An on-chip photonic number resolving detector and a resolving method

Abstract

The application discloses a kind of on-chip photon number resolution photodetector and resolution method, it is related to superconducting nanowire single-photon detector and integrated photonic device field.The application includes: on-chip optical waveguide, multi-pixel detection array and signal processing unit, wherein, multi-pixel detection array is used to obtain detector response signal, including multiple SNSPD pixels with the preset absorption rate of the goal of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of the distribution of

Description

Technical Field

[0001] This invention belongs to the field of superconducting nanowire single-photon detectors and integrated photonic devices, specifically relating to an on-chip photon number-resolution detector and resolution method. Background Technology

[0002] Photon number-resolved detectors are important measurement devices in quantum optics and integrated quantum photonics, used for specific tasks such as photon statistical measurement, non-classical optical field characterization, detection at quantum communication receivers, and measurement-induced state preparation in optical quantum computing. An ideal photon number-resolved detector not only needs to determine whether a photon has arrived, but also needs to distinguish between one, two, or multiple photon events.

[0003] Existing photon number-resolved detection technologies mainly include transition-edge sensors (TES) and quasi-photon-number-resolving detectors (quasi-PNR detectors) that utilize time multiplexing, spatial multiplexing, and other schemes.

[0004] Conversion edge sensors possess inherent energy resolution capabilities, enabling them to directly distinguish events with different photon counts based on the total energy of absorbed photons. However, conversion edge sensors typically require lower operating temperatures, have slower recovery speeds, and exhibit higher system size, low-temperature loads, and readout complexity, presenting engineering limitations in terms of tight integration with large-scale on-chip photonic circuitry and high-speed repetitive measurements.

[0005] Quasi-photon number-resolved schemes typically distribute incident photons into multiple distinguishable detection modes, which are then detected separately by multiple binary single-photon detectors. The number of incident photons is then inferred from the number and combination of trigger detectors. Spatial multiplexing schemes based on beamsplitter tree networks require distributing incident light into multiple spatial channels. As the target distinguishable photon number increases, the number of beamsplitters, propagation path length, number of channels, on-chip area, and number of cryogenic readout lines all increase, leading to an accumulation of additional insertion loss, channel imbalance errors, and system complexity. Temporal multiplexing schemes map different time modes to detection events using delay lines, but they typically require long delay structures, high-speed timing discrimination, and complex time mode matching, also facing trade-offs between loss, speed, and system size.

[0006] In summary, the conversion edge sensor has good intrinsic photon number resolution capability, but it is limited in terms of high-speed, low-temperature system complexity and large-scale on-chip integration. The quasi-PNR scheme based on multiple binary detectors is more suitable for integrated and high-speed detection, but its photon number resolution fidelity is easily limited by beam splitting loss, channel imbalance, collision probability of multiple photons falling into the same detection mode, and readout scale.

[0007] Therefore, the key technical problem that current on-chip photon number-resolution detectors need to solve is how to improve the photon number resolution fidelity of quasi-PNR detectors while maintaining the advantages of SNSPD such as high speed, low dark count, high efficiency and easy on-chip integration, and reduce its dependence on large-scale beam splitting networks and a large number of independent readout channels. Summary of the Invention

[0008] The purpose of this invention is to propose an on-chip photon number-resolution detector and a resolution method. By establishing a multi-pixel detector array arranged along the waveguide propagation direction, and setting the absorption rate of each SNSPD pixel in the multi-pixel detector array with the goal of making the photon landing point distribution tend to be uniform, multiple distinguishable absorption positions are formed, reducing the collision probability of multiple photons falling into the same detection area, and solving the problem of low photon number resolution fidelity.

[0009] In a first aspect, the present invention discloses an on-chip photon number-resolved detector, comprising: an on-chip optical waveguide for transmitting a test optical signal; a multi-pixel detector array for sequentially absorbing photons from the input test optical signal to generate a detection response signal; wherein the multi-pixel detector array includes a plurality of SNSPD pixels arranged sequentially along the waveguide propagation direction; the absorption rate of each SNSPD pixel is pre-set with the goal of achieving a uniform distribution of photon landing points; the multi-pixel detector array is further configured with a bias circuit and a readout circuit, the readout circuit being used to acquire the detection response signal and obtain at least one of the following: click count, click position, time difference, amplitude, and waveform characteristics, as a discrimination criterion, and to obtain the click distribution of the detector response signal based on a preset discrimination algorithm; and a signal processing unit for reconstructing the photon number distribution of the test optical signal based on the click distribution and the response matrix of the multi-pixel detector array.

[0010] Furthermore, the absorptivity of the plurality of SNSPD pixels is set by designing the geometric parameters of the waveguide structure and / or the geometric parameters of the nanowires, so that the absorptivity of the plurality of SNSPD pixels gradually increases along the waveguide propagation direction; wherein, the geometric parameters of the waveguide structure include waveguide width, thickness, ridge width, sidewall tilt angle and cladding thickness; the geometric parameters of the nanowires include nanowire width, thickness, number of nanowires in a single pixel, lateral position covering the waveguide, angle between the nanowire direction and the waveguide direction, duty cycle, spacing and effective absorption length.

[0011] Furthermore, the on-chip optical waveguide adopts a unidirectional input waveguide structure; the number of nanolines in each pixel of the multi-pixel detector array is set to gradually increase along the waveguide transmission direction, so as to improve the absorption rate of each pixel. satisfy:

[0012]

[0013] Where M is the total number of SNSPD pixels, The pixel numbering is incremented along the waveguide propagation direction.

[0014] Furthermore, the on-chip optical waveguide adopts a bidirectional input waveguide structure based on a beam splitter; the beam splitter adopts a 50:50 symmetrical beam splitting ratio to split the optical signal under test into two paths, which are injected from both ends of the multi-pixel detector array towards each other. The photon absorption probabilities in the two directions are superimposed on the multi-pixel detector array to make the photon landing point distribution tend to be uniform.

[0015] Furthermore, the waveguide structure of the on-chip optical waveguide is configured such that the waveguide width gradually changes along the waveguide transmission direction, and / or the cladding thickness gradually decreases along the waveguide transmission direction, so that the absorption rate of the multiple SNSPD pixels gradually increases along the waveguide transmission direction.

[0016] Furthermore, the multiple SNSPD pixels are comb-type pixels; the nanowires of the comb-type pixels include: straight segments and bent segments; the straight segments are located within the waveguide mode field coverage area and serve as absorption segments; the bent segments are placed in the waveguide mode intensity low region or the outer region of the waveguide; the absorption intensity of each comb-type pixel is adjusted by the number of absorption segments, the length of the absorption segments, the thickness of the absorption segments, the degree of overlap between the absorption segments and the waveguide, and the spacing between adjacent absorption segments.

[0017] Furthermore, the readout circuit employs one of the following readout methods: independent parallel readout, serial multiplexing readout, rising edge readout, delayed line readout, or frequency division multiplexing readout. When the multiple SNSPD pixels are divided into several photon number-resolved detection units, serial multiplexing readout, rising edge readout, or delayed line readout is used within each unit, while independent parallel readout or frequency division multiplexing readout is used between units.

[0018] Secondly, this invention discloses an on-chip photon number resolution method, implemented based on the on-chip photon number resolution detector described in the first aspect, comprising: utilizing multiple SNSPD pixels in a multi-pixel detector array to sequentially absorb photons of the optical signal to be tested according to a preset absorption rate, generating a detection response signal; obtaining at least one of the following based on the detection response signal: click count, click position, time difference, amplitude, and waveform characteristics, as a discrimination criterion, and obtaining the click distribution of the detector response signal based on a preset discrimination algorithm; and based on the click distribution, combining it with the response matrix of the multi-pixel detector array... Reconstruct the photon number distribution of the optical signal under test.

[0019] Furthermore, the response matrix of the multi-pixel detection array elements This represents the probability that n incident photons are classified as either m click events or m photon events, and it is calculated as follows: ;

[0020] ;

[0021] ;

[0022] ;

[0023] In the formula, The maximum number of incident photons. For the number of pixels, for The coefficient of the term, For the first A generation function for each pixel. For photons were the first The probability of absorption per pixel. The actual number of photons reaching this pixel. For the first The probability of a pixel generating a pulse. For the first Quantum efficiency per pixel This represents the probability of generating a hidden mark within a single click.

[0024] Furthermore, it also includes: calibrating the parameters of the on-chip photon number-resolved detector, correcting the response matrix of the multi-pixel detector array based on the actual calibrated photon drop point distribution, quantum efficiency, and dark counting probability, and correcting the discrimination algorithm of the readout circuit based on the actual calibrated discrimination threshold.

[0025] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0026] (1) The on-chip photon number-resolved detector proposed in this invention cascades multiple SNSPD pixels on the same optical waveguide propagation path, and sets the absorption rate of multiple SNSPD pixels with the goal of making the photon landing point distribution tend to be uniform, forming multiple distinguishable absorption positions, reducing the collision probability of multiple photons falling into the same detection area, and the remaining photons that are not absorbed in the previous stage can be completely transmitted to the subsequent detection pixels, realizing the synchronous photon response of multiple pixels while maintaining the high photon capture and detection probability of the whole array, and avoiding the additional loss, area overhead and channel imbalance caused by large-scale beam splitting network;

[0027] (2) The on-chip photon number resolution detector proposed in this invention adopts a bidirectional injection waveguide structure design to improve the defect of unidirectional transmission photons being concentrated and absorbed by the front-end pixel; combined with a preset gradient absorptivity distribution structure, it significantly reduces the probability of multiple photons being captured by the same pixel, significantly improves the photon number resolution fidelity of the multi-pixel detection array, and avoids the stringent process requirements of the unidirectional structure for an extremely wide absorptivity dynamic range. It is compatible with the existing limited photolithography and thin film fabrication process precision, significantly improves the array fabrication yield, and can stably achieve large-scale mass production of photon detection arrays;

[0028] (3) The on-chip photon number-resolved detector proposed in this invention uses comb-type superconducting nanowire detector pixels, placing the nanowire corners outside the optical waveguide to avoid the situation where photons are absorbed in the corner region but no response is generated, thus avoiding corner loss. At the same time, this scheme allows for a wider corner width and the design of the inner bend shape of the corner as a current-flow-oriented shape, alleviating the superconducting current congestion effect; thus simultaneously improving the uniformity of critical current and quantum efficiency of each pixel in the multi-pixel detector array.

[0029] (4) The on-chip photon number resolution method proposed in this invention establishes a response matrix based on the quantum noise mitigation method, which is used to quantitatively characterize various non-ideal defects and noise interference in the detector. Through the matching response matrix calibration algorithm, various non-ideal interference terms such as uneven absorption of array pixels, dark count of devices, non-ideal internal detection efficiency, and readout circuit error are uniformly incorporated into the calculation model, effectively offsetting the inherent bias of the system and improving the photon number inversion and discrimination fidelity under actual working conditions. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of the overall structure of the on-chip photon number-resolved detector provided in Embodiment 1 of the present invention;

[0031] Figure 2 This is a schematic diagram of the bidirectional injection structure and taper structure provided in Embodiment 1 of the present invention;

[0032] Figure 3 This is a three-dimensional structural diagram of the SNSPD pixel provided in Embodiment 1 of the present invention;

[0033] Figure 4 This is a comparison diagram of the photon landing point distribution of three different configurations of photon number-resolved detectors provided in Embodiment 1 of the present invention;

[0034] Figure 5 This is a schematic diagram of the comb structure nanowire provided in Embodiment 2 of the present invention;

[0035] Figure 6 This is a schematic diagram of the comb structure cascade provided in Embodiment 2 of the present invention;

[0036] Figure 7 This is a schematic diagram of the comb structure serial readout scheme provided in Embodiment 2 of the present invention;

[0037] Figure 8 This is the output voltage diagram of the series readout circuit provided in Embodiment 2 of the present invention;

[0038] Figure 9 This is a flowchart of photon number reconstruction provided in Embodiment 3 of the present invention;

[0039] Figure 10 This is a flowchart of obtaining the response matrix based on parameter calibration and matrix generation algorithm provided in Embodiment 3 of the present invention;

[0040] Figure 11 This is the parameter calibration test system provided in Embodiment 3 of the present invention;

[0041] In the figure: 1. First photon incident end; 2. Second photon incident end; 3. Optical waveguide; 4. SNSPD pixel; 5. Readout circuit; 6. Signal processing unit; 7. Gradient width waveguide; 8. Photon incident port; 9. Y-type beam splitter; 11. Substrate; 12. Cladding; 13. Superconducting nanowire; 21. Straight section of nanowire; 22. Bending section of nanowire; 23. First comb structure superconducting nanowire; 24. Second comb structure superconducting nanowire; 25. Third comb structure superconducting nanowire; 27. Series comb structure superconducting nanowire; 28. Parallel resistor; 31. Counter; 32. Pulsed laser source; 33. Attenuator; 34. Polarization controller; 35. Grating coupler; 36. Amplifier. Detailed Implementation

[0042] To better understand the present invention, the following description, in conjunction with the accompanying drawings of the embodiments of the present invention, will further illustrate the present invention, but this is not intended to limit the present invention; various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the design concept of the present invention should fall within the protection scope of the present invention.

[0043] Example 1

[0044] refer to Figure 1 This embodiment discloses an on-chip photon number-resolved detector, including: an on-chip optical waveguide 3, a multi-pixel detection array and a signal processing unit 6; wherein, the multi-pixel detection array is configured with a bias circuit and a readout circuit 5.

[0045] The on-chip optical waveguide 3 is used to transmit the optical signal to be tested; the multi-pixel detector array is used to sequentially absorb photons from the input optical signal to be tested, generating a detection response signal. The detection response signal is acquired and processed by the readout circuit 5 to obtain at least one of the following: click count, click position, time difference, amplitude, and waveform characteristics, which are used as the discrimination criteria, and the click distribution of the detector response signal is obtained based on the preset discrimination algorithm.

[0046] The signal processing unit 6 is used to reconstruct the photon number distribution of the optical signal under test based on the click distribution and the response matrix of the multi-pixel detector array.

[0047] For n incident photons, the photons may be absorbed by n different pixels, or the n incident photons may fall into the same pixel, making the photons indistinguishable.

[0048] To avoid multiple photons falling into the same pixel, in this embodiment, the multi-pixel detector array is configured with M SNSPD pixels arranged sequentially along the waveguide propagation direction, where M ≥ 2. The multiple SNSPD pixels form multiple distinguishable absorption positions along the light propagation path or equivalent propagation path.

[0049] It should be noted that in this embodiment, the absorption rate of each SNSPD pixel 4 is optimized with the goal of making the photon landing point distribution more uniform, so that multiple photons are absorbed by different pixels, thereby increasing the probability of multiple photons being distinguished.

[0050] Next, we will explain the principle that setting the absorption rate of SNSPD pixel 4 with the goal of making the photon drop point distribution more uniform can increase the probability that multiple photons are absorbed by different SNSPD pixels 4.

[0051] Generally, photon number resolution fidelity is calculated by determining the discrimination factor. (Characterizing the probability that each photon will eventually fall into a different pixel) and multi-photon joint detection efficiency The product is obtained, where the distinguishing factor refers to... One photon was allocated to The probability of a distinct pixel is calculated using the following formula:

[0052]

[0053] For multi-photon joint detection efficiency, i.e. The probability that all photons are successfully detected is calculated using the following formula:

[0054]

[0055]

[0056] In the formula, For photons were the first The probability of absorption by 4 SNSPD pixels. For actual absorption of the first The SNSPD pixel 4 is the number of each photon, where M is the number of SNSPD pixels 4. For the first Quantum efficiency of 4 SNSPD pixels.

[0057] According to Maclaurin's inequality:

[0058]

[0059] If and only if When the above expression takes the equal sign, at the same time Follow Since M is monotonically increasing, the first condition for obtaining maximum fidelity is that the photon drop point distribution is constant. Under the condition of satisfying the first condition, according to the above formula for calculating the multiphoton joint detection efficiency, it can be concluded that as the number of SNSPD pixels M and the quantum efficiency increase, the efficiency of the multiphoton joint detection will also increase. As the number of pixels increases, the fidelity monotonically increases. Therefore, the number of SNSPD pixels should be maximized within engineering limits to improve quantum efficiency. It also needs to be as close to 100% as possible.

[0060] Furthermore, if multiple identical SNSPD pixels 4 are unidirectionally cascaded on a single straight waveguide, satisfying... For a given value, the photon landing point distribution is as follows:

[0061] ;

[0062] In the formula, For photons were the first The probability of absorption by a single SNSPD pixel 4 can be seen from the above formula. It's not a fixed value, but rather... The increase of is monotonically decreasing, which does not satisfy the maximum fidelity condition mentioned above. From a physical perspective, this means that the probability of a photon being absorbed by the front SNSPD pixel 4 is much greater than that of the rear SNSPD pixel 4, resulting in most photons being absorbed by the front SNSPD pixel 4, making it impossible to distinguish multiple photons.

[0063] Generally, the absorption rate of SNSPD pixel 4 can be adjusted in the following ways:

[0064] First, the nanowire geometry parameters are adjusted by setting the nanowire width, thickness, number of nanowires in a single pixel, lateral position of the covering waveguide, angle between the nanowire direction and the waveguide direction, duty cycle, spacing, and effective absorption length.

[0065] Second, the waveguide structure geometry parameters are adjusted by setting the waveguide width, thickness, ridge width, sidewall tilt angle, cladding thickness, and evanescent field intensity.

[0066] Third, adjustments can be made by setting material parameters and dielectric spacers, for example, by placing SiO2, Al2O3, SiNx or other low-loss dielectric layers between the nanowire and the waveguide.

[0067] In an embodiment employing a unidirectional input waveguide structure, for M SNSPD pixels 4, if it is desired that the probability of a photon landing on each SNSPD pixel 4 is equal, unidirectional incidence must satisfy:

[0068]

[0069] Solving the above equation yields the general term formula:

[0070]

[0071] in, The pixel numbering is incremented along the waveguide propagation direction. For the first Absorption rate of each SNSPD pixel.

[0072] In one embodiment, the number of nanolines in each pixel of the multi-pixel detector array is set to gradually increase along the waveguide transmission direction so that the absorption rate of the SNSPD pixel 4 of the multi-pixel detector array conforms to the above general formula.

[0073] In this embodiment, to alleviate the problem of concentrated absorption at the front end during unidirectional propagation and to ensure a uniform distribution of photon landing points, the following is provided: Figure 2 The bidirectional injection structure is shown.

[0074] refer to Figure 1 and Figure 2 The bidirectional injection structure is achieved through a beam splitting device with a symmetrical beam splitting ratio. The optical signal to be measured is split into two paths by an on-chip or off-chip beam splitting device. These two signals are injected into each other from the first photon incident end 1 and the second photon incident end 2 located at both ends of the multi-pixel detector array. That is, the light field enters the multi-pixel detector array from two opposite directions, so that the photon absorption probabilities in the two directions are superimposed on the multi-pixel detector array, which is conducive to the photon landing point distribution tending to be uniform.

[0075] The beam splitter can be a Y-beam splitter, a multimode interference beam splitter, a directional coupler, a star coupler, an adjustable beam splitter, an off-chip fiber beam splitter, or other structures that can achieve beam splitting.

[0076] The preferred power ratio of the two inputs is 50:50, but it can also be a non-uniform beam split; the two signal inputs can be entered into the multi-pixel detection array simultaneously, or a preset time delay can be used.

[0077] The photon drop point distribution of a photon number-resolved detector with a bidirectional input structure satisfies:

[0078]

[0079] in, It is the first Absorption rate per pixel The transmittance of the beam splitter. Let be the reflectivity of the beam splitter, and let be the loss of the beam splitter. .

[0080] With the bidirectional injection structure, the absorption probabilities of photons from two directions are superimposed on the same pixel array. Compared with unidirectional propagation, this structure can significantly reduce the tendency of photons to concentrate only on the first few SNSPD pixels at one end, making the photon landing point distribution closer to a uniform distribution, thereby improving the distinguishability of multiphoton events.

[0081] Furthermore, although the photon droplet distribution is close to uniform after introducing the bidirectional injection structure, it is still not uniform. Therefore, in the device design process of this embodiment, starting from the target result, the absorption rate of each SNSPD pixel 4 is designed in reverse by adjusting the geometric structure and material parameters, so that the actual absorption rate is achieved. The optimal value has been achieved.

[0082] It should be noted that, Figure 3 A three-dimensional structural diagram of SNSPD pixel 4 is shown, including on-chip optical waveguide 3, superconducting nanowire 13, silicon dioxide substrate 11 and superconducting nanowire cladding 12, where w is the waveguide width and h is the cladding 12 thickness.

[0083] The theoretical basis is as follows:

[0084] When using bidirectional incidence, the photon drop point distribution must satisfy:

[0085]

[0086] Although the general formula cannot be obtained in the case of bidirectional incidence, it can be solved using Newton's iteration method in practical engineering applications. Compared to unidirectional incidence, bidirectional incidence provides... The dynamic range requirement is lower, making it easier to implement in engineering.

[0087] Further reference Figure 2 In this embodiment, inspired by the taper structure, the waveguide width and cladding thickness are set to be gradually varied to adjust the absorption rate of each SNSPD pixel 4, so that the absorption rate of multiple SNSPD pixels 4 gradually increases along the waveguide transmission direction.

[0088] Figure 2 The diagram illustrates the bidirectional injection structure and the taper structure. After the light field enters the photon incident port 8, it is split into two beams by the Y-type beam splitter 9, which then enter from both ends of the multi-pixel detector array. With the taper structure as a reference, the waveguide in the detector coverage area is made into a gradient width waveguide 7. Figure 2 Taking a silicon waveguide as an example, the two sides of the multi-pixel detector array Y-type beamsplitter 9 are the absorption front end, and the middle position is the absorption back end, such as... Figure 2 As shown, the waveguide width of the absorption front end is wider and the cladding 12 thickness is thicker, which is used to weaken the front-end absorption, while the waveguide width of the absorption back end is narrower and the cladding 12 thickness is thinner, which is used to enhance the back-end absorption.

[0089] In specific embodiments, for rectangular waveguides such as those made of silicon or silicon nitride, the waveguide width is set to gradually narrow along the waveguide propagation direction, while for ridge waveguides such as those made of lithium niobate, the waveguide width is set to gradually widen along the waveguide propagation direction. For any waveguide, the cladding 12 thickness is set to gradually thin along the waveguide propagation direction.

[0090] This embodiment also provides a set of simulation experimental data for comparative experiments. Figure 4 The photon drop point distributions of three different configurations of photon number-resolved detectors are shown. In the figure, c serves as the reference group, representing a straight waveguide integrated photon number-resolved detector with identical pixels, without any optimization. Its photon drop point distribution... The distribution of photon landing points is clearly higher at the front and lower at the back. 'a' represents a photon number-resolved detector employing a bidirectional injection structure. Symmetrical at both ends and more uniform; b is a photon number-resolved detector using a bidirectional injection structure and a taper structure, showing its photon drop point distribution. It tends to be evenly distributed.

[0091] As can be seen from the comparative experimental data, the bidirectional injection structure and the taper structure have significant effects on optimizing the photon landing point distribution.

[0092] The multi-pixel detection array in this embodiment can employ various readout methods.

[0093] The first optional readout method is independent parallel readout, where each SNSPD pixel 4 has an independent bias and an independent readout port. This method has a clear structure and is suitable for situations with a small number of pixels or where complete position resolution is required.

[0094] The second optional readout method is series multiplexing readout. Multiple SNSPD pixels are electrically connected in series or share a single readout link, with each pixel connected in parallel with a resistor, bypass resistor, inductor, or impedance control element. When a pixel absorbs a photon and transitions to a resistive state, its bias current is redistributed through the parallel resistor or bypass structure, generating a identifiable electrical pulse while preventing the entire series array from failing completely. Different pixels or different numbers of pixels triggering the pulse can produce different pulse amplitudes, pulse widths, rise times, recovery times, or combined waveforms.

[0095] The third optional readout method is delayed line readout. Multiple SNSPD cells are connected to different positions on the transmission line, and the trigger signal propagates along the transmission line to both ends. The trigger position is determined by the arrival time difference of the output signals at both ends, and the number of incident photons is inferred from the trigger position, the number of triggers, and the pulse combination.

[0096] The fourth optional readout method is frequency division multiplexing readout. By assigning a unique independent carrier frequency to each SNSPD pixel 4 in the array, the detection electrical signals of different pixels are modulated on their respective dedicated frequency points, and all channel signals are mixed and transmitted through a single common transmission line to achieve parallel readout of multiple signals.

[0097] The fifth optional readout method is rising edge readout. When the detector generates multiple hotspots at different locations, the resistance and dynamic inductance of the hotspots will differ, thus affecting the steepness of the rising edge of the response pulse. Therefore, the number of clicks can be read out by the slope of the rising edge. However, this method can usually distinguish no more than 4 clicks, so it needs to be used in conjunction with other readout methods.

[0098] When there are a large number of unidirectional cascaded SNSPD pixels 4, all SNSPD pixels 4 can be divided into several groups. Each group of SNSPD pixels 4 can constitute a photon number-resolved detection unit. Multiple photon number-resolved detection units are cascaded along the same waveguide. Within a unit, serial multiplexing readout, rising edge readout, or delayed line readout is used. Between units, independent parallel readout or frequency division multiplexing readout is used. Light not absorbed by the preceding detection unit continues to propagate along the waveguide and enters the subsequent detection unit. The response results of each SNSPD pixel 4 or each detection unit are used together to determine the incident photon count, thereby maintaining sufficient photon number discrimination capability while reducing the number of cables and low-temperature thermal load.

[0099] Regarding the preparation and integration methods, this embodiment can be achieved by direct preparation or transfer integration.

[0100] The direct fabrication method involves depositing a superconducting thin film on a pre-fabricated photonic chip and patterning it to form multiple SNSPD pixels. The transfer integration method involves first fabricating an SNSPD thin film device or sheet on a donor substrate, and then transferring and aligning the multi-pixel detector onto the target optical waveguide chip using PDMS stamp pick-and-release or other micro-transfer techniques.

[0101] Transfer integration is particularly suitable for optimizing high-quality superconducting thin films and low-loss photonic chips separately before heterogeneous integration, avoiding process compatibility issues that may be introduced by directly depositing and etching superconducting thin films on the target photonic chip.

[0102] In specific embodiments, the target photonic chip may include silicon, silicon nitride, lithium niobate, silicon dioxide, III-V semiconductor, polymer, or a hybrid integrated platform thereof. Superconducting materials may include NbN, NbTiN, WSi, MoSi, MoGe, TiN, or other superconducting thin films suitable for single-photon detection.

[0103] This embodiment eliminates the need for a large-scale beam splitting network by setting multiple distinguishable absorption sites on a single waveguide, thereby reducing chip area, reducing additional optical path loss, and improving channel imbalance. Bidirectional injection combined with gradient absorptivity distribution disperses photon absorption positions, reduces the probability of multiple photons being captured by the same SNSPD pixel 4, significantly improves quasi-photon number resolution, and greatly enhances on-chip detection efficiency.

[0104] Example 2

[0105] To eliminate corner losses and improve quantum efficiency To get as close to 100% as possible, this embodiment provides an on-chip photon number-resolution detector using comb-type pixels, based on embodiment 1.

[0106] like Figure 5 As shown, in the comb structure, the straight nanowire segment 21 serves as the main absorption segment, located within the effective waveguide mode region, while the bent nanowire segment 22 is located in the region with lower waveguide mode intensity or the outer region of the waveguide. Therefore, light absorption occurs in the straight nanowire segment 21, while the bent nanowire segment 22 does not act as an absorption segment.

[0107] The advantages of comb-type pixels are as follows: placing the bent nanowire segment 22 outside the optical waveguide 3 avoids the situation where the corner region 22 absorbs photons but does not generate a response, thus avoiding corner loss. Simultaneously, this design allows for wider corners and the design of the inner curved side of the corner as a current-flow-oriented shape, alleviating the superconducting current congestion effect; this simultaneously improves the uniformity of critical current and quantum efficiency of each pixel in the multi-pixel detector array. Furthermore, the straight nanowire segment 21 is placed above the optical waveguide 3 for photon detection. The absorption intensity of each comb-type pixel can be adjusted by setting the number of absorption segments, the length of the absorption segments, the overlap between the absorption segments and the waveguide, the spacing between adjacent absorption segments, and the position of the bent segment, thereby balancing low bending loss and designability of the absorption rate.

[0108] In practical applications, comb-structured superconducting nanowires can be arranged according to... Figure 6 The cascaded structures shown are cascaded, with each comb structure superconducting nanowire considered as a pixel, i.e. Figure 6 The cascaded first comb structure superconducting nanowire 23, second comb structure superconducting nanowire 24 and third comb structure superconducting nanowire 25 are each regarded as a pixel, and can be read out by independent bias, amplitude multiplexing, frequency division multiplexing, delay line and other methods.

[0109] Besides cascading, each comb structure superconducting nanowire can also be fabricated as... Figure 7 The superconducting nanowire 27 is a series-connected comb structure, allowing a single comb structure nanowire to be divided into multiple pixels. Each winding nanowire is connected in parallel with a parallel resistor 28. When a pixel fails to quench, current flows through the parallel resistor 28, while other pixels continue to function normally. The output port 31 outputs a value based on the number of SNSPD pixels 4. Figure 8 The voltages shown are of different amplitudes.

[0110] Example 3

[0111] Similar to Embodiments 1 and 2 described above, this embodiment provides an on-chip photon number resolution method. This method is implemented based on the on-chip photon number resolution detector described in Embodiment 1 or 2. (Refer to...) Figure 8 The method includes the following steps:

[0112] Step 1: Couple the light field to be measured into the detection area.

[0113] Specifically, if the optical field to be measured is already in the optical waveguide 3, then the optical waveguide 3 can be extended directly, and a photon number-resolved detector can be integrated on top of the waveguide. If the optical field to be measured is outside the chip, then it needs to be coupled to the optical waveguide 3 first through a coupler.

[0114] In specific implementations, grating couplers or edge couplers can be used.

[0115] Step 2: The light to be tested propagates along the optical waveguide 3 and passes through multiple SNSPD pixels 4 in sequence. The multiple SNSPD pixels 4 absorb some photons according to the preset absorption rate and generate electrical pulses as detection response signals.

[0116] Step 3: The readout circuit 5 collects the electrical pulses generated by each pixel to obtain the number of clicks, click position, time difference, amplitude or waveform characteristics, which are used as the basis for judgment. Then, based on the preset judgment algorithm, the click distribution of the detector response signal is obtained.

[0117] In the implementation of the series-type amplitude multiplexing readout scheme, the function of the readout circuit 5 is to... Figure 7 The click distribution is extracted from the output amplitude information, and the criterion is the amplitude threshold. For example, the number of clicks corresponding to an output of 0.3mV to 0.5mV is set to 1, the number of clicks corresponding to an output of 0.6mV to 0.8mV is set to 2, and so on. The statistical results of multiple measurements are the click distribution. .

[0118] Step 4: Based on the click distribution, combine the response matrix of the multi-pixel detection array Reconstruct the photon number distribution of the optical signal under test.

[0119] Figure 9 The flowchart of photon number reconstruction is shown. First, the on-chip photon number resolution detector detects the light field to be measured and outputs an electrical pulse. Then, the pulse distribution is obtained through the readout circuit 5. Finally, the signal processing unit 6 reconstructs the photon number distribution based on the reconstruction matrix. The reconstruction matrix is ​​obtained by converting the response matrix.

[0120] It should be noted that the response matrix of the multi-pixel detection array elements express One incident photon was determined to be A click event or The probability of each photon event. The response matrix is ​​obtained in advance based on the actual measured parameters, eliminating the influence of fabrication process errors.

[0121] Therefore, the efficiency and accuracy of parameter calibration of the response matrix directly affect the real-time working capability of the signal processing unit 6 and the detection fidelity of the system.

[0122] The following provides a specific example for solving the response matrix, which actually solves for arbitrary matrix elements. .

[0123] In this embodiment, The calculation method is as follows;

[0124] ;

[0125] in, The process of obtaining it is as follows:

[0126] For each pixel, it can be viewed as a "binary response" model, where no matter how many photons arrive, it will generate at most one electrical pulse, with the probability being:

[0127] ;

[0128] in, For the first Quantum efficiency per pixel The probability of generating a hidden mark within a single click. This represents the actual number of photons that reached this pixel.

[0129] Therefore, using a two-dimensional exponential generating function, each pixel can be characterized by an "impulse generating function":

[0130] ;

[0131] Among them, the coefficient of the linear term The coefficient of the zero-order term represents the probability of producing a response, while the coefficient of the zero-order term represents the probability of not producing a response.

[0132] For the incident photon number is , No. Each pixel absorbs The probability of a photon is:

[0133] ;

[0134] Therefore, for the first For each pixel, the number of incident photons can be counted using the "incident photon generation function":

[0135] ;

[0136] in, The coefficient represents the first Each pixel received The probability of a photon.

[0137] Furthermore, the first The generation function for each pixel can be written as:

[0138] ;

[0139] Multiplying the generating functions of all M pixels together gives:

[0140] ;

[0141] in, for The coefficient of the term.

[0142] refer to Figure 9 When an on-chip photon number-resolved detector is working, it can be mathematically understood as the device performing an operation:

[0143]

[0144] in, For click distribution, This represents the actual incident photon number distribution.

[0145] Therefore, noise mitigation can be summarized by the following formula:

[0146]

[0147] in, The length is (M+1), where M is the number of pixels; The length is (N+1), where N is the maximum photon number cutoff; when M=N, Let be a square matrix, satisfying the reconstruction matrix .

[0148] Furthermore, the method provided in this embodiment also includes: calibrating the parameters of the on-chip photon number-resolved detector, generating a new response matrix based on the actual calibrated photon landing point distribution, quantum efficiency, and dark counting probability, and using it to correct the above response matrix; and correcting the discrimination algorithm of the readout circuit based on the actual calibrated discrimination threshold.

[0149] Figure 10 This demonstrates a flowchart of obtaining the response matrix based on parameter calibration and matrix generation algorithms, without directly calibrating the individual elements of the matrix. Instead, it starts from the physical model of the response matrix, first calibrating the absorptivity, dark count, and quantum efficiency of each SNSPD pixel 4. If it is a bidirectional input structure, the split ratio needs to be calibrated separately.

[0150] Based on these parameters, the response matrix is ​​directly generated using a matrix generation algorithm. This embodiment has demonstrated a method for generating a two-dimensional exponential function. In addition, Monte Carlo simulation models, Poisson statistical models, etc., can also be used.

[0151] Figure 11The test system for parameter calibration is demonstrated. A pulsed laser source 32 generates light pulses, which pass through an attenuator 33 and a polarization controller 34, and are then coupled into an optical waveguide 3 via a grating coupler 35. The pulses then enter a series of SNSPD pixels 4. The output signal of each SNSPD pixel 4 is amplified by an amplifier 36 and then fed into a counter 31 to measure the light count of the measuring device. (No. (count rate of individual pixels), and the number of secret marks. .

[0152] Count rate of M SNSPD pixels 4 Normalization yields the photon number drop point distribution:

[0153] ;

[0154] The probability of a blind count for each click can be calculated using the formula. Seeking, is the emission frequency of the pulsed laser 32.

[0155] Next, using the formula The light response curve of each SNSPD pixel 4 is fitted to obtain the quantum efficiency. .

[0156] After having , , After obtaining the three key parameters, the new response matrix can be obtained using the two-dimensional exponential generating function mentioned earlier.

[0157] This embodiment uses the above-mentioned matching response matrix calibration algorithm to model and incorporate various non-ideal interference terms such as uneven absorption of array pixels, dark count of devices, non-ideal internal detection efficiency, and readout circuit error into the calculation model. This effectively offsets the inherent bias of the system and improves the accuracy of photon number inversion and discrimination under actual working conditions.

[0158] The above embodiments are only for illustrating the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly. They should not be used to limit the scope of protection of the present invention. All equivalent transformations or modifications made in accordance with the spirit and essence of the present invention should be covered by the present invention.

Claims

1. An on-chip photon number-resolved detector, characterized in that, include: On-chip optical waveguide, used to transmit the optical signal to be tested; A multi-pixel detector array is used to sequentially absorb photons from an input optical signal to be tested, generating a detection response signal; wherein, the multi-pixel detector array includes multiple SNSPD pixels arranged sequentially along the waveguide propagation direction; the absorption rate of each SNSPD pixel is pre-set with the goal of making the photon landing point distribution tend to be uniform. The multi-pixel detection array is also configured with a bias circuit and a readout circuit. The readout circuit is used to collect the detection response signal and obtain at least one of the following: number of clicks, click position, time difference, amplitude, and waveform characteristics, as a discrimination criterion, and obtain the click distribution of the detector response signal based on a preset discrimination algorithm. The signal processing unit is used to reconstruct the photon number distribution of the optical signal under test based on the click distribution and the response matrix of the multi-pixel detection array.

2. The on-chip photon number-resolved detector according to claim 1, characterized in that, The absorptivity of the multiple SNSPD pixels is set by designing the geometric parameters of the waveguide structure and / or the geometric parameters of the nanowires, so that the absorptivity of the multiple SNSPD pixels gradually increases along the waveguide transmission direction. The geometric parameters of the waveguide structure include waveguide width, thickness, ridge width, sidewall tilt angle, and cladding thickness; the geometric parameters of the nanowires include nanowire width, thickness, number of nanowires in a single pixel, lateral position covering the waveguide, angle between the nanowire direction and the waveguide direction, duty cycle, spacing, and effective absorption length.

3. The on-chip photon number-resolved detector according to claim 2, characterized in that, The on-chip optical waveguide adopts a unidirectional input waveguide structure; The number of nanolines in each pixel of the multi-pixel detection array is set to gradually increase along the waveguide transmission direction. To maximize the absorption rate of each pixel satisfy: ； Where M is the total number of SNSPD pixels, The pixel numbering is incremented along the waveguide propagation direction.

4. The on-chip photon number-resolved detector according to claim 2, characterized in that, The on-chip optical waveguide adopts a bidirectional input waveguide structure based on a beam splitter. The beam splitter uses a 50:50 symmetrical beam splitting ratio to split the optical signal under test into two paths. These two signals are injected from both ends of the multi-pixel detector array towards each other. The photon absorption probabilities in the two directions are superimposed on the multi-pixel detector array to make the photon landing point distribution tend to be uniform.

5. The on-chip photon number-resolved detector according to claim 2, characterized in that, The waveguide structure of the on-chip optical waveguide is configured such that the waveguide width gradually changes along the waveguide transmission direction, and / or the cladding thickness gradually decreases along the waveguide transmission direction.

6. The on-chip photon number-resolved detector according to claim 1, characterized in that, The plurality of SNSPD pixels are comb type pixels; the nanowires of the comb type pixels include: straight segments and bent segments; The straight segment is located within the waveguide mode field coverage area and serves as an absorption segment; the bent segment is placed in a region with low waveguide mode intensity or on the outer side of the waveguide. The absorption intensity of each comb pixel is adjusted by the number of absorption segments, the length of the absorption segments, the thickness of the absorption segments, the degree of overlap between the absorption segments and the waveguide, and the spacing between adjacent absorption segments.

7. The on-chip photon number-resolved detector according to claim 1, characterized in that, The readout circuit adopts one of the following readout methods: independent parallel readout, serial multiplexing readout, rising edge readout, delay line readout, and frequency division multiplexing readout. When the multiple SNSPD pixels are divided into several photon number-resolved detection units, serial multiplexing readout, delayed line readout or rising edge readout is used within the unit, and independent parallel readout or frequency division multiplexing readout is used between units.

8. An on-chip photon number resolution method, implemented based on the on-chip photon number resolution detector according to any one of claims 1-7, characterized in that, include: Multiple SNSPD pixels in a multi-pixel detector array absorb photons of the optical signal to be measured in sequence according to a preset absorption rate to generate a detection response signal; Based on the detection response signal, at least one of the following is obtained: click count, click location, time difference, amplitude, and waveform characteristics, which are used as the discrimination criteria, and the click distribution of the detector response signal is obtained based on a preset discrimination algorithm; Based on the click distribution, combined with the response matrix of the multi-pixel detection array Reconstruct the photon number distribution of the optical signal under test.

9. The on-chip photon number resolution method according to claim 8, characterized in that, The response matrix of the multi-pixel detection array elements This represents the probability that n incident photons are classified as either m click events or m photon events, and it is calculated as follows: ; ； ； ； In the formula, The maximum number of incident photons. For the number of pixels, for The coefficient of the term, For the first A generation function for each pixel. For photons were the first The probability of absorption per pixel. The actual number of photons reaching this pixel. For the first The probability of a pixel generating a pulse. For the first Quantum efficiency per pixel This represents the probability of generating a hidden mark within a single click.

10. The on-chip photon number resolution method according to claim 8, characterized in that, Also includes: The parameters of the on-chip photon number-resolved detector are calibrated, and the response matrix of the multi-pixel detection array is corrected according to the actual calibrated photon drop point distribution, quantum efficiency, and dark counting probability. The discrimination algorithm of the readout circuit is corrected according to the actual calibrated discrimination threshold.

Images