Miniature spectrometer for high-resolution spatiotemporal random speckle

Abstract

The application discloses a kind of high-resolution space-time domain random speckle microspectrometer, belong to integrated photonics field.The microspectrometer structure includes: light splitting region, interferometer array region, tunable diffraction region.Light splitting region is used to input light rough N-way light path, interferometer array region is used to carry out different degrees of modulation to the phase of incident light.Tunable diffraction region is composed of slab waveguide and electrode array, N-way light enters the region and occurs spatial diffraction, different electrode is loaded with different power voltage, to change the temperature near electrode, and then change the optical path of different light path and decorrelation, finally light forms different response in N-way output port and outputs.After calibrating transmission matrix, unknown spectrum is collected in the light power response of output end, with the aid of algorithm, the reconstruction of unknown spectrum can be realized, by combining spatial speckle in time domain, construct highly decorrelated transmission matrix, thereby improve the resolving power of spectrometer.

Description

Technical Field

[0001] This invention belongs to the field of integrated photonic devices, and more specifically, relates to a high-resolution spatiotemporal random speckle micro-spectrometer. Background Technology

[0002] Spectroscopy is the "fingerprint" for understanding the microscopic structure of matter. Spectrometers play an indispensable role in various fields of modern society, and their applications are no longer limited to traditional scientific research scenarios, but are beginning to expand into fields such as biomedicine, environmental engineering, and aerospace. Traditional benchtop spectrometers have extremely high spectral resolution, but due to the presence of a large number of large dispersive components, their application in many portable spectral analysis scenarios is quite limited. Therefore, the miniaturization of spectrometers has attracted much attention from researchers in recent years.

[0003] Currently, the main types of miniaturized spectrometers are grating-type, narrowband-filtered, Fourier transform-type, and computational reconstruction-type. Grating-type spectrometers require sufficient optical path length to achieve high resolution, making it difficult to achieve compatibility between small size and high resolution. Narrowband-filtered spectrometers require numerous narrowband filters to divide the wavelength into several intervals, also making high integration difficult. Fourier transform-type spectrometers utilize interference modulation of incident light and collect interference patterns, but in this case, resolution is limited by the maximum optical path difference. Computational reconstruction-type spectrometers are a relatively new paradigm in miniaturized spectrometers. They reconstruct unknown spectra by constructing a mapping from spectral channel information to space and using reconstruction algorithms to solve linear equations. Random speckle spectrometers are a prominent area of ​​research in computational reconstruction-type spectrometers. They use disordered structures to obtain random speckles to construct highly incoherent transfer matrices, thereby improving spectral resolution. However, existing research has limited sampling channels, making it difficult to acquire diverse spectral responses and limiting the achievement of high resolution. Therefore, developing a miniature spectrometer capable of acquiring as many highly decorrelated spectral responses as possible and enabling rapid, high-precision spectral analysis using reconstruction algorithms has significant practical value. Summary of the Invention

[0004] In view of the above-mentioned defects or improvement needs of existing micro spectrometer technology, this invention provides a high-resolution spatiotemporal random speckle micro spectrometer, which aims to solve the problem that existing micro spectrometers are unable to achieve rapid and high-precision spectral analysis.

[0005] To achieve the above objectives, the present invention provides a high-resolution spatiotemporal random speckle micro spectrometer chip, comprising a spectroscopic region, an interferometer array region, and a tunable diffraction region.

[0006] The beam-splitting region consists of multiple cascaded Y-branches, roughly dividing the input light into N optical paths and guiding them to the interferometer array region. The interferometer array region is used to modulate the phase of the input light to varying degrees, thereby producing a certain extinction ratio in the optical path transmission spectrum. The tunable diffraction region is used to decorrelate the input light; that is, by applying voltages of different powers to the electrodes, the local refractive index changes, thus altering the optical path length of different optical paths. This makes the spectral responses at the output ends uncorrelated and tend towards randomization. The spectra obtained at different output ends are spatial random speckle patterns. The purpose of decorrelation is to construct a highly uncorrelated transmission matrix, meaning that each column vector of the transmission matrix is ​​linearly independent, thereby maximizing the sampling of information from unknown spectra at each wavelength and achieving high-resolution spectral analysis. The specific spectral analysis process is as follows: First, a known full-pass light is input at the input end, and multiple rounds of random voltages with different powers are applied to the electrodes of the Mach-Zehnder interferometer and the tunable diffraction region. Different spatial speckles are obtained at the output port through the interference of the Mach-Zehnder interferometer and the spatial diffraction of the diffraction region. The spatial speckles are combined in the time domain to obtain the transmission response matrix. When an unknown spectrum is input, the same random power voltage is applied to the Mach-Zehnder interferometer array and the electrodes for modulation, and then the optical power data at the output port is collected. Thus, the unknown spectral information can be reconstructed using a reconstruction algorithm.

[0007] The beam splitting region is obtained by cascading multiple Y-branch layers, which is used to initially split the input light into N channels and then into the subsequent interferometer array region.

[0008] The interferometer array region includes N Mach-Zender interferometers (MZIs) or N micro-ring resonators. Each unit has a thermoelectrode on one arm that is thermally modulated by a randomly applied voltage. Phase modulation causes the optical path lengths of the two arms to be unequal, resulting in optical interference and enhancing the diversity of optical paths. The MZI structure can be further optimized, such as by changing the arm length or bending one arm towards the other. Each path can have a different MZI structure to provide a more diverse transmission spectrum.

[0009] The tunable diffraction region, consisting of a planar waveguide, is a spatial optical diffraction region capable of diffracting N incident lights. This region contains N rows and M columns of electrodes. By applying different random power voltages to different electrodes, the local refractive index near the electrodes is changed, further increasing the spatial diversity and reconfigurability of the optical path. Spatial speckle is obtained at the output port. By applying different batches of voltages in multiple rounds, the temporal diversity and reconfigurability of the optical path are increased. The spatial speckle obtained after applying different batches of voltages are combined to form temporal speckle, thereby realizing the superposition of spatial speckle in the time domain, increasing the sampling channels of the spectrometer, and reducing the correlation between different channels.

[0010] The electrodes in the tunable diffraction region can be square, rectangular, circular, or other polygonal junctions. The arrangement of the electrodes can be a periodic arrangement similar to a matrix, or a random arrangement, thereby breaking the spatial symmetry of the diffraction region and obtaining a more decorrelated random speckle pattern and a high-rank transfer matrix.

[0011] The output port spacing of the tunable diffraction region can be the same, or it can be gradually or randomly arranged to collect as many different transmission responses as possible.

[0012] Furthermore, to establish the transmission matrix, first input the C-band full-pass light and record the spectrum of each output port under different tuning states, which serves as a row of the transmission matrix. The number of rows in the transmission matrix represents the number of sampling channels, and the number of columns represents the number of wavelength channels. After establishing the transmission matrix, input the unknown spectrum and record the optical power response of each output port under different tuning states to construct a system of linear equations.

[0013] Furthermore, the constructed linear equation system can be solved using the CVX algorithm to solve related convex optimization problems, and the hyperparameters included in the K-fold cross-validation algorithm are used. It can also be solved using other algorithms such as Lasso regression and ridge regression to reconstruct the unknown spectrum.

[0014] In summary, compared with the prior art, the above-described technical solutions conceived by this invention can achieve the following beneficial effects:

[0015] 1. The present invention provides a high-resolution spatiotemporal random speckle micro speckle speckle chip that generates dozens or even hundreds of uncorrelated sampling channels by combining spatial random speckles in the time domain and constructing a highly uncorrelated transmission matrix, thereby significantly improving spectral resolution, with a current resolution of at least 1.2 nm.

[0016] 2. The high-resolution spatiotemporal random speckle micro spectrometer chip provided by this invention can cover the traditional communication band of 1530-1565nm, and can play a significant role in the field of optical research.

[0017] 3. The high-resolution spatiotemporal random speckle micro spectrometer chip provided by this invention has multiple output ports that can perform parallel measurements of spectral response. Under the condition of ensuring high resolution, it can complete data acquisition within a few seconds, and the algorithm reconstruction time is within ten seconds. Therefore, it has fast response time and great practical value.

[0018] 4. The high-resolution spatiotemporal random speckle micro spectrometer chip provided by this invention has a simple structure, large process tolerance, does not rely on advanced processes, and can be processed with a precision of 180nm. Compared with the complex structure of traditional micro spectrometers, it only requires a few interferometers and a planar waveguide, and has great potential for mass production. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the structure of a high-resolution spatiotemporal random speckle micro-spectrometer provided by an example of the present invention.

[0020] Figure 2 This is a microstructure diagram of a high-resolution spatiotemporal random speckle microspectrometer provided in this invention.

[0021] Figure 3 These are schematic diagrams of the transfer matrix and autocorrelation function of a high-resolution spatiotemporal random speckle micro-spectrometer provided by an example of the present invention. (a) is a schematic diagram of the transfer matrix, and (b) is a schematic diagram of the autocorrelation function.

[0022] Figure 4 This is a single-peak reconstruction result of a high-resolution spatiotemporal random speckle microspectrometer provided by an example of the present invention.

[0023] Figure 5 This is a bimodal reconstruction result of a high-resolution spatiotemporal random speckle microspectrometer provided by an example of the present invention.

[0024] Figure 6 This is a diagram illustrating the reconstruction result of a broadband unknown light covering the C-band using an autocorrelation function diagram provided by an example of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0026] like Figure 1As shown, an example of the present invention is a high-resolution spatiotemporal random speckle micro-spectrometer, comprising a beam splitting region 1, an interferometer array region 2, and a tunable diffraction region 3. The beam splitting region routes the light path to 16 channels. The MZI array region performs phase modulation on each light path through thermal tuning to improve the extinction ratio. The 16 modulated light paths enter the tunable diffraction region. After entering the planar waveguide in this region, each light path undergoes spatial diffraction and interacts with each other. The electrode array on the planar waveguide changes the local refractive index near the electrodes by applying random voltages of different powers, thereby further modulating the light path, ultimately forming a highly uncorrelated random speckle output response at the output end. The transmission matrix of the micro-spectrometer is first calibrated by inputting a known spectrum. During calibration, the output port is connected to a commercial benchtop spectrometer to record the output spectrum of each port and use it as a row of the transmission matrix. After calibration, an unknown spectrum is input and the output light power under different tuning states is collected sequentially. The unknown spectrum can then be reconstructed by solving a system of linear equations. This system of equations can be expressed as:

[0027] O = AS

[0028] Where A is the pre-calibrated transmission matrix, O is the transmission response of the unknown spectrum, and S is the unknown spectral information to be obtained. After collecting the optical power information O, S can be obtained by solving the inverse problem of this equation. However, since the transmission matrix A is often not a square matrix, direct inversion will amplify noise. Therefore, the CVX algorithm is used to solve the convex optimization problem.

[0029] Minimize{O-AS2+α1Γ1S1+α2Γ2S2}

[0030] Γ1 and Γ2 are the first-order and second-order difference operators, respectively. α1 and α2 are optimized using K-fold cross-validation. By adding a regularization penalty term, overfitting during the reconstruction process is avoided, thus accurately reconstructing the unknown spectrum.

[0031] Figure 2 The image shows a micrograph of the fabricated chip, which includes a portion of the tunable diffraction region, where each electrode can be voltaged via an FPGA power supply.

[0032] Figure 3 (a) shows the constructed transmission matrix. The number of columns in the matrix represents the wavelength channels. Currently, it shows 3501 wavelength channels in a 35nm range across the entire C-band of communication (1530nm-1565nm), divided in steps of 0.01nm. The number of rows in the matrix represents the sampling channels under different tuning states. Currently, it shows 80 sampling channels obtained after applying five rounds of random voltage. From the transmission matrix, it can be seen that there is no obvious correlation between the channels. Figure 3Image (b) shows the image of the autocorrelation function, which measures the degree of decorrelation between sampling channels. The autocorrelation function C(Δλ) can be expressed as:

[0033]

[0034] Where I(λ,k) represents the intensity of the k-th sampling channel and the λ-th wavelength channel, and <·> represents the average value for the wavelength or sampling channel. The half-width at half-maximum (HWHM) of the autocorrelation function represents the wavelength shift required for the correlation between adjacent channels of the spectrometer to decrease to half, and is an indicator of the resolution of a random speckle spectrometer. As can be seen from the figure, the HWHM of the autocorrelation function is approximately 0.35 nm, indicating a high degree of decorrelation of the transfer matrix.

[0035] Figure 4 The reconstruction results of a single narrow-peak unknown spectrum are shown. The full width at half maximum (FWHM) of the narrow peak is 0.12 nm. As can be seen from the figure, the shape of the unknown spectrum has been perfectly reconstructed. The relative error is used as the metric to evaluate the reconstruction accuracy.

[0036]

[0037] Where S0 is the reference spectrum and S is the reconstructed spectrum, and high-precision reconstruction is generally considered to be achieved when ε < 0.1. Figure 4 In the figure, the value of ε is 0.079, indicating that the reconstruction accuracy is very high.

[0038] Figure 5 The reconstruction results of the unknown bimodal spectrum with a 1 nm interval are shown. The reconstruction ε value is 0.039. It can be seen from the figure that most of the details can be perfectly restored.

[0039] Figure 6 The reconstruction results of a broadband unknown spectrum with a bandwidth covering the entire C-band are shown. The reconstruction ε is 0.187. Broadband light is a more challenging reconstruction target. As can be seen from the figure, most of the details are restored very well.

[0040] This invention provides a high-resolution spatiotemporal random speckle micro-spectrometer. By combining a basic MZI and a planar waveguide, highly uncorrelated random speckle patterns are obtained in the time and spatial domains. By pre-calibrating the transfer matrix, accurate spectral reconstruction of unknown spectra can be achieved.

[0041] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A high-resolution spatiotemporal random speckle microspectrometer, characterized in that, It includes the beam-splitting region (1), the interferometer array region (2), and the tunable diffraction region (3). The beam splitting region (1) is used to split the input light into N optical paths and guide them to the interferometer array region (2). The interferometer array region (2) is used to perform phase modulation on the input light to different degrees, thereby generating a certain degree of extinction ratio in the optical path transmission spectrum. The tunable diffraction region (3) is a spatial light diffraction region composed of a planar waveguide and an electrode array. The electrode array is an N-row M-column square array or a randomly arranged array, used to decorrelate the input light. During operation, full-pass light is first input at the input end. Spatial speckle is obtained at the output port through spatial diffraction and modulation of the local refractive index. Random voltages of different powers are applied to the electrodes of the interferometer array region (2) and the tunable diffraction region (3). After each round of voltage application, a set of spatial speckles is obtained. The spatial speckles obtained by applying different batches of random voltages are combined in the time domain to obtain the transmission matrix. When an unknown spectrum is input, the same random power voltage is applied to the interferometer array and the electrodes for modulation, and then the optical power data of the output port is collected. Thus, the unknown spectral information is reconstructed by using a reconstruction algorithm.

2. The high-resolution spatiotemporal random speckle microspectrometer according to claim 1, characterized in that, The beam splitting region (1) includes multiple cascaded Y branches.

3. A high-resolution spatiotemporal random speckle microspectrometer according to claim 1, characterized in that, The beam splitting region (1) is a 1:N multimode interferometer.

4. A high-resolution spatiotemporal random speckle microspectrometer according to claim 1, characterized in that, The interferometer array region (2) includes N Mach-Zehnder interferometers or N micro-ring resonators. Each Mach-Zehnder interferometer has one or both arms with a thermoelectric electrode to apply a voltage of random power for thermal modulation. The two arms are of equal or unequal length, thereby modulating the optical path to different degrees.

5. A high-resolution spatiotemporal random speckle microspectrometer according to claim 1, characterized in that, The optical output section includes N equally spaced or randomly spaced output ports, and an external detector array enables parallel measurement of the N optical signals.

6. A high-resolution spatiotemporal random speckle microspectrometer according to claim 1, characterized in that, The reconstruction algorithm is the CVX algorithm toolbox in Matlab, and it uses the hyperparameters included in the K-fold cross-validation selection algorithm.

Images