In-pixel reconfigurable band-alignment hyperspectral quantum dot image sensor
By integrating a bias-programmable colloidal quantum dot junction detector into a hyperspectral quantum dot image sensor and adjusting the band alignment state using a unipolar bias, the multi-dimensional performance limitations of hyperspectral imaging in existing technologies are overcome, enabling high-resolution, wide-spectral-response, and low-cost integrated applications.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING INST OF TECH
- Filing Date
- 2026-04-16
- Publication Date
- 2026-07-14
AI Technical Summary
Existing hyperspectral imaging technologies struggle to achieve high spatial resolution, high spectral resolution, continuous and stable spectral modulation, and wide spectral response. Furthermore, they are incompatible with standard semiconductor manufacturing processes, limiting their application in portable and large-area imaging arrays.
A hyperspectral quantum dot image sensor with reconfigurable band alignment within each pixel is used. By integrating a bias-programmable colloidal quantum dot junction detector in each pixel, the band alignment state is adjusted by unipolar incremental bias to achieve a continuous spectral response function. It is monolithically integrated with complementary metal-oxide-semiconductor (CMOS) technology through silicon readout circuitry.
It achieves spectral modulation with high spatial resolution, high spectral resolution, wide spectral response and high signal-to-noise ratio, reduces manufacturing costs and process difficulty, supports large-scale integration and portable applications, and expands application scenarios such as non-destructive material identification and convenient on-site analysis.
Smart Images
Figure CN122396080A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of optoelectronic imaging and sensor technology, and in particular to an on-chip integrated computational spectrometer based on the spectral response of a voltage-modulated detector, which uses colloidal quantum dot materials and a reconfigurable band alignment structure, and is suitable for portable hyperspectral imaging and analysis. Background Technology
[0002] High-resolution infrared hyperspectral imaging technology can acquire spatial information of a target while analyzing its spectral characteristics, and has important application prospects in fields such as material identification, precision agriculture, environmental monitoring, and space exploration. However, traditional hyperspectral imaging equipment relies on bulky dispersive optical elements and mechanical scanning components, resulting in large system size, high cost, and difficulty in integration, which limits its application in portable scenarios.
[0003] To address these issues, chip-level computational spectrometer technology has emerged, replacing complex optical paths with computational reconstruction, thus achieving miniaturization of the equipment. Computational spectrometers are mainly divided into two categories: incident spectrum tuned type and detector response tuned type.
[0004] Incident spectral tuning techniques modulate incident light by integrating filters or dispersive elements on the detector, but this usually comes at the cost of spatial resolution and requires stringent micro- and nano-fabrication processes.
[0005] Detector response tuning technology achieves spectral modulation by adjusting the detector's own photoelectric response, avoiding the trade-off between spatial and spectral resolution, and thus holds greater potential. Representative existing technologies mainly include two types:
[0006] The first approach employs a gate field modulation strategy. It utilizes van der Waals heterostructures such as black phosphorus and molybdenum disulfide to construct three-terminal devices, modulating the band structure through gate voltage to modulate the spectral response. However, this technology is currently mostly limited to single-pixel devices, making it difficult to achieve large-area imaging arrays; furthermore, the material fabrication process is complex, stability and repeatability are challenging, and broadband response capabilities (especially in the short-wave infrared band) are limited.
[0007] The second approach employs a bias polarity switching strategy. It utilizes a back-to-back diode structure (such as an organic back-to-back Schottky diode) to switch between different discrete spectral bands by changing the bias polarity. However, this approach has significant drawbacks: First, at low bias voltages, the two diode junctions cancel each other out, resulting in a "dead zone" in the optical response and disrupting the continuity of spectral modulation; second, the spectral response is non-uniform, and the signal-to-noise ratio is low in certain bands; finally, the required bidirectional bias drive is incompatible with mainstream unipolar silicon readout integrated circuits, hindering its application in high-resolution imaging arrays.
[0008] Therefore, existing technologies still lack an on-chip hyperspectral imaging solution that can simultaneously achieve high spatial resolution, high spectral resolution, continuous and stable spectral modulation, wide spectral response, and compatibility with standard semiconductor manufacturing processes. Summary of the Invention
[0009] This invention addresses the shortcomings of existing technologies by providing a hyperspectral quantum dot image sensor with in-pixel reconfigurable band alignment.
[0010] To achieve the above-mentioned objectives, the technical solution adopted by the present invention is as follows:
[0011] A hyperspectral quantum dot image sensor with in-pixel reconfigurable band alignment, comprising a silicon readout circuit and a pixel array monolithically integrated thereon;
[0012] Each pixel in the pixel array includes a bias-programmable colloidal quantum dot junction detector;
[0013] The bias-programmable colloidal quantum dot junction detector includes a top electrode, a bottom electrode, and a stacked colloidal quantum dot junction structure located between the top electrode and the bottom electrode.
[0014] The stacked colloidal quantum dot junction structure is configured to reconstruct its internal band alignment state by applying a unipolar incremental bias voltage between the top electrode and the bottom electrode, thereby generating a continuously evolving spectral response function.
[0015] Furthermore, the stacked colloidal quantum dot junction structure includes a top n-type large bandgap colloidal quantum dot layer, a middle p-type doped small bandgap colloidal quantum dot layer, and a bottom n-type small bandgap colloidal quantum dot layer stacked sequentially.
[0016] The band gap width of the top n-type large band gap colloidal quantum dot layer is greater than that of the bottom n-type small band gap colloidal quantum dot layer.
[0017] Furthermore, under zero or low bias, the external electric field is mainly concentrated in the top n-type large bandgap colloidal quantum dot layer, and the detector mainly responds to shorter wavelength near-infrared photons. As the unipolar incremental bias increases, the external electric field extends to the junction region formed by the middle p-type doped small bandgap colloidal quantum dot layer and the bottom n-type small bandgap colloidal quantum dot layer, activating the response to longer wavelength near-infrared photons and achieving spectral response superposition.
[0018] Furthermore, the hyperspectral quantum dot image sensor is configured to perform an imaging method by applying a series of different unipolar bias voltages V1, V2, ..., V to the bias-programmable colloidal quantum dot junction detector. M And acquire the corresponding imaging measurement value I(V) under each bias voltage. iThe incident spectrum of each pixel is reconstructed based on the sequence of imaging measurements under all bias voltages.
[0019] Furthermore, the steps for reconstructing the incident spectrum include: solving the linear system. ,in For the imaging measurement value vector, For the pre-calibrated spectral response matrix, Let be the incident spectrum vector to be determined, and a smoothing constraint term is introduced to stabilize the solution process.
[0020] Furthermore, the pixel pitch of the pixel array is less than or equal to 15 μm.
[0021] Furthermore, the band gaps of the top n-type large band gap colloidal quantum dot layer and the bottom n-type small band gap colloidal quantum dot layer are engineered to enable the spectral response range of the detector to cover the wavelengths of the near-infrared band.
[0022] Furthermore, the top electrode is a semi-transparent gold electrode, and the bottom electrode is electrically connected to the silicon readout circuit through a metal-filled via.
[0023] Furthermore, under a specific unipolar bias voltage, the stacked colloidal quantum dot junction structure operates through an internal photoconductivity gain mechanism.
[0024] Furthermore, the silicon readout circuit is a unipolar readout circuit fabricated based on complementary metal-oxide-semiconductor (CMOS) technology.
[0025] Compared with the prior art, the advantages of the present invention are as follows:
[0026] First, this invention breaks through the inherent framework of mutual constraints among the core performance parameters in traditional hyperspectral imaging technology, and achieves performance improvement and synergistic optimization in multiple dimensions such as high spatial resolution, high spectral resolution, high reconstruction accuracy, wide spectral response and high detection sensitivity on a single device, thereby achieving significant progress in comprehensive performance at the device level.
[0027] Second, this invention achieves continuous, stable, and high signal-to-noise ratio spectral modulation. By applying a unipolar incremental bias voltage to continuously adjust the band alignment state of the stacked colloidal quantum dot (CQD) junction, a continuously evolving high signal-to-noise ratio spectral response function can be generated. This completely avoids the "response dead zone" problem existing in bias polarity switching technology, ensuring the reliability and continuity of the spectral signal throughout the modulation process, and laying the foundation for high-fidelity spectral reconstruction.
[0028] Third, the technical solution of this invention has strong compatibility and is easy to achieve large-scale integration and mass production. The colloidal quantum dot (CQD) material used in the sensor has solution processability and can be monolithically integrated with silicon readout circuits (ROICs) through processes such as spin coating; at the same time, its unipolar driving scheme is fully compatible with mainstream complementary metal-oxide-semiconductor (CMOS) process standards. This design avoids the need for complex optical components or high-precision micro-nano fabrication, significantly reducing manufacturing costs and process difficulty, and is conducive to promoting the widespread adoption of hyperspectral imaging technology for large-scale, low-cost applications.
[0029] Fourth, the sensor of this invention is powerful and has a wide range of applications. Its broadband sensitivity, extended to short-wave infrared (SWIR), enables it to penetrate obstructions such as fog and smoke, and achieve non-destructive material identification of textiles, agricultural products, and chemicals. The sensor supports dual-mode operation, performing high-sensitivity wide-band imaging under a fixed bias voltage, and hyperspectral calculation and reconstruction in a frame-based voltage-tuned mode. Furthermore, by mapping the characteristic spectra of materials to visible colors, subtle spectral differences can be transformed into intuitive pseudo-color images, greatly improving the convenience of on-site analysis.
[0030] Fifth, this invention employs a highly miniaturized pixel design, greatly satisfying the needs of portable and integrated applications. For example, the size of a single pixel in the sensor is only 15×15μm. 2 Furthermore, by employing monolithic integration technology, it completely eliminates the bulky optical components and moving mechanical structures such as dispersion gratings and filter wheels found in traditional hyperspectral systems. This allows for a significant reduction in the size, weight, and power consumption of the entire imaging system, making it possible to develop compact applications such as handheld analytical devices, smart wearable devices, and space exploration payloads. Attached Figure Description
[0031] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0032] Figure 1 This is a schematic diagram of the structure of the hyperspectral quantum dot image sensor in an embodiment of the present invention; the labels in the figure are: 1: incident light; 2: top electrode; 3: colloidal quantum dot; 4: bottom electrode; 5: silicon readout circuit. Detailed Implementation
[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0034] This embodiment provides a hyperspectral quantum dot image sensor with in-pixel reconfigurable band alignment. For example... Figure 1 As shown, the sensor mainly includes a pixel array and a silicon readout integrated circuit monolithically integrated with the pixel array.
[0035] 1. Overall structure and parameters of the sensor
[0036] The size of the pixel array can be designed according to application requirements, ranging from a 3x3 demonstration array to a 1280x1024 (approximately 1.3 million pixels) imaging array. Individual pixels employ a compact design with a pixel pitch of 15μm and a pixel size of 15×15μm. 2 Incident light 1 enters from the top of the pixel. Each pixel is electrically connected to the silicon readout circuit 5 below through a metal-filled via. This structure also ensures the flatness of the substrate surface, which is beneficial for the uniform deposition of subsequent functional layers.
[0037] 2. Pixel Core Detector Structure
[0038] At the core of each pixel is a bias-programmable colloidal quantum dot junction detector. This detector employs a vertically stacked structure, specifically including:
[0039] Top electrode 2: A semi-transparent gold (Au) thin film with a thickness of about 10-20 nm is used as the top electrode to ensure sufficient light transmittance and conductivity.
[0040] Stacked CQD junction structure: from bottom to top as follows:
[0041] n-type small bandgap CQD layer: As the bottom absorption layer, its colloidal quantum dot 3 has a narrow bandgap, which enables efficient absorption of longer wavelength near-infrared photons.
[0042] p-type doped small bandgap CQD intermediate layer: This layer is p-type doped and serves as an intermediate functional layer connecting the upper and lower n-type layers, which is the key to realizing bandgap reconstruction.
[0043] n-type large bandgap CQD layer: As the top absorption layer, its colloidal quantum dots have a wide bandgap and are responsible for absorbing short-wavelength near-infrared photons.
[0044] The bandgap width of the three-layer CQD decreases sequentially from top to bottom, forming a gradient bandgap structure, thereby achieving continuous absorption of photons in the infrared spectral range.
[0045] Bottom electrode 4: A metal with good conductivity (such as aluminum Al) is used as the bottom electrode and is directly connected to the silicon readout circuit.
[0046] 3. Preparation process
[0047] The manufacturing process of the sensor mainly includes the following steps:
[0048] a. CQD material synthesis: Colloidal quantum dot materials with different sizes (corresponding to different band gaps) were synthesized, and their electrical properties (n-type or p-type doping) were controlled by ligand engineering.
[0049] b. Readout circuit preprocessing: Commercial or custom silicon readout circuit wafers with a pixel pitch of 15μm are selected and subjected to standard cleaning and surface planarization.
[0050] c. CQD Functional Layer Deposition: On the pre-treated readout circuit, CQD layers are sequentially deposited using a solution spin-coating process. First, an n-type small bandgap CQD layer is spin-coated, followed by a p-type doped small bandgap CQD intermediate layer, and finally an n-type large bandgap CQD layer is spin-coated. Each layer undergoes ligand exchange or annealing after deposition to optimize film quality and electrical properties.
[0051] d. Electrode fabrication: A semi-transparent gold (Au) film is deposited on the spin-coated CQD film using a thermal evaporation process, and patterned using a shadow mask to form a pixelated top electrode. The bottom electrode was fabricated during the readout circuit fabrication process.
[0052] 4. Working mechanism and spectral modulation
[0053] The sensor operates by applying a unipolar incremental bias voltage between the top and bottom electrodes, and its band alignment and spectral response evolution process are as follows:
[0054] Zero-bias / low-bias stage: The external voltage mainly drops onto the top n-type large-bandgap CQD layer with the highest impedance, forming a wide depletion region in this layer. At this time, electron-hole pairs generated by the absorption of shorter wavelength near-infrared photons (such as visible light) in this region can be efficiently separated and collected, and the device exhibits a response to short-wavelength light. However, in the bottom n-type small-bandgap CQD layer, due to the lack of an effective electric field, carriers excited by longer wavelength near-infrared photons recombine before collection, thus resulting in no response to long-wavelength light.
[0055] Bias increase stage: As the forward bias (top electrode positive, bottom electrode negative) gradually increases, the depletion region of the top large bandgap layer widens, and the short-wavelength responsivity is enhanced. When the voltage continues to rise to a certain value, the depletion region of the top junction tends to saturate, and the additional potential begins to be applied mainly to the junction between the p-type intermediate layer and the bottom n-type small bandgap layer, thereby establishing a strong electric field in this region. At this time, the carriers generated by the absorption of longer wavelength near-infrared photons (such as infrared light) in the bottom small bandgap layer can also be effectively separated and transported. The device begins to exhibit a response to long wavelengths, which, combined with the continuous short-wavelength response, forms a continuously evolving spectral response function.
[0056] Optimized bias stage: At a specific bias point, the internal photoconductivity gain may be generated inside the device due to the carrier lifetime being longer than the transit time, so that the external quantum efficiency (EQE) exceeds 100%. At this time, the detector rate of the device reaches its peak and maintains extremely high sensitivity throughout the entire operating band.
[0057] 5. Hyperspectral Imaging and Spectral Reconstruction Methods
[0058] Based on the above working principle, the steps for implementing hyperspectral imaging are as follows:
[0059] a. Data Acquisition: The sensor is aligned with the target scene. The control circuit applies a series (e.g., M) of linearly or non-linearly increasing unipolar bias voltages V1, V2, ..., V to the sensor pixel array. M At each bias voltage Below, the silicon readout circuit reads the imaging measurements of the entire array. This measurement is consistent with the incident spectrum. and the detector's spectral response function under the current bias voltage The following relationship must be satisfied:
[0060]
[0061] Integration interval arrive For example, 400-1700nm.
[0062] b. Spectral Reconstruction: For each spatial pixel, the acquired M measurements {I(V1), I(V2), ..., I(V... M The vector I is formed by solving the linear system. (in The incident spectrum is reconstructed using a pre-calibrated M×N spectral response matrix and the N-dimensional discretized spectral vector to be determined. This is an ill-conditioned inverse problem, and direct solutions are sensitive to noise. Therefore, regularization methods such as the Tikhonov regularized least squares method are used for stable solutions. By introducing spectral smoothness constraints, noise and non-physical oscillations in the reconstruction results are suppressed, thus obtaining high-fidelity and physically reasonable spectral curves.
[0063] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Those skilled in the art can make various modifications and variations to the materials, structural parameters, bias sequences, and reconstruction algorithms without departing from the principles of the present invention; these equivalent forms also fall within the protection scope defined by the claims of the present invention.
Claims
1. A hyperspectral quantum dot image sensor with in-pixel reconfigurable band alignment, characterized in that, This includes silicon readout circuitry and a pixel array monolithically integrated on it; Each pixel in the pixel array includes a bias-programmable colloidal quantum dot junction detector; The bias-programmable colloidal quantum dot junction detector includes a top electrode, a bottom electrode, and a stacked colloidal quantum dot junction structure located between the top electrode and the bottom electrode. The stacked colloidal quantum dot junction structure is configured to reconstruct its internal band alignment state by applying a unipolar incremental bias voltage between the top electrode and the bottom electrode, thereby generating a continuously evolving spectral response function.
2. The hyperspectral quantum dot image sensor according to claim 1, characterized in that, The stacked colloidal quantum dot junction structure includes a top n-type large bandgap colloidal quantum dot layer, a middle p-type doped small bandgap colloidal quantum dot layer, and a bottom n-type small bandgap colloidal quantum dot layer stacked sequentially. The band gap width of the top n-type large band gap colloidal quantum dot layer is greater than that of the bottom n-type small band gap colloidal quantum dot layer.
3. The hyperspectral quantum dot image sensor according to claim 2, characterized in that, Under zero or low bias, the external electric field is mainly concentrated in the top n-type large bandgap colloidal quantum dot layer, and the detector mainly responds to shorter wavelength near-infrared photons. As the unipolar increasing bias increases, the external electric field extends to the junction region formed by the middle p-type doped small bandgap colloidal quantum dot layer and the bottom n-type small bandgap colloidal quantum dot layer, activating the response to longer wavelength near-infrared photons and realizing the superposition of spectral responses.
4. The hyperspectral quantum dot image sensor according to claim 1, characterized in that, The hyperspectral quantum dot image sensor is configured to perform the following imaging method: applying a series of different unipolar bias voltages V1, V2, ..., V to the bias-programmable colloidal quantum dot junction detector. M And acquire the corresponding imaging measurement value I(V) under each bias voltage. i The incident spectrum of each pixel is reconstructed based on the sequence of imaging measurements under all bias voltages.
5. The hyperspectral quantum dot image sensor according to claim 4, characterized in that, The steps to reconstruct the incident spectrum include: solving the linear system. ,in For the imaging measurement value vector, For the pre-calibrated spectral response matrix, Let be the incident spectrum vector to be determined, and a smoothing constraint term is introduced to stabilize the solution process.
6. The hyperspectral quantum dot image sensor according to claim 1, characterized in that, The pixel pitch of the pixel array is less than or equal to 15 μm.
7. The hyperspectral quantum dot image sensor according to claim 2, characterized in that, The band gaps of the top n-type large band gap colloidal quantum dot layer and the bottom n-type small band gap colloidal quantum dot layer are engineered to ensure that the spectral response range of the detector covers the wavelengths of the infrared band.
8. The hyperspectral quantum dot image sensor according to claim 1, characterized in that, The top electrode is a semi-transparent gold electrode, and the bottom electrode is electrically connected to the silicon readout circuit through a metal-filled via.
9. The hyperspectral quantum dot image sensor according to claim 2, characterized in that, Under a specific unipolar bias voltage, the stacked colloidal quantum dot junction structure operates through an internal photoconductivity gain mechanism.
10. The hyperspectral quantum dot image sensor according to claim 1, characterized in that, The silicon readout circuit is a unipolar readout circuit fabricated using complementary metal-oxide-semiconductor (CMOS) technology.