A semiconductor-ferroelectric gradient heterojunction-based light sensing and computing integrated device, a preparation method therefor, and an application thereof

Abstract

The application discloses a kind of based on semiconductor-ferroelectric gradient heterojunction photo sensing and computing integrated device and its preparation method and application, by introducing ferroelectric material in semiconductor, in combination with the accurate control of surface-OH content of conductive substrate, composition gradient distribution of ferroelectric material in semiconductor is realized, and then high-precision modulation of the responsivity of photo sensing and computing integrated device is achieved, and its output signal is accurately controlled.The application greatly improves the reconfigurable state number of photo sensing and computing integrated device, significantly enhances the computing capacity and precision of the device, and finally provides an effective implementation scheme for the development of high-precision in-sensor computing device.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of light sensing and computing integrated devices, and particularly relates to a light sensing and computing integrated device based on a semiconductor-ferroelectric gradient heterojunction and a preparation method and application thereof. BACKGROUND

[0002] With the rapid development of automatic driving, machine vision and other fields, the complexity and data volume of image processing tasks are growing exponentially. Traditional image processing systems adopt an architecture in which sensors, memories and processing units are physically separated, and signals need to be frequently transmitted between units, which not only leads to significant energy loss and data delay, but also increases hardware costs due to the need for multi-component integration, making it difficult to adapt to power and resource limited application scenarios such as edge computing and mobile devices.

[0003] To solve the above problems, the prior art designs a visual sensor simulating the morphology of human retina, which integrates an artificial neural network to perform signal accumulation operations with reconfigurable response characteristics as neural network weights, realizing the preliminary fusion of perception and calculation. However, the core limitation of existing such sensors is that most implementation schemes need to rely on continuous gate voltage maintenance function: the weight value is controlled by an external power supply, which causes the weight adjustment accuracy to be disconnected from the physical characteristics of the device itself, and the weight needs to be stored remotely and transmitted to each sensor unit through complex external circuits, which seriously hinders the direct integration of storage and calculation functions inside the sensor. At the same time, the independent control of each neural network unit needs to rely on complex supporting circuits, which makes it neither practically feasible nor easy to achieve long-term stable operation when deploying such schemes on a large scale in power and resource limited environments, seriously restricting their large-scale application process.

[0004] In-sensor computing technology integrates perception, storage and processing functions in a single device module, and mechanisms such as ferroelectric polarization, ion migration and ion-induced local field modulation have become a research hotspot, providing an important idea to solve the bottleneck of traditional architecture. Such systems achieve weight storage by regulating the polarization intensity or the proportion of different state regions, but the core challenge is how to realize continuous physical modulation of the state to ensure the accuracy of weight adjustment. In current neuromorphic computing research, weight processing relies mainly on curve fitting or interpolation methods, which do not fully verify the physical feasibility of the assumed weights in actual devices, resulting in overestimation of computing performance, and ignoring the direct impact of weight adjustment accuracy on neuromorphic computing accuracy. More critically, even under optimistic estimates, the number of states that existing systems can distinguish is still limited to less than 6 bits, making it difficult to meet the demand for computing accuracy in high-precision image processing, intelligent perception and other scenarios, and restricting the industrialization application process of in-sensor computing technology. SUMMARY

[0005] To solve the technical problems of the existing sensor internal computing system that the weight adjustment precision is out of touch with physical feasibility and the number of distinguishable states is limited, the application aims to provide a photoreception and calculation integrated device based on a semiconductor-ferroelectric gradient heterojunction and a preparation method and application thereof, specifically by introducing ferroelectric materials into the semiconductor, combining with the accurate regulation of the content of -OH on the surface of the conductive substrate, realizing the composition gradient distribution of the ferroelectric materials in the semiconductor, and then achieving high-precision modulation of the responsivity of the photoreception and calculation integrated device, effectively improving the number of reconfigurable states of the photoreception and calculation integrated device, significantly enhancing the computing ability and precision of the device, and finally providing an effective implementation scheme for the development of high-precision in-sensor computing devices.

[0006] The above-mentioned purpose of the application is realized by the following technical solutions.

[0007] The application provides a preparation method of a photoreception and calculation integrated device based on a semiconductor-ferroelectric gradient heterojunction, which comprises the following steps:

[0008] (1) The surface of the conductive substrate is subjected to hydrophilic treatment, and the contact angle of the surface of the treated conductive substrate is 15°-60°;

[0009] (2) The semiconductor solution and the ferroelectric material solution are mixed to prepare a photoelectric-ferroelectric material mixed solution; the semiconductor is perovskite, lead sulfide (PbS) or lead iodide (PbI2); the ferroelectric material is polyvinylidene fluoride-trifluoroethylene (P(VDF-TrFE)), barium titanate (BaTiO3) or bismuth ferrite (BiFeO3); the concentration of the ferroelectric material in the ferroelectric material solution is 0.05-0.2 M; and the volume ratio of the semiconductor solution to the ferroelectric material solution is (200-400):(8-15);

[0010] (3) The photoelectric-ferroelectric material mixed solution is coated on the surface of the hydrophilic-treated conductive substrate to form a semiconductor-ferroelectric gradient heterojunction layer;

[0011] (4) A metal top electrode is evaporated on the semiconductor-ferroelectric gradient heterojunction layer to obtain the photoreception and calculation integrated device based on the semiconductor-ferroelectric gradient heterojunction.

[0012] The present application adopts a solution processing process, which is simple to operate, controllable in cost, and easy to mass produce; by regulating and controlling the content of -OH on the surface of the conductive substrate (characterized by the contact angle), the composition gradient distribution of the ferroelectric material in the semiconductor is accurately realized, without relying on complex and precise equipment, and the process compatibility is excellent; and the parameters of each step are convenient to regulate and control, which can stably reproduce the high-consistency semiconductor-ferroelectric gradient heterojunction structure, effectively guaranteeing the performance stability and reliability of the device batch production. On this basis, the present application further proposes and realizes a polarization energy focusing strategy, which is realized by accurately designing the composition gradient distribution structure of the ferroelectric material in the semiconductor, and establishes the correlation mechanism between the composition distribution of the semiconductor-ferroelectric and the adjustment accuracy of the calculation weight in the sensor.

[0013] Further, in step (1), the surface hydrophilic treatment adopts an ozone cleaning method, and the treatment equipment is a surface ozone cleaning machine.

[0014] Further, in step (1), the conductive substrate is a transparent conductive substrate, such as indium tin oxide (ITO), fluorine-doped tin oxide (FTO), or aluminum-doped zinc oxide (AZO).

[0015] Further, in step (1), the surface hydrophilic treatment time is 5-30 min.

[0016] Preferably, the contact angle of the treated conductive substrate surface is 15°-40°.

[0017] Further, in step (2), the semiconductor solution is a perovskite precursor solution, a lead sulfide solution, or a lead iodide solution. Perovskite material is the core choice for optical sensing devices due to its high light absorption coefficient and excellent carrier mobility; lead iodide, as a core component of perovskite, can be used alone as a semiconductor matrix, which meets the needs of simplifying the device structure.

[0018] Further, the preparation method of the perovskite precursor solution comprises the following steps: dissolving lead iodide (PbI2) and formamidinium iodide (FAI) in an organic solvent to obtain solution A, the organic solvent is N,N-dimethylformamide (DMF) and / or dimethyl sulfoxide (DMSO); adding cesium iodide solution to solution A, stirring at 70-100 ℃ for 15-20 min to obtain the perovskite precursor solution.

[0019] Further, the concentration of PbI2 in the perovskite precursor solution is 1-2 M, and the concentration of FAI is 1-2 M.

[0020] Further, the solution A further comprises lead bromide (PbBr2) and / or methylamine bromide (MABr).

[0021] Further, the concentration of PbBr2 in the perovskite precursor solution is 0.2-0.5 M, and the concentration of MABr is 0.2-0.5 M.

[0022] Further, the volume ratio of the solution A to the cesium iodide solution is (80-130):(4-6).

[0023] Further, the cesium iodide solution is prepared by dissolving cesium iodide (Csl) in DMSO.

[0024] Further, the concentration of the cesium iodide solution is 1-5 M.

[0025] Further, in step (2), the ferroelectric material solution is prepared by dissolving a ferroelectric material in an organic solvent, and the organic solvent is DMF and / or DMSO.

[0026] As an organic ferroelectric material, P(VDF-TrFE) has the advantages of good flexibility and easy solution processing; as inorganic ferroelectric materials, BaTiO3 and BiFeO3 have high polarization strength and excellent stability, and the ferroelectric material can be flexibly selected according to the flexible requirements of the device, the stability of the working environment, and other application scenarios.

[0027] In a specific embodiment, in step (2), the ferroelectric material is dissolved in an organic solvent, and stirred for 20-30 min to obtain a ferroelectric material solution.

[0028] Further, in step (2), the semiconductor solution and the ferroelectric material solution are stirred for 10-20 min for mixing.

[0029] The introduction of the ferroelectric material solution into the semiconductor solution enables the mixed system to have adjustable ferroelectric polarization function, thereby forming a photoelectric-ferroelectric material mixed solution. The polarization direction and strength of the ferroelectric material can be regulated by an external electric field, thereby changing the carrier transport characteristics of the semiconductor, and finally realizing the reconfigurability of the device response.

[0030] Further, in step (3), the photoelectric-ferroelectric material mixed solution is spin-coated on the surface of the hydrophilic-treated conductive substrate, and a semiconductor-ferroelectric gradient heterojunction layer is obtained after annealing treatment.

[0031] In a specific embodiment, a two-step spin-coating method is adopted, chlorobenzene is added as an anti-solvent before the end of spin-coating, and annealing treatment is performed at 80-120 ℃ for 10-20 min after the end of spin-coating.

[0032] Further, in step (4), the speed of evaporating the metal top electrode is 0.1-0.3 Å / s.

[0033] Further, in step (4), the metal top electrode is a Cu electrode, an Au electrode, or a Cu / Au mixed electrode.

[0034] Further, the thickness of the Cu electrode is 1-3 nm, and the thickness of the Au electrode is 7-10 nm.

[0035] Further, the Cu / Au mixed electrode is a transparent electrode, and high light transmittance can be achieved.

[0036] In the specific embodiment, a Cu layer with a thickness of 1-3 nm is first deposited on the surface of the semiconductor-ferroelectric gradient heterojunction layer at a speed of 0.1-0.3 Å / s, and then an Au layer with a thickness of 7-10 nm is deposited on the surface of the Cu layer at the same speed (0.1-0.3 Å / s), so as to realize the preparation of the Cu / Au mixed electrode.

[0037] The second aspect of the present application provides a semiconductor-ferroelectric gradient heterojunction-based photo-sensing and computing integrated device prepared by the preparation method of the first aspect.

[0038] The semiconductor-ferroelectric gradient heterojunction-based photo-sensing and computing integrated device provided by the present application is a reconfigurable photodiode, and the core advantage is derived from the synergistic effect of the semiconductor-ferroelectric gradient heterojunction: the gradient distribution of the ferroelectric material forms a continuously changing polarization electric field in the device, which can be precisely controlled by an external pulse voltage, thereby realizing gradient modulation of carrier transport efficiency, and finally achieving high-precision reconfiguration of responsivity. Practical tests show that the linearity of the device is as high as 0.999, and 1024 distinguishable states (corresponding to 10 bits of precision) are supported, which is far superior to existing similar devices (usually 4-6 bits), and can realize more precise weight modulation and more complex computing tasks. Based on a 3x3 device array, the present application successfully demonstrates the synchronous implementation of in-situ image sensing and processing: each device in the array corresponds to a pixel point of the image, and by independently controlling the responsivity of each device (which is mapped to the matrix elements of the image processing kernel function), the sensing and processing operations can be completed simultaneously without external data transmission and computing units.

[0039] The third aspect of the present application provides an application of the semiconductor-ferroelectric gradient heterojunction-based photo-sensing and computing integrated device of the second aspect in neuromorphic computing, such as image processing.

[0040] Further, the image processing method of the semiconductor-ferroelectric gradient heterojunction-based photo-sensing and computing integrated device comprises the following steps:

[0041] S1, under the irradiation of a light source, a pulse voltage is applied to the semiconductor-ferroelectric gradient heterojunction-based photo-sensing and computing integrated device for modulation;

[0042] S2, detecting and recording the current of the photo-sensing and computing integrated device based on the semiconductor-ferroelectric gradient heterojunction after the pulse voltage modulation, and calculating the responsivity;

[0043] S3, taking the responsivity as a matrix element of an image processing kernel function to realize the image processing.

[0044] The photo-sensing and computing integrated device based on the semiconductor-ferroelectric gradient heterojunction can realize high-precision image processing.

[0045] Further, in S1, the wavelength of the light source is 600-700 nm, preferably 650 nm.

[0046] Further, in S1, the pulse voltage is applied by using a semiconductor analyzer.

[0047] Further, in S1, the amplitude of the pulse voltage is 1-3 V, the pulse width is 0.5-2 ms, and the pulse frequency is 0.1-1 Hz.

[0048] Further, in S2, the semiconductor analyzer is used to detect and record the current.

[0049] Further, in S2, the calculation method of the responsivity is R=I / (PxA), wherein R is the responsivity (unit: A / W), I is the device current value, P is the incident light power, and A is the effective light receiving area of the device.

[0050] In the present application, the light responsivity of the photo-sensing and computing integrated device can be directly mapped to the light weight value in the neuromorphic computing, and the precise modulation of the weight is realized by regulating the responsivity.

[0051] Further, in S3, the kernel function of the image processing is a 3x3 matrix (adapted to a 3x3 device array).

[0052] Further, in S3, the image processing includes at least one of image sharpening processing, image embossing processing, and image contour extraction processing. The specific implementation principle is that the matrix elements of different processing kernel functions (such as the Laplacian kernel for sharpening, the gradient kernel for embossing, and the Sobel kernel for contour extraction) are one-to-one mapped with the responsivity of the device, and the responsivity of each device is regulated to make the signal output by the array be the processed image signal directly, without the need for external algorithms and computing units.

[0053] In the specific embodiment, the image processing method includes the following steps:

[0054] S1, under the illumination of a 650 nm light source, a series of pulse voltages are generated by using a semiconductor analyzer to modulate the photo-sensing and computing integrated device;

[0055] S2. Collect the current values ​​of the photosensitive computing device before and after pulse modulation using a semiconductor analyzer, and complete the responsivity calibration according to the formula R=I / (P×A);

[0056] S3. In the image processing process, the matrix elements of the target kernel function are mapped and matched with the responsivity to ultimately achieve customized image processing.

[0057] In a specific implementation, the semiconductor analyzer used is the Keithley-2400.

[0058] Compared with the prior art, the above-described technical solution of the present invention has the following advantages:

[0059] 1. This invention achieves precise design of the compositional gradient distribution of ferroelectric materials in semiconductors by accurately controlling the -OH content on the surface of a transparent conductive substrate. This significantly increases the number of reconfigurable states of the computing devices within the sensor. Combined with 10-bit precision reconfigurable states and a high linearity of 0.999, it significantly improves the accuracy and complexity of in-sensor computing, enabling it to support more advanced image processing and pattern recognition tasks.

[0060] 2. This invention integrates sensing, storage and processing functions into a single module, effectively reducing the need for external data transmission and significantly reducing device power consumption. This allows the device to complete real-time high-quality image processing without relying on external computing hardware, adapting to the application scenarios requiring low power consumption and high integration.

[0061] 3. This invention uses mature processes such as solution treatment and vacuum evaporation, which do not require complex and precision equipment, are easy to operate and have controllable costs; the parameters of each step are easy to control, and the device structure with high consistency can be stably reproduced, making it easy to scale up mass production. Attached Figure Description

[0062] Figure 1 The images show the X-ray diffraction (XRD) patterns, photoluminescence (PL) patterns, and surface scanning electron microscope (SEM) images of the perovskite-P(VDF-TrFE) gradient heterojunction layer (doped P(VDF-TrFE)) in Example 1 and the perovskite semiconductor layer (undoped P(VDF-TrFE)) in Comparative Example 1; where a is a comparison of XRD patterns, b is a comparison of PL patterns, c is a surface SEM image of the perovskite-P(VDF-TrFE) gradient heterojunction layer, and d is a surface SEM image of the perovskite semiconductor layer.

[0063] Figure 2 The images show a schematic diagram and a cross-sectional SEM image of the photosensitive computing device based on a perovskite-P(VDF-TrFE) gradient heterojunction prepared in Example 1; where a is a schematic diagram and b is a cross-sectional SEM image.

[0064] Figure 3 The ITO substrate water contact angle data graph and internal element longitudinal distribution graph of the photodetector-integrated device based on the perovskite-P(VDF-TrFE) gradient heterojunction prepared in Examples 1-4 and the photodetector-integrated device based on the perovskite-P(VDF-TrFE) heterojunction prepared in Comparative Example 2; wherein a is the ITO substrate water contact angle data graph, b is the internal element longitudinal distribution graph of Example 1, c is the internal element longitudinal distribution graph of Example 2, d is the internal element longitudinal distribution graph of Example 3, e is the internal element longitudinal distribution graph of Example 4, and f is the internal element longitudinal distribution graph of Comparative Example 2.

[0065] Figure 4 The response degree reconfigurable change curve graph, memory time test result graph of reconfigurable response degree, and response speed test result graph of the photodetector-integrated device based on the perovskite-P(VDF-TrFE) gradient heterojunction prepared in Example 1 under different pulse voltage stimuli; wherein a is the response degree reconfigurable change curve graph under different pulse voltage stimuli, b is the memory time test result graph of reconfigurable response degree, c is the response speed test result graph under positive response degree, and d is the response speed test result graph under negative response degree.

[0066] Figure 5 The structure schematic diagram, XRD graph of the gradient heterojunction layer, and the reconfigurable response degree change curve graph of the photodetector-integrated device based on the PbI2-BaTiO3 gradient heterojunction prepared in Example 5 and the photodetector-integrated device based on the PbS-BiFeO3 gradient heterojunction prepared in Example 6; wherein a is the structure schematic diagram of the photodetector-integrated device based on the PbI2-BaTiO3 gradient heterojunction and the XRD graph of the PbI2-BaTiO3 gradient heterojunction layer, b is the reconfigurable response degree change curve graph of the photodetector-integrated device based on the PbI2-BaTiO3 gradient heterojunction, c is the structure schematic diagram of the photodetector-integrated device based on the PbS-BiFeO3 gradient heterojunction and the XRD graph of the PbS-BiFeO3 gradient heterojunction layer, and d is the reconfigurable response degree change curve graph of the photodetector-integrated device based on the PbS-BiFeO3 gradient heterojunction.

[0067] Figure 6 The response degree matrix and effect display graph of the device array for different image processing in Application Examples 1-3; wherein a is sharpening processing, b is embossing processing, and c is contour extraction processing. DETAILED DESCRIPTION

[0068] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the description herein is for describing particular embodiments only and is not intended to be limiting of the application. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0069] The present application is further described so that examples thereof can be better understood by reference to the following figures and examples, but the present application should not be construed as limited thereto.

[0070] The experimental methods used in the following examples are routine methods unless otherwise specified, and the materials, reagents, etc. used are commercially available unless otherwise specified.

[0071] Example 1

[0072] A preparation method of a photodetector-integrated device based on a perovskite-P(VDF-TrFE) gradient heterojunction, comprising the following steps:

[0073] (1) The ITO substrate was placed in an ultraviolet-ozone environment and treated with a surface ozone cleaning machine for 20 min to make the ITO substrate surface hydrophilic. The water contact angle of the treated ITO substrate surface was 15°.

[0074] (2) 200 μL of perovskite precursor solution was mixed with 10 μL of P(VDF-TrFE) solution with a concentration of 0.1 M and stirred for 10 min to prepare a perovskite-P(VDF-TrFE) mixed solution; the preparation method of the perovskite precursor solution (wherein the PbI2 concentration is 1.15 M, the FAI concentration is 1.1 M, the PbBr2 concentration is 0.2 M, and the MABr concentration is 0.2 M) is as follows: PbI2, FAI, PbBr2, and MABr were dissolved in 0.96 mL of DMF, 40 μL of CsI solution (solvent: DMSO) with a concentration of 1.5 M was added, and stirred at 70 ℃ for 20 min; the preparation method of the P(VDF-TrFE) solution is as follows: P(VDF-TrFE) was dissolved in DMF and stirred for 20 min.

[0075] (3) The perovskite-P(VDF-TrFE) mixed solution is spin-coated on the surface of the hydrophilic treated ITO substrate by a two-step spin-coating method, first at a speed of 1000 rpm for 10 s, and then at a speed of 4000 rpm for 35 s. 25 s before the end of spin-coating, 100 μL of chlorobenzene is added as an anti-solvent, and after spin-coating, annealing treatment is carried out at 100 °C for 10 min to form a perovskite-P(VDF-TrFE) gradient heterojunction layer.

[0076] (4) A 1 nm thick Cu layer is first deposited on the surface of the perovskite-P(VDF-TrFE) gradient heterojunction layer at a speed of 0.1 Å / s, and then a 10 nm thick Au layer is deposited on the surface of the Cu layer at a speed of 0.1 Å / s to realize the preparation of a Cu / Au hybrid electrode, and a light-sensing and calculating integrated device based on a perovskite-P(VDF-TrFE) gradient heterojunction is obtained.

[0077] Example 2

[0078] A method for preparing a light-sensing and calculating integrated device based on a perovskite-P(VDF-TrFE) gradient heterojunction, which is basically the same as that of Example 1, except that in step (1), the ITO substrate is treated in a UV-ozone environment using a surface ozone cleaning machine for 15 min to make the surface of the ITO substrate hydrophilic, and the water contact angle of the treated ITO substrate surface is 25°.

[0079] Example 3

[0080] A method for preparing a light-sensing and calculating integrated device based on a perovskite-P(VDF-TrFE) gradient heterojunction, which is basically the same as that of Example 1, except that in step (1), the ITO substrate is treated in a UV-ozone environment using a surface ozone cleaning machine for 10 min to make the surface of the ITO substrate hydrophilic, and the water contact angle of the treated ITO substrate surface is 35°.

[0081] Example 4

[0082] A method for preparing a light-sensing and calculating integrated device based on a perovskite-P(VDF-TrFE) gradient heterojunction, which is basically the same as that of Example 1, except that in step (1), the ITO substrate is treated in a UV-ozone environment using a surface ozone cleaning machine for 5 min to make the surface of the ITO substrate hydrophilic, and the water contact angle of the treated ITO substrate surface is 40°.

[0083] Example 5

[0084] A method for preparing a light-sensing and calculating integrated device based on a PbI2-BaTiO3 gradient heterojunction, comprising the following steps:

[0085] (1) The FTO substrate was placed in an ultraviolet-ozone environment and treated with a surface ozone cleaning machine for 20 min to make the FTO substrate surface hydrophilic. The contact angle of the FTO substrate surface after treatment was 15°.

[0086] (2) 200 μL of PbI2 solution was mixed with 10 μL of BaTiO3 solution with a concentration of 0.1 M and stirred for 10 min to prepare a PbI2-BaTiO3 mixed solution. The preparation method of the PbI2 solution was as follows: 1 M PbI2 was dissolved in 1 mL of DMF and stirred at 70°C for 20 min. The preparation method of the BaTiO3 solution was as follows: BaTiO3 nanoparticles were dispersed in DMF and stirred for 20 min.

[0087] (3) The PbI2-BaTiO3 mixed solution was spin-coated on the surface of the hydrophilic treated FTO substrate at a speed of 2000 rpm for 10 s, and then annealed at 100°C for 10 min after spin-coating to form a PbI2-BaTiO3 gradient heterojunction layer.

[0088] (4) A 1 nm thick Cu layer was first deposited on the surface of the PbI2-BaTiO3 gradient heterojunction layer at a speed of 0.1 Å / s, and then a 10 nm thick Au layer was deposited on the surface of the Cu layer at a speed of 0.1 Å / s to realize the preparation of a Cu / Au hybrid electrode, and a light-sensing and counting integrated device based on the PbI2-BaTiO3 gradient heterojunction was obtained.

[0089] Example 6

[0090] A method for preparing a light-sensing and counting integrated device based on a PbS-BiFeO3 gradient heterojunction, comprising the following steps:

[0091] (1) The FTO substrate was placed in an ultraviolet-ozone environment and treated with a surface ozone cleaning machine for 20 min to make the FTO substrate surface hydrophilic. The contact angle of the FTO substrate surface after treatment was 15°.

[0092] (2) 200 μL of PbS precursor solution was mixed with 10 μL of BiFeO3 solution with a concentration of 0.1 M and stirred for 10 min to prepare a PbS-BiFeO3 mixed solution; the preparation method of the PbS precursor solution was as follows: 0.6 mol of PbS and 0.2 mol of DPhTA (N,N'-diphenyl thiourea) were dissolved in 9 mL of DMF, after the solid was dissolved, 1 mL of n-butylamine was injected, then 3 mL of toluene was used as an anti-solvent, the solution was centrifuged at 9500 rpm for 15 min, and the precipitate was completely dried under a nitrogen atmosphere, and then the PbS precursor solution with a concentration of 500 mg / mL was prepared with DMF; the preparation method of the BiFeO3 solution was as follows: BiFeO3 nanoparticles were dispersed in 1 mL of DMF and stirred for 20 min.

[0093] (3) The PbS-BiFeO3 mixed solution was spin-coated on the surface of the hydrophilic treated FTO substrate at a speed of 2000 rpm for 10 s, and then annealed at 100 ℃ for 10 min after spin-coating to form a PbS-BiFeO3 gradient heterojunction layer.

[0094] (4) A 1 nm thick Cu layer was first deposited on the surface of the PbS-BiFeO3 gradient heterojunction layer at a speed of 0.1 Å / s, and then a 10 nm thick Au layer was deposited on the surface of the Cu layer at a speed of 0.1 Å / s to prepare a Cu / Au hybrid electrode, thereby obtaining a PbS-BiFeO3 gradient heterojunction based light-sensing and counting integrated device.

[0095] Comparative Example 1

[0096] A preparation method of a perovskite semiconductor based light-sensing and counting integrated device, comprising the following steps:

[0097] (1) The ITO substrate was placed in an ultraviolet-ozone environment and treated with a surface ozone cleaning machine for 20 min to make the ITO substrate surface hydrophilic, and the contact angle of the treated ITO substrate surface was 15°.

[0098] (2) PbI2, FAI, PbBr2 and MABr were dissolved in 0.96 mL of DMF, 40 μL of CsI solution with a concentration of 1.5 M (solvent: DMSO) was added, and stirred at 70 ℃ for 20 min to obtain a perovskite precursor solution, wherein the concentration of PbI2 was 1.15 M, the concentration of FAI was 1.1 M, the concentration of PbBr2 was 0.2 M, and the concentration of MABr was 0.2 M.

[0099] (3) The perovskite precursor solution was spin-coated on the surface of the hydrophilic treated ITO substrate by a two-step spin-coating method, first at a speed of 1000 rpm for 10 s, and then at a speed of 4000 rpm for 35 s. 25 s before the end of spin-coating, 100 μL of chlorobenzene was added as an anti-solvent, and after spin-coating, annealing treatment was carried out at 100 °C for 10 min to form a perovskite semiconductor layer.

[0100] (4) A 1 nm thick Cu layer was first deposited on the surface of the perovskite semiconductor layer at a speed of 0.1 Å / s, and then a 10 nm thick Au layer was deposited on the surface of the Cu layer at a speed of 0.1 Å / s to realize the preparation of a Cu / Au hybrid electrode, and a light-sensing and counting integrated device based on a perovskite semiconductor was obtained.

[0101] Comparative Example 2

[0102] A method for preparing a light-sensing and counting integrated device based on a perovskite-P(VDF-TrFE) heterojunction, which is basically the same as Example 1, the difference is that in step (1), the ITO substrate is not treated by a surface ozone cleaning machine, and the water contact angle of the ITO substrate surface is 65°.

[0103] Test Example 1

[0104] The light-sensing and counting integrated devices prepared in Examples 1-6 and Comparative Examples 1-2 were characterized for thin film properties, composition distribution and device structure, and the test results are as follows:

[0105] Figure 1 The XRD patterns, PL patterns and surface SEM patterns of the perovskite-P(VDF-TrFE) gradient heterojunction layer (doped with P(VDF-TrFE)) in Example 1 and the perovskite semiconductor layer (not doped with P(VDF-TrFE)) in Comparative Example 1, Figure 1 reflect the changes in the properties of the thin film material before and after P(VDF-TrFE) doping. As can be seen from Figure 1 , the XRD spectra of the two groups of materials have no obvious difference, indicating that the introduction of P(VDF-TrFE) does not change the crystallization characteristics of the perovskite, and proves that the doping of the ferroelectric material does not destroy the integrity of the crystal structure of the perovskite. As can be seen from Figure 1 , the PL spectra of the perovskite-P(VDF-TrFE) gradient heterojunction layer and the perovskite semiconductor layer have no difference except intensity, indicating that the doping of P(VDF-TrFE) does not change the band gap width of the perovskite, nor does it introduce additional defect states, which ensures that the intrinsic photoelectric properties of the perovskite are not affected. As can be seen from Figure 1As can be seen from c and d, there is no significant difference between the two morphologies, indicating that the introduction of P(VDF-TrFE) does not significantly change the grain size and morphology of the perovskite, and does not introduce defects such as pinholes that affect the performance of the device, ensuring the density and integrity of the thin film.

[0106] Figure 2 The structure diagram and cross-sectional SEM diagram of the optical and arithmetic integrated device based on the perovskite-P(VDF-TrFE) gradient heterojunction prepared in Example 1 are shown in FIG. 2. As can be seen from FIG. 2a, the optical and arithmetic integrated device is in a photodiode configuration, which can guarantee excellent light response performance of the device. As can be seen from FIG. 2b, the optical and arithmetic integrated device has tight contact and clear interface between each functional layer, and the perovskite layer has good morphology characteristics without obvious interlayer peeling or defects, which effectively guarantees the efficient transport of carriers between layers and provides structural support for high responsivity and high stability of the device. Figure 2 Figure 2

[0107] Figure 3 The water contact angle data diagram of the ITO substrate of the optical and arithmetic integrated device based on the perovskite-P(VDF-TrFE) gradient heterojunction prepared in Examples 1-4 and the optical and arithmetic integrated device based on the perovskite-P(VDF-TrFE) heterojunction prepared in Comparative Example 2, and the internal element longitudinal distribution spectrum tested by secondary ion mass spectrometry are shown in FIG. 3. As can be seen from FIG. 3a, as the ultraviolet ozone treatment time is shortened, the water contact angle of the ITO substrate gradually increases, and the content of -OH groups on the substrate surface gradually decreases, proving that the content of -OH on the substrate surface can be precisely controlled by adjusting the ozone treatment time. As can be seen from FIG. 3b-f, as the content of -OH on the ITO substrate surface decreases, the distribution of fluorine (F) elements (characteristic elements of P(VDF-TrFE)) in the perovskite gradually increases from top to bottom, and transitions to uniform distribution in the internal film, directly verifying the regulation mechanism of the content of -OH on the substrate on the gradient distribution of ferroelectric materials. Figure 3 Figure 3

[0108] Application Example 1

[0109] The optical and arithmetic integrated device based on the perovskite-P(VDF-TrFE) gradient heterojunction prepared in Example 1 was subjected to image sharpening (Sharpen) processing, and the processing method included the following steps:

[0110] S1, under the illumination of a 650 nm light source, a pulse voltage with an amplitude of 1.8 V, a pulse width of 1 ms, and a pulse frequency of 0.2 Hz was generated by the pulse mode of Keithley-2400 to modulate the optical and arithmetic integrated device;

[0111] ​​​​S2, the current value of the photo-sensing and computing integrated device before and after different pulse modulation was collected by Keithley-2400, and the responsivity was calibrated according to the formula R=I / (PxA);

[0112] S3, the light weight value (i.e. the responsivity) of the device was regulated by applying different numbers of pulse voltages, and a 3x3 device array was used to simulate the convolution kernel to complete the image processing operation. After each device in the array was accurately modulated to the corresponding responsivity by pulse voltage, the light signal was input and the output current of the array was detected, and the image reconstruction was completed based on the output current value, and finally the Sharpen processing of the target image was realized.

[0113] Figure 4 The responsivity reconfigurable change curve diagram of the photo-sensing and computing integrated device based on the perovskite-P(VDF-TrFE) gradient heterojunction prepared in Example 1 under the stimulation of different pulse voltages, the memory time test result diagram of the reconfigurable responsivity, and the response speed test result diagram. From Figure 4 it can be seen that under the modulation of-1.8 V, 1 ms pulse voltage, the responsivity of the photo-sensing and computing integrated device shows excellent reconfigurable characteristics, and finally realizes 1024 distinguishable states (corresponding to 10 bits accuracy), which verifies the high precision and wide range of weight regulation. From Figure 4 b in it can be seen that under the condition of room temperature without packaging and continuous light, the memory time of the photo-sensing and computing integrated device can reach 5000 s, indicating that it has stable state retention capability. From Figure 4 c and d in it can be seen that whether the response is positive or negative, the photo-sensing and computing integrated device shows ultrafast response speed, and the response time is in the order of ns, which meets the low delay demand of real-time in-sensing calculation.

[0114] Application Example 2

[0115] The photo-sensing and computing integrated device based on PbI2-BaTiO3 gradient heterojunction prepared in Example 5 was used for image embossing (Emboss) processing, and the processing method included the following steps:

[0116] S1, under the illumination of 650 nm light source, the photo-sensing and computing integrated device was modulated by Keithley-2400 pulse mode with pulse voltage amplitude of 1.8 V, pulse width of 1 ms and pulse frequency of 0.2 Hz;

[0117] S2, the current value of the photo-sensing and computing integrated device before and after different pulse modulation was collected by Keithley-2400, and the responsivity was calibrated according to the formula R=I / (PxA);

[0118] S3, the light weight value (i.e. responsivity) of the device is regulated by applying different number of pulse voltage, and the 3x3 device array is used to simulate the convolution kernel to complete the image processing operation. After the devices in the array are precisely modulated to the corresponding responsivity by pulse voltage, the input light signal is inputted and the output current of the array is detected, and the image reconstruction is completed based on the output current value, and finally the Emboss processing of the target image is realized.

[0119] Application Example 3

[0120] The image outline extraction (Outline) processing is performed on the light sensing and computing integrated device based on PbS-BiFeO3 gradient heterojunction prepared in Example 6, and the processing method includes the following steps:

[0121] S1, under the irradiation of a 650 nm light source, the light sensing and computing integrated device is modulated by a pulse voltage with an amplitude of 1.8 V, a pulse width of 1 ms and a pulse frequency of 0.2 Hz in the pulse mode of Keithley-2400;

[0122] S2, the current values of the light sensing and computing integrated device before and after different pulse modulation are collected by Keithley-2400, and the responsivity is calibrated according to the formula R=I / (PxA);

[0123] S3, the light weight value (i.e. responsivity) of the device is regulated by applying different number of pulse voltage, and the 3x3 device array is used to simulate the convolution kernel to complete the image processing operation. After the devices in the array are precisely modulated to the corresponding responsivity by pulse voltage, the input light signal is inputted and the output current of the array is detected, and the image reconstruction is completed based on the output current value, and finally the Outline processing of the target image is realized.

[0124] Figure 5 The structure schematic diagram of the light sensing and computing integrated device based on PbI2-BaTiO3 gradient heterojunction prepared in Example 5 and the light sensing and computing integrated device based on PbS-BiFeO3 gradient heterojunction prepared in Example 6, the XRD pattern of the gradient heterojunction layer and the reconfigurable responsivity change curve diagram. From Figure 5 It can be seen from a and b in FIG. 6 that the PbI2-BaTiO3 gradient heterojunction layer has good crystallinity, and the light sensing and computing integrated device based on PbI2-BaTiO3 gradient heterojunction can realize 1024 reconfigurable responsivity states. From Figure 5 It can be seen from c and d in FIG. 7 that the PbS-BiFeO3 gradient heterojunction layer has good crystallinity, and the light sensing and computing integrated device based on PbS-BiFeO3 gradient heterojunction can realize 1024 reconfigurable responsivity states.

[0125] Figure 6 The responsivity matrix of the device array for different image processing and the effect display diagram of Application Examples 1-3 are shown.Figure 6 The ideal operator processing result and the experimental operator processing result are displayed, and the actual performance and the precision level of the light sensing and processing integrated device in different image processing scenes are intuitively embodied.

[0126] Obviously, the above embodiments of the present application are only examples for clearly illustrating the present application, and are not intended to limit the implementation modes of the present application. Those skilled in the art should understand that on the basis of the above description, other different forms of changes or variations can also be made. Here, all the implementation modes cannot be exhausted. Any modification, equivalent replacement and improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the claims of the present application.

Claims

1. A method for fabricating an integrated photosensitive computing device based on a semiconductor-ferroelectric gradient heterojunction, characterized in that, Includes the following steps: (1) The conductive substrate is subjected to surface hydrophilic treatment, and the contact angle of the conductive substrate surface after treatment is 15°-60°; (2) A semiconductor solution is mixed with a ferroelectric material solution to prepare a photoelectric-ferroelectric material mixed solution; wherein the semiconductor is perovskite, lead sulfide or lead iodide; the ferroelectric material is polyvinylidene fluoride-trifluoroethylene, barium titanate or bismuth ferrite; and the concentration of the ferroelectric material in the ferroelectric material solution is 0.05-0.2 M. The volume ratio of the semiconductor solution to the ferroelectric material solution is (200-400):(8-15); (3) The photoelectric-ferroelectric material mixed solution is coated on the surface of a hydrophilically treated conductive substrate to form a semiconductor-ferroelectric gradient heterojunction layer; (4) A metal top electrode is deposited on the semiconductor-ferroelectric gradient heterojunction layer to obtain the photosensitive computing device based on the semiconductor-ferroelectric gradient heterojunction.

2. The preparation method according to claim 1, characterized in that, In step (2), the semiconductor solution is a perovskite precursor solution or a lead iodide solution; the preparation method of the perovskite precursor solution includes the following steps: dissolving lead iodide and formamidinium iodide in an organic solvent to obtain solution A, wherein the organic solvent is N,N-dimethylformamide and / or dimethyl sulfoxide; adding cesium iodide solution to solution A, and stirring at 70-100 °C for 15-20 min to obtain the perovskite precursor solution.

3. The preparation method according to claim 2, characterized in that, Solution A also includes lead bromide and / or methylamine bromide.

4. The preparation method according to claim 1, characterized in that, In step (2), the ferroelectric material solution is prepared by dissolving the ferroelectric material in an organic solvent, wherein the organic solvent is N,N-dimethylformamide and / or dimethyl sulfoxide.

5. The preparation method according to claim 1, characterized in that, In step (3), the photoelectric-ferroelectric material mixed solution is spin-coated onto the surface of a hydrophilically treated conductive substrate, and after annealing, a semiconductor-ferroelectric gradient heterojunction layer is obtained.

6. The preparation method according to claim 1, characterized in that, In step (4), the vapor deposition rate of the metal top electrode is 0.1-0.3 Å / s; the metal top electrode is a Cu electrode, an Au electrode, or a Cu / Au mixed electrode.

7. A photosensitive computing device based on a semiconductor-ferroelectric gradient heterojunction, prepared by the preparation method according to any one of claims 1-6.

8. The application of the photosensitive computing device based on a semiconductor-ferroelectric gradient heterojunction as described in claim 7 in neuromorphic computing.

9. The application according to claim 8, characterized in that, The image processing method based on a semiconductor-ferroelectric gradient heterojunction photosensitive computing device includes the following steps: S1. Under illumination by a light source, a pulse voltage is applied to the photosensitive computing device based on a semiconductor-ferroelectric gradient heterojunction for modulation. S2. Detect and record the current of the photosensitive computing device based on the semiconductor-ferroelectric gradient heterojunction after pulse voltage modulation, and calculate the responsivity; S3. Image processing is achieved by using the responsivity as the matrix element of the image processing kernel function.

10. The application according to claim 9, characterized in that, The image processing includes at least one of image sharpening, image embossing, and image contour extraction.

Images