Preparation method of filter chip for SF6 gas identification

Abstract

The invention relates to the technical field of infrared filter chip preparation, and discloses a preparation method of a filter chip for SF6 gas identification, which comprises the following steps of: respectively designing corresponding film system theoretical stacking sequences based on a detection wave band and a reference wave band of SF6 gas, and simulating and generating a standard spectrum track reflecting an expected spectrum state after deposition of each film layer according to the film system theoretical stacking sequences; sequentially depositing each film layer in an iteration mode, obtaining an actual measurement spectrum curve of the current film system after each layer is deposited, extracting characteristic parameters of the actual measurement spectrum curve, and comparing the characteristic parameters with theoretical values of corresponding positions in a standard spectrum track; according to the deviation value obtained through calculation and the variation trend of the deviation value, process parameters of the subsequent film layer to be deposited are dynamically adjusted, and closed-loop control is formed. The problem of spectral performance deviation caused by error accumulation in a traditional preparation process is effectively solved, the spectral precision and product consistency of the filter chip can be remarkably improved, and then the accuracy and reliability of SF6 gas detection are enhanced.

Description

Technical Field

[0001] This invention relates to the field of infrared filter chip fabrication technology, and in particular to a method for fabricating a filter chip for SF6 gas identification. Background Technology

[0002] In the field of power equipment safety monitoring, sulfur hexafluoride (SF6) gas is widely used due to its excellent insulation properties. However, its leakage poses a serious threat to equipment safety and the environment. Infrared imaging-based gas leak detection technology, especially utilizing the strong absorption characteristics of SF6 gas in specific infrared bands (such as 10.55 μm), has become a key means to achieve long-distance, visualized live-line detection. One of the core components of this technology is the infrared filter chip, which acts like a precision spectral sieve. This dual-channel infrared filter chip only allows light from the characteristic absorption band of SF6 gas and a reference band for comparison (such as 8.05 μm) to pass through, thus making the invisible gas visible in the camera. Currently, these high-performance filter chips are mainly achieved by depositing dozens of thin films of different thicknesses and materials (i.e., film structures) on a silicon substrate. The fabrication process directly determines the performance and reliability of the final product.

[0003] However, the fabrication of this filter chip faces a severe challenge: how to ensure that the spectral performance of the final product is highly consistent with the initial theoretical design. In the complex process of multilayer film deposition, any tiny process fluctuations—such as nanoscale deviations in film thickness or subtle changes in the material's refractive index—will generate cumulative errors between the film layers.

[0004] Existing technologies typically employ an extreme value method of "depositing one layer and measuring one layer" to control the thickness of a single film layer. While this method can control the thickness of a single film layer, it lacks global monitoring of the overall spectral evolution process. This localized, post-hoc control is difficult to compensate for the cascading effects of errors in the preceding film layers on subsequent film systems. Consequently, the fabricated filter chips generally suffer from problems such as center wavelength drift and reduced transmittance, which reduces the contrast and signal-to-noise ratio of gas imaging and severely affects the accuracy and sensitivity of SF6 leak detection. Summary of the Invention

[0005] Therefore, the technical problem to be solved by the present invention is to overcome the defect of spectral performance degradation caused by the accumulation of interlayer errors in the preparation of infrared filter chips in the prior art, and to provide a method for preparing a filter chip for SF6 gas identification. This method can dynamically correct the deposition process by introducing closed-loop control of standard spectral trajectory, thereby obtaining a measured spectral curve that is highly consistent with the theoretical design, and improving the accuracy and reliability of gas detection.

[0006] To address the aforementioned technical problems, this invention provides a method for fabricating a filter chip for SF6 gas identification. The filter chip includes a first film structure and a second film structure deposited on a silicon substrate. The first film structure corresponds to the detection band of SF6 gas, and the second film structure corresponds to the reference band. The fabrication method includes the following steps: Based on the detection band and the reference band, theoretical stacking sequences of the first and second membrane structures were designed respectively. Based on the theoretical stacking sequence, the real-time theoretical spectral curve of the filter chip after each film layer is deposited during the coating process is simulated and calculated. The characteristic parameters of each real-time theoretical spectral curve are extracted, and all characteristic parameters are arranged in the coating order to form a standard spectral trajectory. According to the theoretical stacking sequence, the first layer of the first or second film system is deposited on a silicon substrate. Each subsequent film layer is deposited in an iterative manner. After each film layer is deposited, the following operations are performed: the transmittance of the currently deposited film system is measured using an online spectral measurement system to obtain the current measured spectral curve, and the current measured characteristic parameters corresponding to the standard spectral trajectory are extracted; the current measured characteristic parameters are compared with the theoretical characteristic parameters of the corresponding film layer positions in the standard spectral trajectory, the deviation value is calculated, and the deposition control parameters of the subsequent film layers are dynamically adjusted according to the deviation value and its changing trend. Repeat the above iterative deposition process until both the first and second membrane structures are completely deposited, so that the spectral evolution of the entire membrane structure during the deposition process follows the standard spectral trajectory generated by simulation calculation.

[0007] In one embodiment of the present invention, based on the probe band and the reference band, theoretical stacking sequences of the first film structure and the second film structure are designed, including the following steps: High-refractive-index materials and low-refractive-index materials were selected, and the optical constants of the high-refractive-index materials and low-refractive-index materials in the detection band and reference band were determined respectively. Construct a basic film unit, including a resonant layer at the center, with multiple reflective layers symmetrically arranged on both sides. The reflective layers are formed by alternating stacking of high-refractive-index materials and low-refractive-index materials at a thickness of one-quarter of the optical thickness. Based on the target center wavelength of the detection band, the initial stacking sequence of the first membrane structure is designed by cascading the basic membrane units multiple times; based on the target center wavelength of the reference band, the initial stacking sequence of the second membrane structure is designed by cascading the basic membrane units multiple times. The initial stacking sequences of the first and second membrane structures are set as the optimization starting points, respectively. With passband rectangularity and peak transmittance as optimization objectives, the optimization algorithm is run to iteratively adjust the thickness of each membrane layer in the two initial stacking sequences. When the optimization results simultaneously satisfy the matching of the passband center of the first membrane structure with the probe band and the matching of the passband center of the second membrane structure with the reference band, and the passband rectangularity and peak transmittance of both membrane structures reach the preset index, the final membrane thickness sequence obtained at this time is determined as the theoretical stacking sequence of the first membrane structure and the second membrane structure, respectively.

[0008] In one embodiment of the present invention, the feature parameters include the center wavelength and full width at half maximum (FWHM). The feature parameters extracted from each real-time theoretical spectral curve specifically include: Identify the peak point with the highest transmittance in the real-time theoretical spectral curve and record its corresponding wavelength as the center wavelength; On the real-time theoretical spectral curve, find two wavelength points where the transmittance is half of the peak transmittance, and calculate the width between these two wavelength points as the full width at half maximum (FWHM).

[0009] In one embodiment of the present invention, after constructing the standard spectral trajectory, a dynamic tolerance range is further set for the theoretical characteristic parameters of each film layer position in the trajectory; the dynamic tolerance range is determined according to the sensitivity of the current film layer in the overall film system structure, and the higher the sensitivity, the smaller the tolerance range. When comparing the current measured characteristic parameters with the theoretical characteristic parameters of the corresponding film position in the standard spectral trajectory, if the deviation value exceeds the dynamic tolerance range, the operation of dynamically adjusting the deposition control parameters of the subsequent film to be deposited is triggered.

[0010] In one embodiment of the present invention, the deposition control parameters of the subsequent film layer to be deposited are dynamically adjusted based on the deviation value and its changing trend, specifically including: When the deviation value indicates that the center wavelength of the current measured spectrum has systematically shifted relative to the theoretical value, the theoretical design thickness of all subsequent films to be deposited is increased or decreased proportionally according to the direction and magnitude of the shift, so as to compensate for the systematic shift on a macroscopic level.

[0011] In one embodiment of the present invention, the variation law of deviation value after several consecutive film layers are analyzed. If it is determined that the deviation trend is caused by a single film layer deposition error and its influence is localized, then the deposition control parameters of the subsequent one or several specific film layers with the highest optical coupling with the erroneous film layer are finely adjusted for localized correction.

[0012] In one embodiment of the present invention, the deposition control parameter is the deposition rate, and dynamically adjusting the deposition control parameters of subsequent films to be deposited specifically includes: If the deviation value indicates insufficient film density, the deposition rate of subsequent films to be deposited should be reduced. If the deviation value indicates that the film stress is too high, the deposition rate of the subsequent film layers to be deposited should be increased.

[0013] In one embodiment of the present invention, after each film layer is deposited, the fluctuation of the current measured spectral curve is continuously monitored, and the feature parameter extraction and comparison operation is performed only when the fluctuation amplitude of the current measured spectral curve tends to stabilize.

[0014] In one embodiment of the present invention, spectral reference calibration is performed before depositing the first film structure and the second film structure. When depositing the first film structure and the second film structure respectively, the same calibrated online spectral measurement system is used to measure the transmittance of the currently deposited film structure.

[0015] In one embodiment of the present invention, after all film layers have been deposited, the prepared filter chip is subjected to a final spectral scan using a high-precision Fourier transform infrared spectrometer in a constant temperature and humidity environment. The final measured spectrum is compared with the theoretically designed spectrum across the entire band. Based on the degree of agreement of the comparison results, the filter chip is classified into performance levels.

[0016] The technical solution of the present invention has the following advantages compared with the prior art: The method for fabricating a filter chip for SF6 gas identification described in this invention no longer views the deposition of each film layer in isolation, but rather regards the entire coating process as a complete growth process and pre-sets a standard growth trajectory for it—that is, a standard spectral trajectory composed of the real-time theoretical spectral curves that should exist after each layer is deposited, which are simulated based on the theoretical film system sequence.

[0017] At the outset of the fabrication process, a first film layer is deposited on a silicon substrate. Starting from this point, an iterative cycle begins. After each film layer is deposited, the next layer is not immediately started. Instead, an integrated online spectral measurement system performs a full-body scan of the current semi-finished product, obtaining its current measured spectral curve. This measured curve is then compared with the theoretical curve at the corresponding position on the standard growth trajectory, calculating the deviation value of the characteristic parameters. This deviation value is no longer an isolated number but reveals the degree and trend of the overall film structure deviating from the expected direction. Based on this diagnostic result, the deposition control parameters of subsequent films are proactively and proactively adjusted dynamically. Subsequent operations compensate for and correct the deviations generated earlier, forming a self-correcting closed loop. By repeating this iterative process until all film layers are deposited, this invention effectively constrains the spectral evolution of the film structure firmly to a pre-set theoretical trajectory.

[0018] Through the synergistic effect of these technical features, this invention fundamentally overcomes the problem of spectral performance degradation caused by error accumulation in existing technologies. Its most significant benefit is a substantial improvement in the fabrication precision and product consistency of the filter chip, ensuring a high degree of agreement between the final measured spectral curve and the theoretical design. The filter chip fabricated using this method can more accurately separate the characteristic information of SF6 gas, thereby significantly improving the signal-to-noise ratio and recognition accuracy of the subsequent binocular imaging system when detecting SF6 gas. Simultaneously, this forward-looking adjustment strategy reduces reliance on extremely stringent initial conditions of the coating process, improving process robustness and yield, and laying a solid foundation for high-performance, low-cost SF6 gas leak detection equipment. Attached Figure Description

[0019] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings, wherein: Figure 1 This is a flowchart of the steps involved in the fabrication of the filter chip for SF6 gas identification according to the present invention. Figure 2 This is a flowchart illustrating the steps involved in designing the theoretical stacking sequence of the first and second membrane structures according to the present invention. Detailed Implementation

[0020] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.

[0021] In existing technologies, the fabrication process of infrared filter chips mainly employs a layer-by-layer deposition combined with an extreme value method for controlling the thickness of a single layer. This method terminates deposition by monitoring the thickness of a single layer to reach a preset threshold. While this ensures the thickness accuracy of a single layer, it cannot globally track the evolution of the optical performance of the entire film system. Due to the complex coupling effects of the optical properties of multilayer thin films, even small errors in preceding layers can have a cumulative effect on the spectral response of subsequent layers through optical interference, ultimately leading to defects such as passband shift and decreased transmittance in the finished product. This defect is particularly pronounced in dual-channel filter chips, where the spectral coupling effect between the two channels further amplifies process errors, causing a deterioration in the signal-to-noise ratio of the gas detection system.

[0022] Through experimental analysis of the optical superposition mechanism of multilayer films, it was found that controlling the thickness of a single layer alone is insufficient to offset the error accumulation effect. To solve this problem, the following two major ideas are proposed: A spectral evolution trajectory model was established to dynamically compare the real-time optical response during the coating process with theoretical predictions. Further, we envision using a closed-loop feedback mechanism to correct process parameters in real time during the deposition process, ensuring that the optical properties of the actual coating process always develop along a preset trajectory. This approach breaks through the passive control mode of traditional processes and instead adopts an active optical performance tracking strategy.

[0023] Therefore, based on the above concept, and referring to Figure 1 As shown, this application proposes an iterative deposition method based on standard spectral trajectory tracking. This method first establishes a standard spectral trajectory through theoretical simulation, then measures the spectral characteristics in real time after each layer is deposited and compares them with the trajectory nodes. The subsequent process parameters are dynamically adjusted according to the deviation. Through closed-loop control, the spectral evolution of the actual coating process is always close to the theoretical trajectory, thereby eliminating the error accumulation effect.

[0024] Among them, the theoretical stacking sequence refers to the sequence of film materials and thicknesses determined through optical simulation. Specifically, it can be designed using the transfer matrix method combined with optimization algorithms to construct film structures with target spectral responses. The standard spectral trajectory refers to the set of theoretical spectral characteristic parameters for each stage of the coating process. Specifically, it can be generated through time-domain spectral simulation technology and serves as a benchmark for the evolution of optical performance during the coating process. Dynamically adjusting deposition control parameters refers to correcting subsequent process conditions based on real-time spectral deviations. Specifically, it can be implemented using adaptive control algorithms to compensate for existing process deviations and suppress error propagation.

[0025] Specifically, the fabrication process first designs theoretical stacking sequences for the first and second film systems based on the probe and reference wavelengths, respectively. Then, a theoretical model of the dual-channel film system is established through optical simulation. Based on the theoretical stacking sequence, the real-time theoretical spectral curves of the filter chip after each film layer deposition are simulated and calculated. Characteristic parameters of each real-time theoretical spectral curve are extracted, and all characteristic parameters are arranged in the deposition sequence to form a standard spectral trajectory. After deposition begins on the silicon substrate, transmission spectral measurements are performed after each film layer is completed. The characteristic parameters of the current spectrum are extracted, and these measured features are compared with the theoretical values ​​of corresponding nodes in the standard trajectory. When a characteristic parameter deviates from a predetermined range, the deposition rate or thickness design of subsequent film layers is automatically adjusted. For example, if a continuous shift of the center wavelength towards longer wavelengths is detected, the theoretical thickness of subsequent film layers can be reduced proportionally, and the spectral response can be pulled back to the predetermined trajectory by changing the optical interference conditions. This closed-loop control runs throughout the entire deposition process, ensuring that the overall optical performance of the film system after dozens of thin film layers are deposited still accurately matches the design requirements.

[0026] Compared to existing technologies, traditional processes only focus on the static control of single-layer film thickness, while this solution establishes a dynamic tracking mechanism for optical performance throughout the entire process. Existing technologies are like building a wall piece by piece according to blueprints without monitoring the overall tilt of the wall. This solution, however, dynamically adjusts the construction plan by monitoring the changes in the center of gravity of the building structure in real time. This shift from local control to global regulation effectively suppresses the chain propagation of film layer errors and solves the spectral coupling problem in the fabrication of dual-channel filter chips.

[0027] Through the above technical solution, this application effectively controls the optical performance drift during the coating process, reducing the deviation between the actual passband center wavelength of the dual-channel filter chip and the design value by an order of magnitude, while ensuring the spectral isolation between the two channels and controlling inter-channel crosstalk within a negligible range. This fabrication method significantly improves the transmittance uniformity and passband edge steepness of the filter chip, providing a core optical component with stable spectral characteristics for SF6 gas detection systems.

[0028] Reference Figure 2 As shown, this application further proposes a method for designing a theoretical stacking sequence of a first film structure and a second film structure based on the detection band and reference band, respectively. The method includes the following steps: selecting high-refractive-index and low-refractive-index materials and determining their optical constants in the detection and reference bands; constructing a basic film unit, including a resonant layer at the center, with multiple reflective layers symmetrically arranged on both sides, the reflective layers being formed by alternating stacking of high-refractive-index and low-refractive-index materials at a quarter optical thickness; and, based on the target center wavelength of the detection band, initially stacking the first film structure by cascading the basic film unit multiple times. Stack sequence design: Based on the target center wavelength of the reference band, the initial stacking sequence of the second membrane structure is designed by cascading the basic membrane units multiple times; the two initial stacking sequences are set as optimization starting points, and the optimization algorithm is run with passband rectangularity and peak transmittance as optimization targets to iteratively adjust the thickness of each membrane layer in the two initial stacking sequences; when the optimization results simultaneously satisfy the matching of the passband center of the first membrane structure with the probe band, the matching of the passband center of the second membrane structure with the reference band, and the passband rectangularity and peak transmittance both reach the preset indicators, the final membrane thickness sequence is determined as the theoretical stacking sequence.

[0029] Among them, high-refractive-index and low-refractive-index materials refer to two coating materials with significant differences in refractive index, specifically achieved by combining zinc sulfide and yttrium fluoride. Their differentiated optical properties can form an effective optical interference structure. The basic film unit refers to a periodic combination of film layers with a symmetrical reflective layer structure, specifically achieved by alternately depositing high- and low-refractive-index material layers with a quarter optical thickness on both sides of the resonant layer. This structure can enhance the transmission selectivity of specific wavelengths. Multiple cascading refers to repeatedly stacking the basic film unit according to a specific pattern. Specifically, a mathematical recursive model can be used to calculate the number of cascades, thereby expanding the passband width and improving the spectral response sharpness. The optimization algorithm refers to a calculation method based on iterative adjustment of the film thickness according to the objective function. Specifically, genetic algorithms or particle swarm optimization algorithms can be used to achieve the global optimum of the passband performance index through multi-parameter collaborative optimization.

[0030] Specifically, after determining the optical constants of the high and low refractive index materials, a basic film system unit containing a resonant layer and a symmetrical reflective layer is first constructed. For example, when the center wavelength of the detection band is 10.55 μm, the physical thickness of the resonant layer corresponding to a quarter wavelength can be calculated. The reflective layer is composed of alternating stacks of high and low refractive index materials, with the thickness of each layer set to a quarter of the optical thickness of the corresponding material at the target wavelength. By cascading the basic film system unit multiple times, an initial stacking sequence is formed. Subsequently, with passband rectangularity and peak transmittance as optimization objectives, an optimization algorithm is run to fine-tune the thickness of each film layer in the initial stacking sequence. When the optimized film system structure simultaneously meets the requirements of passband center wavelength matching, passband edge steepness, and transmittance, the theoretical stacking sequence is obtained.

[0031] Compared with existing technologies, traditional methods only calculate film thickness using empirical formulas without considering multi-objective collaborative optimization, resulting in insufficient passband edge steepness and large transmittance fluctuations. This solution introduces basic film system units and optimization algorithms to simultaneously improve passband rectangularity and transmittance while ensuring the accuracy of the passband center wavelength, thus solving the problem of difficulty in achieving both spectral performance indicators in existing technologies.

[0032] Through the above technical solution, this application can precisely control the film stacking structure of the dual-channel filter chip, so that the passband center of the two film structures is strictly matched with the SF6 gas absorption peak and the reference band, respectively, and at the same time obtain a passband shape with high rectangularity and stable peak transmittance, providing a reliable theoretical design basis for subsequent coating processes.

[0033] Specifically, the feature parameters described in this embodiment include the center wavelength and the full width at half maximum (FWHM). The feature parameters of each real-time theoretical spectral curve are extracted, specifically including: identifying the peak point with the highest transmittance in the real-time theoretical spectral curve and recording its corresponding wavelength as the center wavelength; finding two wavelength points on the real-time theoretical spectral curve with transmittance half that of the peak transmittance, and calculating the width between these two wavelength points as the full width at half maximum (FWHM).

[0034] The center wavelength refers to the wavelength corresponding to the peak point of highest transmittance in the real-time theoretical spectral curve. This can be achieved by automatically identifying the highest point of the transmittance curve and recording its wavelength coordinates using a spectral analysis algorithm. This parameter characterizes the center position of the passband in the current deposition state of the filter chip, ensuring its match with the target band. The full width at half maximum (FWHM) refers to the width between two wavelength points corresponding to when the transmittance drops to half of the peak value. This can be achieved by searching for the position where the transmittance equals 50% of the peak value on the real-time theoretical spectral curve and calculating the wavelength difference. This parameter quantifies the passband width, reflecting the precision of the film stacking in controlling the spectral shape.

[0035] Specifically, after each film layer is deposited during the coating process, a real-time theoretical spectral curve is generated through simulation. Peak points are identified by traversing transmittance data points, for example, using a local maximum detection algorithm to pinpoint the wavelength corresponding to the highest transmittance. For calculating the full width at half maximum (FWHM), a 50% threshold for peak transmittance is first determined. Then, the wavelengths where transmittance first falls below this threshold are searched along the real-time theoretical spectral curve; the difference between these two points is the FWHM. These two parameters are continuously recorded and arranged according to the coating sequence to form a standard spectral trajectory, providing a benchmark for real-time deviation analysis during subsequent deposition processes.

[0036] Compared to existing technologies, traditional methods only monitor whether the thickness of a single layer reaches a preset value. This approach, however, extracts two key spectral features—the center wavelength and full width at half maximum (FWHM)—to directly reflect the dynamic evolution of passband characteristics during multilayer film stacking. Existing technologies cannot capture the cumulative errors caused by optical coupling effects between film layers. This approach, by tracking global changes in the standard spectral trajectory, can detect systematic deviations at an early stage and trigger adjustments, preventing errors from propagating to subsequent layers.

[0037] Through the above technical solution, this application achieves real-time tracking and quantitative evaluation of the spectral characteristics during the coating process of filter chips. Specifically, continuous monitoring of the center wavelength can promptly detect passband shifts caused by film thickness or refractive index deviations, while dynamic analysis of the full width at half maximum (FWHM) can effectively identify spectral broadening caused by film stacking asymmetry or interface scattering. This dual-parameter monitoring mechanism based on spectral characteristics provides a reliable basis for dynamically adjusting the deposition process, thereby improving the consistency of the spectral performance of the filter chip.

[0038] In the actual fabrication process, to avoid overly sensitive adjustments, a dynamic tolerance mechanism is introduced to achieve graded control of film sensitivity. This ensures the accuracy of critical films while reducing the process cost of non-critical films. Specifically, this is achieved through the following method: a dynamic tolerance range is set for the theoretical characteristic parameters of each film position in the trajectory; the dynamic tolerance range is determined based on the sensitivity of the current film in the overall film system structure, with higher sensitivity resulting in a smaller tolerance range; when comparing the current measured characteristic parameters with the theoretical characteristic parameters of the corresponding film position in the standard spectral trajectory, if the deviation exceeds the dynamic tolerance range, the deposition control parameters of subsequent films to be deposited are dynamically adjusted.

[0039] The dynamic tolerance range refers to the maximum range within which measured characteristic parameters are allowed to deviate from theoretical characteristic parameters. The width of this range is dynamically set according to the optical contribution weight of the film layer in the overall film system. Specifically, it can be achieved by statistical analysis of historical deposition data or by calculating the sensitivity coefficient using an optical simulation model. Sensitivity refers to the degree of influence of a specific film layer on the overall spectral performance. Specifically, it can be quantitatively evaluated by simulating the perturbation amplitude of different film layer thickness deviations on the final passband characteristics using an optical transfer matrix.

[0040] Specifically, during the film deposition process, after each film layer is deposited, the online spectral measurement system acquires the transmittance curve of the current film system in real time and extracts the center wavelength and full width at half maximum (FWHM) as characteristic parameters. These measured characteristic parameters are immediately compared with the theoretical characteristic parameters at the corresponding positions in the standard spectral trajectory. When the deviation exceeds the preset dynamic tolerance range, the system automatically adjusts the deposition rate or thickness design value of subsequent film layers according to the sensitivity level of the current film layer. For example, for a high-sensitivity film layer located at the core of the resonant cavity, its tolerance range is set to ±0.5 nm, while the tolerance range of the edge reflective layer can be widened to ±2 nm. This differentiated tolerance control allows the process fluctuations of key film layers to be corrected in a timely manner, preventing errors from being transmitted to subsequent film layers.

[0041] Through the above technical solution, this application effectively solves the problem of spectral performance degradation caused by error accumulation during multilayer film deposition. By dynamically adjusting the process tolerance of different sensitivity film layers, the overall process stability can be optimized while ensuring the accuracy of key film layers, thereby improving the spectral consistency and fabrication yield of the filter chip.

[0042] Specifically, this application proposes a dynamic adjustment of deposition control parameters for subsequent film layers based on deviation values ​​and their changing trends, including the following two dynamic adjustment strategies: The first dynamic adjustment strategy is as follows: when the deviation value indicates that the center wavelength of the current measured spectrum has systematically drifted relative to the theoretical value, the theoretical design thickness of all subsequent films to be deposited is increased or decreased proportionally according to the direction and magnitude of the drift, so as to compensate for the systematic drift on a macroscopic level.

[0043] Systematic drift refers to the phenomenon where the center wavelength continuously deviates from the theoretical value due to deviations in coating equipment parameters or environmental factors. This can be identified by detecting the wavelength shift trend after the deposition of multiple consecutive film layers using an online spectral measurement system. For example, if the center wavelengths of three consecutive film layers all shift towards longer wavelengths exceeding a threshold, it is considered systematic drift. The theoretical design thickness refers to the target thickness value of each film layer calculated using a film structure optimization algorithm. This can be achieved through simulation and iterative calculations using optical thin film design software. For instance, bandgap characteristics can be optimized by adjusting the thickness ratio of high-refractive-index material layers to low-refractive-index material layers.

[0044] Specifically, during the coating process, when the online spectral measurement system detects a continuous unidirectional shift in the center wavelength of the deposited film, it is determined that there is a systematic error caused by temperature drift or abnormal evaporation source rate within the coating cavity. At this time, based on the ratio between the measured shift and the theoretical value, the thickness of all subsequent undeposited film layers is scaled up as a whole. For example, if the measured center wavelength is 2% shorter than the theoretical value, the design thickness of each subsequent film layer is increased by 2%. By globally adjusting the film layer thickness sequence, the overall optical performance of the final film structure returns to the theoretical design trajectory, avoiding the error being amplified step by step during the multilayer stacking process.

[0045] Compared with existing technologies, traditional extreme value methods only correct errors for single-layer film thickness and cannot eliminate the cumulative errors caused by coating process parameter drift. In contrast, this solution monitors the spectral evolution trend in real time and performs global thickness compensation for subsequent film layers when a systematic deviation is detected. This achieves error correction from the overall structure level of the film system and effectively suppresses the impact of center wavelength drift on the final filtering performance.

[0046] The second dynamic adjustment strategy is to analyze the variation of deviation values ​​after the deposition of several consecutive film layers. If it is determined that the deviation trend is caused by a single film layer deposition error and its influence is localized, then the deposition control parameters of the subsequent one or several specific film layers with the highest optical coupling to the erroneous film layer are finely adjusted for localized correction.

[0047] Among these, the variation pattern of deviation values ​​after the deposition of several consecutive film layers refers to identifying the evolution pattern of deviations by monitoring the characteristic parameter deviation sequences of multiple adjacent film layers after deposition. Specifically, a sliding window algorithm can be used to fit the trend of deviation values ​​of adjacent film layers to distinguish between systematic and local errors. The local impact caused by a single film layer deposition failure refers to the fact that after the thickness or refractive index of a certain film layer deviates from the theoretical value, its optical effect only significantly interferes with a specific subsequent film layer. Specifically, the degree of mutual influence between film layers can be quantified by establishing an optical coupling matrix between film layers to accurately locate the subsequent film layer that needs correction. The specific subsequent film layer with the highest optical coupling degree refers to the subsequent film layer in the multilayer film structure that has a strong optical interference with the erroneous film layer. Specifically, it can be determined by calculating the electric field intensity distribution or transmission phase matching degree between film layers to screen out the correction target most sensitive to the error. Fine-tuning the deposition control parameters refers to making minor adjustments to the deposition rate or deposition time of the selected subsequent film layer. Specifically, a PID control algorithm can be used to generate compensation commands based on the deviation amplitude to eliminate local errors with minimal intervention.

[0048] Specifically, when the online spectral measurement system detects an abnormal deviation in the spectral characteristic parameters of a certain film layer after deposition, it first traces the deviation data of the first three films and calculates the rate of change of deviation to determine whether there is a single-point sudden increase trend. If it is confirmed that the abnormality is caused by the deposition error of the current film layer, it calls the pre-stored optical coupling model to screen out the two subsequent films that have a strong coupling relationship with the current film layer. For example, when the fifth high-refractive-index film is deposited too thickly, the model recognizes that the optical phase of the seventh low-refractive-index film will be most disturbed. At this time, the deposition time of the seventh layer is shortened by a predetermined proportion, while the parameters of other subsequent films remain unchanged. This local correction operation avoids excessive adjustment of the overall film system structure by suppressing the transmission of errors between specific films.

[0049] Compared to existing technologies, traditional extremum methods only perform post-hoc corrections on the thickness of a single film layer, failing to identify the source and propagation path of errors. This often leads to new accumulated errors due to the use of global compensation strategies. Our proposed solution, by establishing a coupling relationship model between film layers, can accurately locate subsequent film layers affected by errors, narrowing the correction scope from the entire film system to a specific local region. This significantly reduces process adjustment costs while maintaining spectral performance.

[0050] In this embodiment, the deposition control parameter is the deposition rate. The deposition control parameters of the subsequent films to be deposited are dynamically adjusted, specifically including: if the deviation value indicates that the film density is insufficient, the deposition rate of the subsequent films to be deposited is reduced; if the deviation value indicates that the film stress is too high, the deposition rate of the subsequent films to be deposited is increased.

[0051] The deposition rate refers to the amount of material deposited on the substrate surface per unit time. It can be achieved by adjusting the sputtering power or gas flow rate in ion beam sputtering processes. The deposition rate directly affects the density of the film's microstructure and the distribution of internal stress. Insufficient film density indicates the presence of pores or a loose structure within the film, which can be identified by increased scattering loss at non-characteristic wavelengths in the transmission spectrum. Excessive film stress refers to internal stress between the film and the substrate caused by differences in thermal expansion coefficients or lattice mismatch, which can be judged by edge warping or cracking.

[0052] Specifically, during the coating process, when the online spectral measurement system detects a deviation between the center wavelength or full width at half maximum (FWHM) of the current film layer and the theoretical value, the physical cause of the film layer defect can be determined by analyzing the type of deviation. For example, if the transmittance of the measured spectral curve outside the target band is abnormally high, it indicates that the film layer is not dense enough. In this case, by reducing the deposition rate of subsequent film layers, the migration time of material particles on the substrate surface can be extended, promoting the formation of a denser stacked structure. Conversely, if microcracks or substrate deformation appear on the film layer surface, it indicates that the film layer stress is too high. In this case, increasing the deposition rate can reduce the heat accumulation during the film growth process and alleviate the stress concentration phenomenon. By adjusting the deposition rate in real time, differentiated compensation can be implemented for different defect types, so that the optical performance of the film system gradually returns to the theoretical trajectory.

[0053] By dynamically adjusting the deposition rate, the film quality can be corrected in real time without interrupting the coating process, ensuring that the dual-channel filter chip has stable transmittance and spectral resolution in both the SF6 characteristic absorption band and the reference band.

[0054] In actual processes, after the physical vapor deposition of each film layer is completed, the temperature distribution in the vacuum chamber is not yet completely uniform, and the microstructure of the newly deposited film layer is still in a metastable state. If feature extraction and comparison are performed at this time, measurement errors are likely to occur due to transient changes on the film layer surface. To solve this problem, this application further proposes to continuously monitor the fluctuation of the current measured spectral curve after each film layer is deposited, and only perform feature parameter extraction and comparison operations after the fluctuation amplitude of the current measured spectral curve tends to stabilize.

[0055] The continuous monitoring of fluctuations in the measured spectral curve refers to the continuous sampling of transmittance multiple times using an online spectral measurement system after the deposition of a single film layer, forming a time-series trajectory of spectral data changes. This can be achieved using a high-sampling-rate spectral sensor in conjunction with a data acquisition module, for example, acquiring transmittance data 10 times per second and recording the wavelength-transmittance curve. This monitoring process is used to capture transient changes in optical properties on the film surface caused by thermal stress release or material crystallization after coating. The stabilization of fluctuation amplitude means that the change in the morphological parameters of the measured spectral curve decreases to within a preset threshold range. This can be achieved using the standard deviation calculation method; for example, a stable state is determined when the standard deviation of the center wavelength after five consecutive samples is less than 0.2 nanometers. This criterion is used to exclude short-term interference caused by temperature fluctuations or residual gas adsorption within the coating chamber.

[0056] Through the above technical solution, this application solves the problem of measurement data distortion caused by material phase transformation hysteresis during the coating process, ensuring that the optical performance evaluation of each film layer is based on real data under stable conditions. This technical means can significantly reduce the risk of misadjustment of control parameters due to improper measurement timing and improve the control accuracy of the film structure deposition process.

[0057] Specifically, this application further proposes to perform spectral reference calibration before depositing the first and second film structures, and to use the same calibrated online spectral measurement system to measure the transmittance of the currently deposited film structure when depositing the first and second film structures respectively.

[0058] Spectral benchmark calibration refers to calibrating the wavelength accuracy and transmittance linearity of the online spectral measurement system using standard samples. Specifically, a standard filter with known optical constants can be used as a benchmark. Before deposition, the measured data of the online spectral measurement system is compared with the calibrated spectrum of the standard filter, and the system parameters are adjusted until the measurement error is below a preset threshold. A calibrated online spectral measurement system refers to the same spectral measurement equipment after a unified calibration process, where the optical probe, light source, and detector maintain fixed installation positions and parameter configurations during both film deposition processes, thereby ensuring consistency of the measurement conditions between the two measurements.

[0059] Specifically, before depositing the first film structure, a standard filter is placed on the silicon substrate, and an online spectral measurement system is activated for scanning. If the measured center wavelength of the spectrum deviates from the calibration value of the standard filter, the wavelength calibration parameters of the system are adjusted until the deviation is eliminated. After calibration, the system is locked as the sole device for subsequent film layer measurements. When depositing the second film structure, the same system under the same calibration state is directly used for transmittance measurement, avoiding additional errors introduced by changing equipment or adjusting the optical path. For example, when depositing the first film structure, the system may experience detector response drift due to changes in ambient temperature. In this case, calibration can correct the drift in real time, and the deposition process of the second film structure directly inherits the corrected parameters, thereby maintaining the consistency of the measurement reference for the two channels.

[0060] This application ensures that the transmittance measurement data of the dual-channel membrane structure is based on the same reference, avoiding the spectral response mismatch of the two channels due to differences in the measurement system, thereby improving the dual-channel contrast accuracy of the filter chip in SF6 gas identification.

[0061] Specifically, after all film layers are deposited, the prepared filter chip is subjected to a final spectral scan using a high-precision Fourier transform infrared spectrometer in a constant temperature and humidity environment. The final measured spectrum is compared with the theoretically designed spectrum across the entire band, and the filter chip is classified into performance levels based on the degree of agreement between the comparison results.

[0062] The constant temperature and humidity environment refers to stable testing conditions where the temperature is controlled at 20±0.5℃ and the relative humidity is maintained at 45%±3%. This can be achieved by linking a constant temperature chamber with a humidity control device. This environment eliminates spectral measurement errors caused by temperature and humidity fluctuations, ensuring the consistency of data acquisition standards. Full-band comparison involves calculating the transmittance deviation between the measured spectrum and the theoretical spectrum point-by-point within the 8-12μm range. This can be achieved by fitting the similarity index of the two curves using the least squares method. Performance level classification involves setting multiple quality thresholds based on the distribution of deviation values. For example, a product is classified as Grade A when the center wavelength shift is less than 0.1% and the passband transmittance fluctuation is less than 2%, thus establishing an objective screening standard.

[0063] Specifically, after completing the dual-channel film deposition, the filter chip is transferred to a constant temperature and humidity test chamber and left to stand for two hours. Once the internal stress of the chip is released and the substrate temperature is balanced with the environment, the spectrometer is activated for a full-band scan. During the scan, a step-by-step measurement mode is used, collecting transmittance data points at 0.1 μm intervals to generate continuous measured spectral curves. The measured data is then imported into analysis software, and point-by-point matching with the theoretically designed spectrum is calculated. The three core parameters—passband center wavelength, full width at half maximum (FWHM), and stopband rejection ratio—are evaluated. Based on the preset error tolerance range, the products are classified into three grades: A, B, and C. Grade A products can be directly used in high-precision gas detection equipment, Grade B products require secondary calibration, and Grade C products are deemed unqualified and require rework.

[0064] This application further establishes a full-band performance verification mechanism for the final product, which can accurately identify membrane structure defects caused by process fluctuations and prevent products with hidden optical distortions from flowing into the subsequent assembly stage. This graded screening mechanism significantly improves the batch consistency of filter chips and ensures that each product leaving the factory can meet the stringent requirements of SF6 gas detection system for spectral selectivity.

[0065] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A method for fabricating a filter chip for SF6 gas identification, the filter chip comprising a first film structure and a second film structure deposited on a silicon substrate, the first film structure corresponding to the detection band of SF6 gas, and the second film structure corresponding to the reference band, characterized in that, The preparation method includes the following steps: Based on the detection band and the reference band, theoretical stacking sequences of the first and second membrane structures were designed respectively. Based on the theoretical stacking sequence, the real-time theoretical spectral curve of the filter chip after each film layer is deposited during the coating process is simulated and calculated. The characteristic parameters of each real-time theoretical spectral curve are extracted, and all characteristic parameters are arranged in the coating order to form a standard spectral trajectory. According to the theoretical stacking sequence, the first layer of the first or second film system is deposited on a silicon substrate. Each subsequent film layer is deposited in an iterative manner. After each film layer is deposited, the following operations are performed: the transmittance of the currently deposited film system is measured using an online spectral measurement system to obtain the current measured spectral curve, and the current measured characteristic parameters corresponding to the standard spectral trajectory are extracted; the current measured characteristic parameters are compared with the theoretical characteristic parameters of the corresponding film layer positions in the standard spectral trajectory, the deviation value is calculated, and the deposition control parameters of the subsequent film layers are dynamically adjusted according to the deviation value and its changing trend. Repeat the above iterative deposition process until both the first and second membrane structures are completely deposited, so that the spectral evolution of the entire membrane structure during the deposition process follows the standard spectral trajectory generated by simulation calculation.

2. The method for fabricating a filter chip for SF6 gas identification according to claim 1, characterized in that: Based on the probe band and the reference band, theoretical stacking sequences of the first and second membrane structures are designed, including the following steps: High-refractive-index materials and low-refractive-index materials were selected, and the optical constants of the high-refractive-index materials and low-refractive-index materials in the detection band and reference band were determined respectively. Construct a basic film unit, including a resonant layer at the center, with multiple reflective layers symmetrically arranged on both sides. The reflective layers are formed by alternating stacking of high-refractive-index materials and low-refractive-index materials at a thickness of one-quarter of the optical thickness. Based on the target center wavelength of the detection band, the initial stacking sequence of the first membrane structure is designed by cascading the basic membrane units multiple times; based on the target center wavelength of the reference band, the initial stacking sequence of the second membrane structure is designed by cascading the basic membrane units multiple times. The initial stacking sequences of the first and second membrane structures are set as the optimization starting points, respectively. With passband rectangularity and peak transmittance as optimization objectives, the optimization algorithm is run to iteratively adjust the thickness of each membrane layer in the two initial stacking sequences. When the optimization results simultaneously satisfy the matching of the passband center of the first membrane structure with the probe band and the matching of the passband center of the second membrane structure with the reference band, and the passband rectangularity and peak transmittance of both membrane structures reach the preset index, the final membrane thickness sequence obtained at this time is determined as the theoretical stacking sequence of the first membrane structure and the second membrane structure, respectively.

3. The method for fabricating a filter chip for SF6 gas identification according to claim 1, characterized in that: The characteristic parameters include the center wavelength and full width at half maximum (FWHM). The characteristic parameters extracted for each real-time theoretical spectral curve specifically include: Identify the peak point with the highest transmittance in the real-time theoretical spectral curve and record its corresponding wavelength as the center wavelength; On the real-time theoretical spectral curve, find two wavelength points where the transmittance is half of the peak transmittance, and calculate the width between these two wavelength points as the full width at half maximum (FWHM).

4. The method for fabricating a filter chip for SF6 gas identification according to claim 1, characterized in that: After constructing the standard spectral trajectory, the method further includes setting a dynamic tolerance range for the theoretical characteristic parameters of each film layer position in the trajectory; the dynamic tolerance range is determined based on the sensitivity of the current film layer in the overall film system structure, and the higher the sensitivity, the smaller the tolerance range. When comparing the current measured characteristic parameters with the theoretical characteristic parameters of the corresponding film position in the standard spectral trajectory, if the deviation value exceeds the dynamic tolerance range, the operation of dynamically adjusting the deposition control parameters of the subsequent film to be deposited is triggered.

5. The method for fabricating a filter chip for SF6 gas identification according to claim 1, characterized in that: Based on the deviation value and its changing trend, the deposition control parameters of the subsequent film layers to be deposited are dynamically adjusted, specifically including: When the deviation value indicates that the center wavelength of the current measured spectrum has systematically shifted relative to the theoretical value, the theoretical design thickness of all subsequent films to be deposited is increased or decreased proportionally according to the direction and magnitude of the shift, so as to compensate for the systematic shift on a macroscopic level.

6. The method for fabricating a filter chip for SF6 gas identification according to claim 5, characterized in that: Analyze the variation pattern of deviation values ​​after several consecutive film layers are deposited. If it is determined that the deviation trend is caused by a single film layer deposition error and its effect is localized, then the deposition control parameters of the subsequent one or several specific film layers with the highest optical coupling to the erroneous film layer are finely adjusted for localized correction.

7. The method for fabricating a filter chip for SF6 gas identification according to claim 1, characterized in that: The deposition control parameter is the deposition rate. The deposition control parameters for subsequent layers to be deposited are dynamically adjusted, specifically including: If the deviation value indicates insufficient film density, the deposition rate of subsequent films to be deposited should be reduced. If the deviation value indicates that the film stress is too high, the deposition rate of the subsequent film layers to be deposited should be increased.

8. The method for fabricating a filter chip for SF6 gas identification according to claim 1, characterized in that: After each film layer is deposited, the fluctuation of the current measured spectral curve is continuously monitored. Only when the fluctuation amplitude of the current measured spectral curve tends to stabilize is the feature parameter extraction and comparison operation performed.

9. The method for fabricating a filter chip for SF6 gas identification according to claim 1, characterized in that: Before depositing the first and second film structures, spectral reference calibration is performed. When depositing the first and second film structures respectively, the same calibrated online spectral measurement system is used to measure the transmittance of the currently deposited film structures.

10. The method for fabricating a filter chip for SF6 gas identification according to claim 1, characterized in that: After all film layers were deposited, the prepared filter chip was subjected to a final spectral scan using a high-precision Fourier transform infrared spectrometer in a constant temperature and humidity environment. The final measured spectrum is compared with the theoretically designed spectrum across the entire band. Based on the degree of agreement of the comparison results, the filter chip is classified into performance levels.

Images