阅读说明：本技术 一种利用全光控忆阻器实现非易失性布尔逻辑的方法 (Method for realizing nonvolatile Boolean logic by using all-optical control memristor ) 是由 诸葛飞 杨静 胡令祥 于 2021-04-21 设计创作，主要内容包括：本发明公开一种利用全光控忆阻器实现非易失性布尔逻辑的方法,所述忆阻器包括两个电极端口：顶电极和底电极,以及位于两电极间的中间介质层,其特征在于,所述方法为通过所述底电极层或顶电极层输入光信号,其中包括增加所述忆阻器电导的光写入模式和降低所述忆阻器电导的光擦除模式；在所述光写入模式下,所述光信号为紫外光和可见光；在所述光擦除模式下,所述光信号为可见光和近红外光；所述逻辑的输入为忆阻器的初始电导态和光信号；所述逻辑的输出为忆阻器的最终电导态。本发明公开的方法利用的全光信号调控相比电信号调控,具有带宽高,功耗低,串扰小等优点,在实施光电逻辑运算方面具有很大优势。(The invention discloses a method for realizing nonvolatile Boolean logic by using an all-optical control memristor, wherein the memristor comprises two electrode ports: the method is characterized in that an optical signal is input through the bottom electrode layer or the top electrode layer, and the optical signal comprises an optical writing mode for increasing the conductance of the memristor and an optical erasing mode for reducing the conductance of the memristor; in the optical writing mode, the optical signals are ultraviolet light and visible light; in the optical erasing mode, the optical signals are visible light and near infrared light; the inputs of the logic are the initial electrical conductivity and optical signal of the memristor; the output of the logic is the final electrical conduction state of the memristor. Compared with electric signal regulation, the all-optical signal regulation and control method disclosed by the invention has the advantages of high bandwidth, low power consumption, small crosstalk and the like, and has great advantages in the aspect of implementing photoelectric logic operation.)

1. A method of implementing non-volatile boolean logic with an all-optical controlled memristor, the memristor including two electrode ports: the method is characterized in that an optical signal is input through the bottom electrode layer or the top electrode layer, and the method comprises a light writing mode for increasing the conductance of the memristor and a light erasing mode for reducing the conductance of the memristor;

in the optical writing mode, the optical signals are ultraviolet light and visible light;

in the optical erasing mode, the optical signals are visible light and near infrared light;

the inputs of the logic are the initial electrical conductivity and optical signal of the memristor;

the output of the logic is the final electrical conduction state of the memristor.

2. The method of implementing non-volatile Boolean logic with an all-optical controlled memristor according to claim 1, wherein an initial electrical conductance state of the memristor and an optical signal applied on the memristor are defined as two input variables p, q, and a final electrical conductance state of the memristor is defined as an output variable p';

defining a threshold current magnitude of the memristor high/low electrical conductance transition as I_th；

The logic input variables p-0 and q-0 respectively represent that the initial electric conduction state of the device is a low electric conduction state and no optical signal is applied;

the logic input variables p ═ 1 and q ═ 1 represent that the initial electrical conductivity of the device is a high electrical conductivity state and that an optical signal is applied, respectively;

the logic output variables p '0 and p' 1 represent the final electric conduction state of the device to be a low electric conduction state and a high electric conduction state respectively;

the applying of the optical signal includes two modes: in the optical writing mode, an optical signal in the optical writing mode is green light; in the optical erasing mode, the optical signal in the optical erasing mode is red light.

3. The method for implementing nonvolatile Boolean logic with an all-optical memristor according to claim 1, wherein the intermediate dielectric layer is an oxide material, and the thickness of the intermediate dielectric layer is 5nm-500 nm.

4. The method of claim 1, wherein an optical signal with a wavelength of 350nm to 550nm is input before the memristor photo-erase mode is adjusted by the optical signal input.

5. The method for implementing nonvolatile Boolean logic with the all-optical memristor according to claim 1, wherein in the optical writing mode, the optical signals are ultraviolet light and visible light, and the wavelength is 300nm-750 nm.

6. The method of claim 1, wherein in the optical erasing mode, the optical signals are visible light and near infrared light, and the wavelength is 600nm to 800 nm.

7. The method of claim 2, wherein the threshold I is set to be equal to the threshold I_thIs 0.75nA-6 nA.

8. The method of implementing non-volatile Boolean logic with all-optical controlled memristors according to claim 2, wherein the Boolean logic function implementation method based on a single memristor comprises: inputting the initial electrical conduction state and the optical signal of the memristor, and outputting the final electrical conduction state of the memristor.

9. The method for implementing nonvolatile Boolean logic by using an all-optical control memristor according to claim 8, wherein the initial electrical conducting state and the optical signal of the memristor are used as input, the final electrical conducting state is used as output, 16 Boolean logic functions are implemented by the following operation method, and the functions are classified into six categories according to the complexity of the implementation of the functions, specifically as follows:

the first type: the logic function can be realized by directly inputting p and q;

1) or logic, the initial electrical conductivity of the memristor is used as input p, a green light signal is used as input q, and the final electrical conductivity is used as output p', and the p and the q are directly input to realize the OR logic;

2) the negative substance inclusion logic is realized by taking the initial electrical conductivity of the memristor as an input p, taking a red light signal as an input q and taking the final electrical conductivity as an output p', and directly inputting p and q;

the second type: the need to control the optical signal;

1) the memristor comprises an AND logic, a control light signal and a control light signal, wherein the AND logic is implemented by taking the initial electrical conduction state of the memristor as an input p, taking a green light signal as an input q, taking the final electrical conduction state as outputs p', p and q, and then applying a red light signal as the control light signal;

2) the true logic is realized by taking the initial electrical conductivity of the memristor as an input p, taking a red light signal as an input q, and taking the final electrical conductivity as outputs p', p and q to be input, and then applying a green light signal as a control light signal;

3) the pseudo logic is implemented by taking the initial electrical conductivity of the memristor as an input p, taking a red light signal as an input q, and applying the red light signal as a control light signal after the final electrical conductivity is input as outputs p', p and q;

in the third category: a light forming operation is required;

1) the method comprises the following steps of (1) anti-substance inclusion logic, wherein an initial electrical conduction state of a memristor is used as an input p, a red light signal is used as an input q, and a final electrical conduction state is used as an output p', a green light signal is applied before the q input to execute a light forming operation, so that the substance inclusion logic is realized;

2) q logic, taking the initial electrical state of the memristor as an input p, a green light signal as an input q, and a final electrical state as an output p', applying a red light signal before the q input to execute a light forming operation, and realizing the q logic;

the fourth type: the initial electrical conductivity of the device needs to be measured;

1) the NAND logic is used for taking the initial electrical conduction state of the memristor as an input p, taking a red light signal as an input q and taking the final electrical conduction state as an output p', determining the application mode of a control optical signal according to the initial electrical conduction state of the device, and applying a green light signal as the control optical signal if p is 0; if p is 1, not applying and realizing NAND logic;

2) the method comprises the steps of essentially containing logic, wherein after an initial electrical conduction state of a memristor serves as an input p, a green light signal serves as an input q, and a final electrical conduction state serves as an output p', q is input, a red light signal is applied, a green light signal is applied according to the initial electrical conduction state of a device, and if p is 0, the green light signal is applied; if p is 1, no green signal is applied, and finally, the actual implication logic is realized;

3) the negative substance contains logic, the initial electrical conductivity state of the memristor is used as an input p, a green light signal is used as an input q, the final electrical conductivity state is used as an output p', the application mode of the control optical signal is determined according to the initial electrical conductivity state of the device, and if p is 0, the control optical signal is not applied and is directly output; if p is 1, continuously applying the red light signal twice, and finally realizing the inverse negative substance inclusion logic;

4) the p logic takes the initial electrical conduction state of the memristor as an input p, a green light signal as an input q and the final electrical conduction state as an output p', reads the initial electrical conduction state of the device, and applies a red light signal if p is 0; if p is 1, no red light signal is applied, and finally p logic is realized;

5) the non-p logic is used for taking the initial electrical conduction state of the memristor as an input p, taking a red light signal as an input q and taking the final electrical conduction state as an output p', reading the initial resistance state p, determining and controlling the application mode of an optical signal according to the initial electrical conduction state of the device, and applying a green light signal and not applying a red light signal if p is 0; if p is 1, applying a red light signal, not applying a green light signal, and finally realizing non-p logic;

the fifth type: both the initial electrical conductivity of the device and the photo-formation operation are required to be measured;

1) non-q logic, which takes the initial electrical conductivity of the memristor as input p, takes a red light signal as input q, takes the final electrical conductivity as output p', determines the application mode of the light signal in the light forming operation according to the initial electrical conductivity of the device before the q input, and applies a green light signal to execute the light forming operation if p is 0; if p is 1, no green light signal is applied, and finally non-q logic is realized;

2) the method comprises the following steps that NOR logic is used for determining the application mode of a light signal in the light forming operation according to the initial electric conduction state of a device before the initial electric conduction state of the memristor is used as an input p, a red light signal is used as an input q, and the final electric conduction state is used as an output p', q is input, and if p is 0, a green light signal is applied to execute the light forming operation; if p is 1, no green light signal is applied, after q is input, the application mode of the control light signal is determined according to the initial electric conductivity of the device, and if p is 0, no red light signal is applied; if p is 1, applying a red light signal, and finally realizing NOR logic;

the sixth type: in addition to the need for photo-forming operations, the need to measure the initial and intermediate electrical conductivity states of the device;

1) the memristor comprises an XOR logic, a logic circuit and a logic circuit, wherein the initial electrical conduction state of the memristor is used as an input p, a green light signal is used as an input q, and the final electrical conduction state is used as an output p'; before q is input, determining the application mode of the optical signal in the optical forming operation according to the initial electrical conductivity state of the device, and if p is 0, applying a green optical signal and not applying a red optical signal; if p is 1, applying red light signal and not applying green light signal, after q input, determining the applying mode of following control light signal according to the current electric conducting state of the device, if in high electric conducting state, applying red light signal and not applying green light signal; if the voltage is in a low-conductivity state, applying a green light signal, not applying a red light signal, and finally realizing exclusive or logic;

2) the same or logic is used for taking the initial electrical conductivity of the memristor as an input p, taking a green light signal as an input q and taking the final electrical conductivity as an output p'; before q is input, determining the application mode of the optical signal in the optical forming operation according to the initial electrical conductivity state of the device, and if p is 0, applying a green optical signal and not applying a red optical signal; if p is 1, applying red light signal and not applying green light signal, after q input, determining and controlling the application mode of light signal according to the current electric conduction state of the device, if in high electric conduction state, applying red light signal and not applying green light signal; if the device is in the low-conductivity state, applying a green light signal, not applying a red light signal, reading the conductivity state of the device again, and if the device is in the high-conductivity state, applying a red light signal, not applying a green light signal; if the current is in a low-conductivity state, a green light signal is applied, a red light signal is not applied, and finally p' is output to realize the same or logic.

10. A logic cascade method, according to the logic realization method using the all-optical control memristor in claims 1 to 9, wherein the output of the previous logic operation is saved in the form of a conductance value, the conductance value is used as an initial conductance state input p, an optical signal is used as another input q, and the final conductance state is used as an output p', so as to directly realize the logic cascade.

Technical Field

The invention relates to the technical field of photoelectronics, in particular to a method for realizing nonvolatile Boolean logic by using an all-optical control memristor.

Background

The memristor is a fourth circuit element, except for a resistor, a capacitor and an inductor, discovered by professor zeitchy begonia in 1971 when the relation among charge, current, voltage and magnetic flux is studied. In 2008, after the sandwich-structure memristor is made by using titanium dioxide as a dielectric layer for the first time in a Hewlett packard laboratory, researchers have conducted extensive research on the memristor, and in recent years, the practical application of the memristor is more concerned. With the rise of artificial intelligence, the traditional von Neumann architecture based on CMOS devices has the characteristic that data needs to be transferred back and forth between a computing unit and a storage unit due to the separation of the computing unit and the storage unit, and the requirement of the current Internet of things and big data era on intelligent development is increasingly difficult to meet. The memristor is generally a simple sandwich structure, the conductance of the memristor can be switched between a high state and a low state, if the low conductance state is defined as '0', and the high conductance state is defined as '1', logic operation can be naturally performed, and the memristor has the excellent characteristics of non-volatility, simple structure, low power consumption and the like, so that the memristor has unique advantages in the non-volatile logic field, and the so-called von neumann bottleneck can be effectively relieved. Certainly, to solve the problem fundamentally, a novel computing system integrating basic devices, circuits, architectures, systems and the like needs to be developed through collaborative innovation from multiple layers.

A cost-integrated architecture was first proposed in 1969. To date, a whole body can be roughly divided into three categories: (1) a non-volatile logic operation; (2) a non-volatile arithmetic operation; (3) neuromorphic calculations based on synaptic length range plasticity. In terms of logical operation, Hewlett packard proposed the use of Pt/Ti/TiO in 2010₂The Pt memristor circuit physically realizes the substantial implication logic for the first time, only 3 memristor devices are needed to realize the NAND logic, and the whole memory calculation process is completed by the memristor devices. This work fully demonstrates the great potential of memristive devices in the field of computing integration. The existing logic implementation methods based on memristors mainly have two main categories: non-state logic and state logic. For the non-state logic, two voltage signals are usually used as input, a conductance value is used as output, different physical quantities are provided, and because the input signals are voltages, the output signals need to be converted into voltage signals and then used as input again when the next logic is carried out, and the logic cascade cannot be directly carried out; and for state logic, input and outputAnd are expressed as physical quantities. In general, the electrical conductivity of a memristor can be reversibly modulated by applying a voltage signal. Chinese patents CN109994139A and CN 111061454 a disclose the implementation of 16 non-volatile boolean logic functions with voltage regulation applied across the device in a single unipolar and a single bipolar memristor, respectively. As is well known, optical signals have a series of advantages such as high bandwidth, small crosstalk, high speed, and good parallelism, compared with electrical signals. In order to fully utilize the advantages of light and further enrich the condition turnover of devices, the logic operation is more rapid and flexible, the conductance of the memristor can be regulated and controlled by light and electricity mixed signals, and various logic functions are realized. Chinese patent publication No. CN 109860389 a discloses that photoelectric dual drive is used to implement three binary logic functions of and, or and not in a memristive cell. All the above steps are realized by using full electric signals or photoelectric signals together. So far, the reversible regulation and control of the device electrical conductivity state is carried out only by using optical signals, and 16 kinds of Boolean logics are realized by using the reversible regulation and control, which is not reported yet.

Disclosure of Invention

The invention provides a method for realizing nonvolatile Boolean logic by using an all-optical control memristor, wherein the memristor realizes 16 Boolean logics by using optical signals and can directly carry out logic cascade.

A method of implementing non-volatile boolean logic with an all-optical controlled memristor, the memristor including two electrode ports: inputting an optical signal through the top electrode or the bottom electrode, wherein the optical signal comprises an optical writing mode for increasing the conductance of the memristor and an optical erasing mode for reducing the conductance of the memristor;

in the optical writing mode, the optical signals are ultraviolet light and visible light;

in the optical erasing mode, the optical signals are visible light and near infrared light;

the inputs of the logic are the initial electrical conducting state and the optical signal of the memristor respectively;

the output of the logic is the final electrical conduction state of the memristor.

According to the invention, reversible regulation and control can be carried out on the conductance of the device through optical signals with different wavelengths on a single memristor, the conductance of the memristor can be increased or reduced by utilizing illumination, and after the optical signals are removed, the conductance state of the memristor has good non-volatility, so that the logic can integrate the processing and the storage of data.

The initial electrical conductivity state of the memristor and the optical signal applied to the memristor are defined as two input variables p and q, respectively, and the final electrical conductivity state of the memristor is defined as an output variable p'.

Defining a threshold current magnitude of the memristor high/low electrical conductance transition as I_th；

The logic input variables p-0 and q-0 are respectively defined as the initial electrical conductivity state of the device is a low electrical conductivity state and no optical signal is applied;

the logic input variables p ═ 1 and q ═ 1 are defined as the initial electrical conductivity state of the device being a high electrical conductivity state and the optical signal being applied;

the logic output variables p '0 and p' 1 represent the final electrical conductivity of the device as low and high electrical conductivity states, respectively.

The applying of the optical signal includes two modes: the optical writing mode applies an optical signal for increasing the conductance of the memristor, and is further preferably green light; in the optical erasing mode, an optical signal for reducing the conductance of the memristor is applied, and the optical erasing mode is preferably red light.

The intermediate dielectric layer is made of an oxide material, and the thickness of the intermediate dielectric layer is 5nm-500nm, and is further preferably 50 nm.

A single-layer oxide film prepared in a pure argon environment is selected as an intermediate medium layer, and compared with a double-layer oxide utilizing an anoxic layer and an oxygen-rich layer, the structure is simpler and is easy to prepare. Meanwhile, when the memristor realizes the optical erasing, the device conductance is continuously reduced along with the illumination time, and when the memristor prepared by the double-layer oxide film executes the optical erasing mode, the device conductance is firstly increased and then reduced along with the illumination time, so that the realization of the Boolean logic function is not facilitated.

Before the optical erasing mode is executed, an optical signal with the wavelength of 350nm-550nm is input. The shorter wavelength optical signal tunes the electrical conductance of the memristor to a relatively higher electrical conductance state to better achieve optical erasure.

In the optical writing mode, the optical signals are ultraviolet light and visible light, the wavelength is 300nm to 750nm, and preferably, the optical signals are green light, and the wavelength is 480nm to 550 nm.

In the light erasing mode, the light signal is visible light and near infrared light, the wavelength is 600nm-800nm, and more preferably, the light signal is red light, and the wavelength is 600nm-750 nm.

The main mechanism of conductance regulation is that the oxide forms two back-to-back schottky junctions with the metal electrodes on both sides. When in equilibrium, the number of electrons passing through the schottky junction in both directions is equal and the net current is zero. When a bias is present, one junction is in forward bias and the other junction is in reverse bias. In general, the device conductance is determined by the schottky junction being in reverse bias. In addition, a large number of oxygen vacancy defects are present in the oxide. The width of the schottky junction is primarily determined by the concentration of ionized oxygen vacancies, i.e., the more ionized oxygen vacancies, the narrower the schottky junction and vice versa the wider. Under the irradiation of a green signal, oxygen vacancies in the oxide are ionized to generate electrons and charged oxygen vacancies, so that the electric conductance of the device is increased. A red signal is then applied to tunnel electrons in the electrode or directly across the schottky barrier into the conduction band of the oxide, recombining with ionized oxygen vacancies, thereby reducing the conductance of the device.

The threshold value I_thIs 0.75nA-6 nA.

Through the optical signal input of proper wavelength, can make the memristor have new electric conduction state, after the optical signal is withdrawed, in the longer time, the electric conduction state of memristor has nonvolatile to distinguish the final electric conduction state of memristor through the threshold value stability, guaranteed the realization of boolean logic.

The invention provides a method for realizing 16 Boolean logic functions based on a single all-optical control memristor, which comprises the following steps: and the initial electrical conduction state and the optical signal of the memristor are used as two inputs, and the final electrical conduction state of the memristor is used as an output, so that the Boolean logic function is realized. According to the complexity of function implementation, 16 kinds of boolean logic are divided into six categories, specifically as follows:

the first type: the logic function can be realized by directly inputting p and q.

1) OR (OR) logic, the initial electrical conductivity of the memristor is used as input p, a green light signal is used as input q, and the final electrical conductivity is used as output p', and the input p and the output q are directly used for realizing OR logic;

2) negative parenchyma inclusion (NIMP) logic, wherein the initial electrical conductivity of the memristor is used as input p, a red light signal is used as input q, and the final electrical conductivity is used as output p', and the negative parenchyma inclusion logic can be realized by directly inputting p and q;

the second type: the need to control the optical signal;

1) AND (AND) logic, wherein the initial electrical conductivity of the memristor is used as an input p, a green light signal is used as an input q, AND a red light signal is applied as a control light signal after the final electrical conductivity is used as outputs p', p AND q to realize AND logic;

2) true (TURE) logic, which is implemented by inputting the initial electrical state of the memristor as input p, a red light signal as input q, and a final electrical state as output p', p, and q, and then applying a green light signal as a control light signal;

3) FALSE (FALSE) logic, which is implemented by applying a red light signal as a control light signal after inputting an initial electrical conduction state of the memristor as an input p, a red light signal as an input q, and a final electrical conduction state as outputs p', p, and q;

in the third category: a light forming operation is required;

1) an inverse essence implication (RIMP) logic to implement RIMP logic with an initial electrical conductivity state of the memristor as an input p, a red signal as an input q, and a final electrical conductivity state as an output p', applying a green signal to perform a light forming operation before q input;

2) q logic, taking the initial electrical state of the memristor as an input p, a green light signal as an input q, and a final electrical state as an output p', applying a red light signal before the q input to execute a light forming operation, and realizing the q logic;

the fourth type: the initial electrical conductivity of the device needs to be measured;

1) and the NAND (NAND) logic takes the initial electrical conduction state of the memristor as an input p, a red light signal as an input q and the final electrical conduction state as an output p', and determines the application mode of the control light signal according to the initial electrical conduction state of the device. If p is 0, applying a green light signal as a control light signal; if p is 1, not applying and realizing NAND logic;

2) the method comprises the steps that (IMP) logic is adopted, an initial electrical conductivity state of a memristor is used as an input p, a green light signal is used as an input q, a final electrical conductivity state is used as an output p', after the q is input, a red light signal is applied, a green light signal is applied according to the initial electrical conductivity state of a device, and if p is 0, the green light signal is applied; if p is 1, no green light signal is applied, and finally IMP logic is realized;

3) and the inverse negative substance inclusion (RNIMP) logic takes the initial electrical conductivity state of the memristor as an input p, a green light signal as an input q and the final electrical conductivity state as an output p', and determines the application mode of the control optical signal according to the initial electrical conductivity state of the device. If p is 0, the control optical signal is not applied and the control optical signal is directly output; if p is 1, continuously applying the red light signal twice, and finally realizing RNIMP logic;

4) the p logic takes the initial electrical conduction state of the memristor as an input p, a green light signal as an input q and the final electrical conduction state as an output p', reads the initial electrical conduction state of the device, and applies a red light signal if p is 0; if p is 1, no red light signal is applied, and finally p logic is realized;

5) and non-p (NOT p) logic, taking the initial electrical conduction state of the memristor as an input p, taking a red light signal as an input q, taking the final electrical conduction state as an output p', reading the initial resistance state p, and determining and controlling the application mode of the optical signal according to the initial electrical conduction state of the device. If p is 0, applying a green light signal and not applying a red light signal; if p is 1, applying a red light signal, not applying a green light signal, and finally realizing non-p logic;

the fifth type: both the initial electrical conductivity of the device and the photo-formation operation are required to be measured;

1) a non-q (NOT q) logic, which takes the initial electrical conductivity of the memristor as an input p, takes a red light signal as an input q, and takes the final electrical conductivity as an output p', determines the application mode of the light signal in the light forming operation according to the initial electrical conductivity of the device before the q is input, and applies a green light signal to execute the light forming operation if p is 0; if p is 1, no green light signal is applied, and finally non-q logic is realized;

2) a NOR (NOR) logic that determines an application manner of the optical signal in the light forming operation according to an initial electrical conductive state of the device before the initial electrical conductive state of the memristor is used as an input p, a red optical signal is used as an input q, and a final electrical conductive state is used as an output p', q, and if p is 0, the green optical signal is applied to perform the light forming operation; if p is 1, no green light signal is applied, after q is input, the application mode of the control light signal is determined according to the initial electric conductivity of the device, and if p is 0, no red light signal is applied; if p is 1, applying a red light signal, and finally realizing NOR logic;

the sixth type: in addition to the need for photo-forming operations, the need to measure the initial and intermediate electrical conductivity states of the device;

1) exclusive-or (XOR) logic with the memristor initial electrical conductivity state as input p, the green signal as input q, and the final electrical conductivity state as output p'; before q is input, determining the application mode of the optical signal in the optical forming operation according to the initial electrical conductivity state of the device, and if p is 0, applying a green optical signal and not applying a red optical signal; if p is 1, a red signal is applied and a green signal is not applied. q, after inputting, determining the application mode of the following control optical signal according to the current electric conduction state of the device, and if the device is in a high electric conduction state, applying a red optical signal and not applying a green optical signal; if the voltage is in a low-conductivity state, applying a green light signal, not applying a red light signal, and finally realizing exclusive or logic;

2) exclusive-Nor (NXOR) logic to take the initial electrical conductivity state of the memristor as input p, a green signal as input q, and a final electrical conductivity state as output p'; before q is input, determining the application mode of the optical signal in the optical forming operation according to the initial electrical conductivity state of the device, and if p is 0, applying a green optical signal and not applying a red optical signal; if p is 1, a red signal is applied and a green signal is not applied. q, the application mode of the optical signal is determined and controlled according to the current electric conduction state of the device after the input. If the signal is in a high-conductivity state, applying a red light signal, and not applying a green light signal; if in the low electrical state, a green signal is applied and no red signal is applied. Reading the electric conduction state of the device again, and if the electric conduction state is the high electric conduction state, applying a red light signal and not applying a green light signal; if the current is in a low-conductivity state, a green light signal is applied, a red light signal is not applied, and finally p' is output to realize the same or logic.

The invention also provides a logic cascading mode:

the logic realizing method by using the all-optical control memristor is characterized in that the output of the previous step of logic operation is stored in a form of conductance, the conductance is used as an initial conductance input p, an optical signal is used as another input q, and the final conductance state is used as an output p', so that logic cascade connection is directly realized.

Compared with the prior art, the invention has the main advantages that:

1) in a single memristor, all-optical signals are used for replacing existing electric signals to reversibly regulate and control the conductance of the memristor, and the optical power density of the optical signals is in mu W/cm²The magnitude and the power consumption are low, the change of the microstructure of the device is not easy to cause, and compared with the traditional voltage regulation and control means, the stability and the reliability of the memristor device can be improved to a certain extent.

2) The regulation and control result of the electrical conductivity of the memristive device by using the optical signal is nonvolatile, namely after illumination is removed, the regulated and controlled electrical conductivity can be directly stored in the device for a long time without disappearance, the characteristic is combined with Boolean logic, the storage and operation of data are integrated, and the storage and energy consumption problems caused by frequent data transportation are greatly reduced; the memristor is simple in structure, has excellent expandability and high compatibility with modern complementary metal oxide semiconductor technology, and has practical application value.

3) By defining new logic input, one logic input is defined as an initial electric conduction state of the device, the other input is defined as an optical signal, and the logic output is defined as a final electric conduction state, the full 16 Boolean logic functions can be realized in a single memristive device, the device has advantages in the aspects of functional completeness and high efficiency, and in addition, the input signal and the output signal are both provided in the electric conduction form, and when the next logic step is carried out, initialization is not needed, logic cascade can be directly realized, and the working efficiency is greatly improved.

Drawings

Fig. 1 is a schematic structural diagram of a memristor of the present invention, in which: 1-a top electrode layer, 2-an intermediate dielectric layer, 3-a bottom electrode layer, and 4-a substrate;

FIG. 2 is a graph of current-voltage characteristics of memristors prepared in examples in the dark and after illumination with 530nm green light for 30s, respectively (step size: 10 mV);

FIG. 3 is a graph of the photoresponse of memristors used for realizing logic functions under different longer wavelengths after being irradiated by green light, wherein the visible light with the wavelength of 650nm-800nm is adopted as the optical signal with the longer wavelength, and the optical power densities are all 36 muW/cm²；

FIG. 4 is a photoresponse diagram of the decrease of the conductance of a memristor used for realizing logic functions in a high conductance state, wherein visible light of 650nm-800nm is adopted as an optical signal, and the optical power density is 36 muW/cm²；

FIG. 5 is a continuous reversible regulation diagram of the conductance of memristors used to implement logic functions under stimulation of optical signals; wherein the conductance value is continuously increased by adopting 50 green light pulses of 530 nm; the conductance was continuously decreased by using 50 red light pulses of 650 nm. The optical power density is 36 mu W/cm²(ii) a The pulse width and the pulse interval of the green light pulse are both 1s, and the pulse width and the pulse interval of the red light pulse are both 5 s;

fig. 6 is an experimental result diagram of an implementation method for implementing 16 boolean logic functions in a single memristor by using an optical signal according to an embodiment of the present invention, where (a) is an OR logic implementation result diagram, (b) is an AND logic implementation result diagram, (c) is an RIMP logic implementation result diagram, (d) is a NAND logic implementation result diagram, (e) is a NOT q logic implementation result diagram, AND (f) is an XOR logic implementation function diagram;

fig. 7 is a graph showing the results of the non-linear fitting when p' is 1 in 6 logic functions selected in the example, and the fitting function selected is y₀+Aexp(-t/τ₁)+Bexp(-t/τ₂)+Cexp(-t/τ₃) Wherein y is₀,A,B,C,τ₁,τ₂And τ₃Are all constants. Wherein (a)₁) Is OR logic where p is 0, q is 1, (a)₂) Is OR logic where p is 1, q is 0, (a)₃) The result of the nonlinear fitting when p is 1 and q is 1 in OR logic;

(b) the result of the nonlinear fitting when p is 1 AND q is 1 in AND logic is obtained;

(c₁) Is that p is 0, q is 0 in RIMP logic, (c)₂) Is that p is 1, q is 0 in RIMP logic, (c)₃) The result of nonlinear fitting when p is 1 and q is 1 in RIMP logic is obtained;

(d₁) For NAND logic, p is 0, q is 0, and (d)₂) For NAND logic, p is 0, q is 1, and (d)₃) The result of the nonlinear fitting when p is 1 and q is 0 in the NAND logic is obtained;

(e₁) NOT q logic p is 0, q is 0, (e)₂) The result of nonlinear fitting when NOT q logic p is 1 and q is 0;

(f₁) For XOR logic p ═ 0, q ═ 1, (f)₁) The result of the nonlinear fitting when XOR logic p is 1 and q is 0 is shown.

Detailed Description

The invention is further described with reference to the following drawings and specific examples. It should be understood that these examples are for illustrative purposes only and are not intended to limit the scope of the present invention. The following examples are conducted under conditions not specified, usually according to conventional conditions, or according to conditions recommended by the manufacturer.

In the present embodiment, a memristor structure is provided, as shown in fig. 1, and includes, from bottom to top, a substrate 4, a bottom electrode layer 3, an intermediate dielectric layer 2, and a top electrode layer 1, where optical signals are all input through the top electrode layer 1 in the following embodiments.

The substrate of the memristor is a thermal oxidation silicon wafer, the bottom electrode layer is made of platinum, and the thickness of the bottom electrode layer is 200 nm; the oxide layer material adopts a single-layer zinc oxide (ZnO) film prepared in a pure argon environment by magnetron sputtering, and the thickness of the single-layer zinc oxide (ZnO) film is 50 nm; the top electrode layer material is gold, and the thickness of the gold is 10 nm. Optical signals used in the experiment are all input through the top electrode layer and optical work under all wavelengthsThe rate density is 36 mu W/cm². Fig. 2 is a current-voltage characteristic curve before and after the memristor is illuminated, and the device is characterized in that the device is changed from a high-conductivity state to a low-conductivity state under positive and negative voltages. The overall conductance of the memristor increased after 30s of 530nm illumination compared to the curve measured in the dark. Fig. 3 shows the photoresponse characteristics of the memristor after being irradiated by green light at 530nm for 60s for different long wavelengths. It can be seen that illumination at 530nm increased the device conductance, and after light removal, the device exhibited a persistent photoconductive effect (PPC), as shown by the dark environment curve. In contrast, the 650nm-800nm light can reduce the conductance of the device, and the shorter the wavelength, the better the effect of the reduction of the conductance of the device. FIG. 4 is a photo-response characteristic of the memristor in a photo-erase mode. In the experiment, the green light of 530nm is firstly utilized to adjust the memristor to a higher electric conduction state, and then the light of 650nm-800nm is applied, so that the electric conduction of the memristor can be obviously reduced, and the electric conduction reduction effect is optimal under the wavelength of 650 nm. Therefore, the optical signal in the optical writing mode used in this embodiment is 530nm of green light, and the optical signal in the optical erasing mode is 650nm of red light. Fig. 5 shows that the memristor realizes continuous regulation and control of conductance under full optical signals, wherein 50 green light pulses of 530nm are used to realize continuous increase of the conductance of the device, 50 red light pulses of 650nm are used to realize continuous decrease of the conductance of the device, the pulse width and the pulse interval of the green light pulses are both 1s, and the pulse width and the pulse interval of the red light pulses are both 5 s. Based on the all-optical control memristor, 16 nonvolatile Boolean logic functions are realized by using different optical signals. We define inputs p and q as the initial electrical conduction state and the optical signal of the memristor, respectively, output p' as the final electrical conduction state of the device, the threshold is set to 100nS, the value of the electrical conduction is greater than 100nS for the logic variable "1", and less than 100nS for the logic variable "0"; the application of an optical signal represents a logical "1" and the absence of an optical signal represents a logical "0", which logical "1" includes two definitions: one is to apply 530nm green light to increase the memristor conductance, and the other is to apply 650nm red light to decrease the memristor conductance.

The 16 boolean logic functions are classified into 6 classes according to the complexity of the logic operation, and the following description will be made by selecting one example from each of the 6 classes:

fig. 6(a) is a diagram showing the result of implementing the OR logic function. The initial conductance and the final conductance of the device are respectively defined as an input p and an output p', and an optical signal of 530nm is used as an input q, when p is 0 and q is 0, the device is in a low-conductance state initially, and a result of 0 can be output without applying an additional control optical signal; when p is 0 and q is 1, the device is initially in a low-conductance state, then an input optical signal of 530nm is applied to increase the conductance of the device, the current is gradually attenuated due to a PPC effect generated after light removal, but the current is not attenuated to be below a threshold value within a long period of time, so that the final output result is '1'; similarly, the output results of p ═ 1, q ═ 0, p ═ 1 and q ═ 1 can be obtained by analogy, and according to the experimental results, only when two inputs are both "0", the final output result of the device is "0", and the rest of the output results are both "1".

Fig. 6(b) is a schematic diagram of implementing AND logic. The initial and final conductances of the device are defined as input p and output p', 530nm optical signals as input q, respectively, and in this logic a 650nm control optical signal is introduced. When p is 0 and q is 0, after the device is irradiated by an optical signal of 650nm from a low-conductivity state, the conductivity of the device is slightly increased, but the amplitude is low and cannot exceed the set threshold value of 100nS, so that the final logic output result is '0'; when p is 0 and q is 1, the device is initially in a low-conductance state, then an input optical signal of 530nm is applied to increase the conductance of the device, and after light is removed, the current is gradually attenuated to be below a threshold value due to a PPC effect, so that an output result of 1 is obtained; similarly, output results of p ═ 1, q ═ 0, p ═ 1, and q ═ 1 can be obtained, and according to experimental results, only when two inputs are both "1", the final output result of the device is "1", and the rest of the output results are both "0".

FIG. 6(c) is a functional diagram of the RIMP logic. An optical signal of 650nm is used as input q, and light of 530nm is used as a control optical signal. Before q is input, the conductance of the device is changed to a high conductance state by using an optical signal of 530nm, and a photo-forming operation is performed. For example, when p is 0 and q is 0, the device changes from a low-conductivity state to a high-conductivity state upon illumination with 530nm light, resulting in a final logic output of "1"; when p is 0 and q is 1, the device increases the conductance of the device by an optical signal of 530nm from a low-conductance state, then inputs an optical signal of q, namely 650nm, and the final output result is 0 as the application of 650nm light energy in a high-conductance state reduces the conductance of the device; similarly, output results of p ═ 1, q ═ 0, and p ═ 1, and q ═ 1 can be obtained, and only when p ═ 0, and q ═ 1, the output result of the device is "0", and the remaining output results are "1".

FIG. 6(d) is a functional diagram of the NAND logic. A 650nm red signal is used as input q, a 530nm green light is used as control light signal and is determined by the initial electrical conductivity of the device, and if p is 0, the 530nm light signal is applied, otherwise it is not applied. Specifically, when p is 0 and q is 0, the device is changed from a low-conductivity state to a high-conductivity state by the irradiation of light of 530nm from the low-conductivity state, and a logic result "1" is finally output; similarly, output results are obtained where p is 0, q is 1, p is 1, q is 0 and p is 1, q is 1, and only when p is 1 and q is 1, the output result is "0" and the remaining output results are "1".

Fig. 6(e) is a functional diagram of implementing NOT q logic. Before a 650nm red signal is input as input q, a light forming operation is performed using a 530nm green signal and is determined by the initial electrical conductivity of the device, and if p is 0, the green signal is applied, otherwise, it is not applied. Specifically, when p is 0 and q is 0, the device is irradiated by 530nm light from the low-conductivity state to change the device conductivity from the low-conductivity state to the high-conductivity state, and finally a logic result "1" is output; similarly, output results of p ═ 0, q ═ 1, p ═ 1, q ═ 0, and p ═ 1, and q ═ 1 can be obtained, and only when p and q are the same, the output result is "1", and the remaining output results are "0".

FIG. 6(f) is a schematic diagram of the implementation of the XOR logic function. Implementing this logic requires not only the photo-forming operation and reading of the initial electrical conductivity state of the device, but also the reading of the intermediate electrical conductivity state of the device. Specifically, a light signal of 530nm is taken as an input q, green light of 530nm plus red light of 650nm is taken as a set of control light signals, the first set of control light signals being determined by the initial electrical state of the device, and a light forming operation is performed. If p is 0, applying a green light signal and not applying a red light signal; if p is 1, a red signal is applied and a green signal is not applied. The other group of control optical signals is determined by the electrical conductivity of the device read after the input q, and if the control optical signals are in a high electrical conductivity state, the control optical signals apply red light signals and do not apply green light signals; if the device is in a low electrical conductivity state, then a green signal is applied and a red signal is not applied. Specifically, when p is 0 and q is 0, the device is first changed from a low-conductivity state to a high-conductivity state by green light irradiation, and then changed to the low-conductivity state by a red light signal, so as to obtain an output result of "0"; similarly, an output result of p ═ 0, q ═ 1, p ═ 1, q ═ 0, and p ═ 1, and q ═ 1 can be obtained, and when p and q are the same, the output result is "0", and p and q are different, the output result is "1".

Due to the good PPC effect of the device, the conductance of the device is gradually attenuated after light is removed. In the case where p' is 0, the attenuation causes the device conductance to become lower and lower than the set threshold (100nS), and the final output result is not affected. However, in the case of p' 1, the continuous decay of the device conductance brings the final conductance state closer to the threshold (100nS), and it is likely or not that the final conductance state may decay below the threshold for a longer time, which may cause inaccuracy in the logic output result. In order to confirm that the attenuation of the device conductance does not affect the final output result of the logic, the curve of 6 types of logic with p' 1 in the embodiment is fitted, and the fitting result is shown in fig. 7, and the fitting function selected is y-y₀+Aexp(-t/τ₁)+Bexp(-t/τ₂)+Cexp(-t/τ₃) Wherein y is₀,A,B,C,τ₁,τ₂And τ₃Are all constant, wherein, as shown in FIG. 7 (a)₁) FIG. 7 (f)₂) The result shows that the corresponding device electric conductivity state can be stabilized above the threshold value for a long time when p' is 1 in 6 logics, thereby ensuring the accurate output of the logics.

Furthermore, it should be understood that various changes and modifications can be made by one skilled in the art after reading the above description of the present invention, and equivalents also fall within the scope of the invention as defined by the appended claims.