on the stage at this time.

2.The study

HP's discovery of the memristor was published in the journal Nature in 2008. In 2009, it proved that the CrossLatch system can be easily stacked to form a three-dimensional memory. The "switch" between each wire is about 3nm x 3nm, and the switching time is less than 0.1ns. The overall operating speed is about the same as DRAM, but the number of switching is not as good as DRAM. It can’t replace DRAM now, but it has the amazing potential capacity of 1 cm² 100 gigabit (GB), 1cm³ 1 petabit (data storage unit 1PB=1000TB).

Crossbar Latch is not only used to store data. Its grid-like design and switches between each intersection mean that the entire group of grids can be logicalized to some extent. The original Crossbar Latch paper has already mentioned how to use grids to simulate the three logic gates of AND, OR, and NOT. The combination of several grids can even perform operations such as addition. This opens a window for getting rid of transistors and entering the next generation. Many people think that the leap of the memristor computer relative to the transistor is as big as the leap of the transistor relative to the vacuum tube. On the other hand, some people are discussing the possibility of the circuit adjusting its state in real-time to meet the computing requirements. This, coupled with the memory capacity of the memristor, means that the arithmetic circuit and the memory circuit can coexist at the same time and can be adjusted as needed. This has completely exceeded the design logic of this generation of computers, and if it can be developed in this way, it may represent the birth of a new generation of intelligent robots.

The combination of memristor and Crossbar Latch represents a new advancement in computer technology, which may allow us to continue the life of Moore's Law once again and move towards a future ruled by robots.

Researchers at Hewlett-Packard Laboratories believe that RRAM is what Chua calls a memristor. The RRAM device based on TiO2 reported by him was published in the journal Nature on May 1, 2008.

RRAM enables mobile phones to be used for several weeks or longer in the future without charging; it can start the personal computer immediately after it is turned on. The notebook computer still remembers the last used information long after the battery runs out. The memristor will also challenge the flash memory commonly used in handheld electronic devices because it has the ability to remember data even after the power is turned off. RRAM will memorize information faster than flash memory, consume less power, and take up less space. Memristors work in a similar way to the human brain. Hewlett-Packard says that perhaps someday computer systems can use memristors to memorize and associate certain patterns like humans.

RRAM has paved the way for the manufacture of non-volatile storage devices, instant PCs, more energy-efficient computers, and analog computers that process and contact information in a way similar to the human brain. In the future, it may even be achieved by greatly improving transistors. The functional density has a major impact on the development of electronic science.

The researchers said that the most interesting feature of the memristor device is that it can remember the amount of charge flowing through it. Professor Cai’s original idea was that the resistance of the memristor depends on how much charge passes through the device. In other words, if the charge flows in one direction, the resistance will increase; if the charge flows in the reverse direction, the resistance will decrease. Simply put, the resistance of such a device at any moment is a function of time - or how much charge passes forward or backward through it. The confirmation of this simple idea will have a profound impact on computing and computer science.

3.Breakthrough 

Dr. Thomas of Bielefeld University and his colleagues produced a memristor with learning ability in 2012. In 2013, Andy Thomas used this memristor as a key component of the artificial brain, and his research results are published in the Journal of Physics D: Applied Physics.

Memristor developed by Bielefeld University

Andy Thomas explained that because of the similarity between the memristor and the synapse, it is an excellent material for making artificial brains—and thus creating a new generation of computers. “It allows us to build extremely energy-efficient , durable, and self-learning processor.” Thomas’ article summarizes his own experimental results and draws on the results of other biological and physics researches, expounding for the first time how this neural-like computer can transform natural phenomena into technology system, and several principles that should be followed in it. These principles include that the memristor should "pay attention" to the previous electronic pulse like a synapse; and only when the stimulation pulse exceeds a certain amount, the neuron will respond, and the memristor should be the same.

HP memristor structure diagram and physical microscope diagram

Memristors can continuously increase or decrease resistance. Thomas explained: "This is also the basis for how the memristor functions in the process of learning and forgetting in the artificial brain."

Ⅲ Prospects

A faster and more energy-efficient instant PC is expected to be made

The simplest application of the memristor is as a non-volatile impedance memory (RRAM). The biggest problem that the current dynamic random access memory faces is that when you turn off the PC power, the dynamic random access memory forgets what was there. So when you turn on the computer, you have to sit there and wait until everything that needs to run the computer is loaded from the hard disk to the dynamic random access memory. With non-volatile random access memory, that process will be instantaneous, and your PC will return to the same state it was when you turned it off.

According to the researchers, the memristor allows mobile phones to be used for weeks or longer without recharging, and it also allows laptop computers to retain information long after the battery is exhausted. The memristor is also expected to challenge the flash memory commonly used in digital devices because it has the ability to save information even after the power is turned off. Chips made with this new discovery will save information faster than flash memory, consume less power, and take up less space.

Paving the way for the development of analog computers

Memristors can also allow computers to understand the way they collect data in the past, which is similar to the way the human brain collects and understands a series of things. It allows computers to be smarter in finding the data they have saved. For example, based on the information collected in the past, the memristor circuit can tell a microwave oven the heating time for different foods. Currently, many researchers are trying to write computer codes that run on standard machines to simulate brain functions. They use a large number of machines with huge processing power, but they can only simulate a small part of the brain. Researchers claim that they can stimulate the brain or stimulate certain functions of the brain in a way different from writing computer programs, that is, relying on the construction of hardware that simulates brain functions based on a memristor. The basic principle is that instead of 1 and 0, almost all states in gray like different light and dark are used instead. Such a computer can do many things that digital computers are not very good at-such as making decisions, judging that one thing is bigger than another, or even learning. Such hardware can be used to improve facial recognition technology, which should be thousands to millions of times faster than running programs on digital computers.