A method for integrating ceramic-based thin film circuit products

Abstract

The present invention relates to the field of semiconductor processing technology, and more specifically to a method for integrating ceramic-based thin-film circuit products. The method comprises: obtaining an etching matrix; obtaining a fractal dimension sequence for each row of data and constructing an anomaly index; converting each row of data into a frequency domain signal, performing a partitioned analysis, and constructing a local heterogeneity evaluation parameter; analyzing the coordinated changes in different types of data at each moment to construct a local consistency parameter, and combining similarities to construct a synergy factor; improving the differential coefficient of PID control; and achieving regulation of ion beam energy. The present invention aims to overcome the oscillation and instability caused by the heterogeneity of the electroplated film surface and the chain changes in parameters, improve the accuracy of ion beam energy regulation, and ensure the integration quality of new ceramic-based thin-film circuit products.

Description

Technical Field

[0001] The present application relates to the field of semiconductor processing technology, and in particular to a method for integrating ceramic-based thin film circuit products. Background Art

[0002] Ceramic-based thin-film circuit integration technology leverages the high-temperature stability, corrosion resistance, and mechanical strength of ceramic-based materials, combined with thin-film processing, to demonstrate exceptional performance in specialized applications. Its exceptional performance in high-temperature environments, corrosion resistance, and high-frequency capabilities make it an ideal choice for electronic device design in areas with extremely demanding performance requirements, such as military, healthcare, and aerospace. This technology not only offers significant advantages in miniaturization and high-density interconnection, but also provides strong support for improved system stability and innovative designs. In the future, as technology continues to advance, new ceramic-based thin-film circuit integration technologies are expected to demonstrate their importance in a wider range of applications, driving the development of electronic device technology.

[0003] In the preparation process of ceramic-based thin-film circuits, after the photolithography and circuit steps are completed, local etching will be performed to remove the seed layer and the base layer to complete the circuit integration. During the etching process, the ceramic substrate contains different material components to produce heterogeneity, resulting in inconsistent etching rates of the ion beam in different parts. In addition, due to the synergistic effect of various parameters under the etching condition, temperature fluctuations and vacuum changes during etching will cause chain changes in the process environment, further leading to unstable etching during ceramic-based thin-film circuit integration, resulting in errors, causing the actual characteristics of the ion beam to deviate from the theoretically designed parameters, thereby affecting the accuracy of etching. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention provides an integration method of ceramic-based thin film circuit products to solve the existing problems.

[0005] The present invention provides a method for integrating a ceramic-based thin film circuit product using the following technical solutions:

[0006] One embodiment of the present invention provides a method for integrating a ceramic-based thin film circuit product, the method comprising the following steps:

[0007] Obtaining various types of etching data, and using the various types of etching data as rows of an etching matrix;

[0008] Detrended fluctuation analysis is used to obtain a fractal dimension sequence for each row of data in the etching matrix; the element distribution characteristics of the fractal dimension sequence for each row of data are analyzed, and combined with a mutation point detection algorithm, the data trend within the local range of the mutation point data is analyzed to obtain an anomaly index for each row of data; the frequency domain signal for each row of data is obtained, and based on the frequency distribution range and amplitude changes in the frequency domain signal for each row of data, the stability parameter for each row of data is determined, and combined with the anomaly index, a local heterogeneity assessment parameter for each row of data is obtained;

[0009] Taking each data in each row of the etching matrix as the center, the initial length of the window is preset, and the window is evenly increased towards both ends with a preset step size. The local consistency parameter between any two rows of data in the etching matrix is ​​determined based on the consistency of the data changes in the window after the window corresponding to each data is increased; the synergy factor of the etching matrix is ​​determined by comprehensively considering the distribution similarity between different rows of data, the local consistency parameter, and the local heterogeneity evaluation parameter;

[0010] The differential coefficient of PID control is improved according to the synergistic factor of the etching matrix; the improved differential coefficient of PID control is combined with the integration of new ceramic-based thin film circuit products.

[0011] Preferably, the method of obtaining the fractal dimension sequence of each row of data in the etching matrix by using detrended fluctuation analysis includes:

[0012] Each row of the etching matrix is ​​used as the input of the multifractal detrended fluctuation analysis, and the fractal dimension curve of each row of data is output; the fractal dimension curve is uniformly sampled to obtain the fractal dimension sequence.

[0013] Preferably, the abnormality index of each row of data is obtained as follows:

[0014] Preset the local data sequence of each mutation point in the fractal dimension sequence of each row of data;

[0015] Calculate the local data eigenvalue of each mutation point based on the degree of element change in the local data sequence of each mutation point and the data increase or decrease trend;

[0016] Obtain the degree of discreteness of elements in the fractal dimension sequence of each row of data; calculate the mean of the local data eigenvalues ​​of all mutation points corresponding to each row of data; and forwardly fuse the degree of discreteness and the mean of the local data eigenvalues ​​corresponding to each row of data as the anomaly index of each row of data.

[0017] Preferably, the calculation of the local data characteristic value of each mutation point is specifically as follows:

[0018] For the fractal dimension sequence of each row of data, the slope of the straight line obtained by fitting the local data sequence of each mutation point is forward fused with the element mean of the first-order difference sequence of the local data sequence corresponding to the mutation point to obtain the local data eigenvalue of each mutation point in the fractal dimension sequence of each row of data.

[0019] Preferably, the stability parameter of each row of data is determined as follows:

[0020] Get all the peaks and troughs of the frequency domain signal for each row of data;

[0021] The absolute value of the amplitude difference between the peaks and troughs of the same sequence is recorded as the first difference; the absolute value of the frequency difference between the peaks and troughs of the same sequence is recorded as the second difference; the average value of the ratio between the first difference and the second difference under all sequences is recorded as the stability parameter of each row of data in the frequency domain.

[0022] Preferably, the local heterogeneity evaluation parameter of each row of data is specifically the product of the local data eigenvalue and the stability parameter.

[0023] Preferably, determining the local consistency parameter between any two rows of data includes:

[0024] Calculate the ratio between the window length increased each time and the maximum window length increased; obtain the negative correlation mapping result of the difference between the extreme values ​​of the data in the window after each increase of the window centered on each data in any two rows of data; use the ratio as the weight of the negative correlation mapping result corresponding to each increase of the any two rows of data, and sum them to obtain the local consistency parameter between the any two rows of data.

[0025] Preferably, the specific process of determining the synergy factor of the etching matrix is ​​as follows:

[0026] Obtaining the similarity measure and the difference of the local heterogeneity evaluation parameter between any two rows of data, and combining the local consistency parameter to obtain the synergy index between the any two rows of data;

[0027] The average of the synergy indexes obtained from all pairwise combinations of row data in the etching matrix is ​​taken as the synergy factor of the etching matrix.

[0028] Preferably, the synergy index between any two rows of data is obtained as follows:

[0029] For any two rows of data in the etching matrix, the similarity measure and the local consistency parameter are forward fused and then divided by the absolute value of the difference of the local heterogeneity evaluation parameter between the any two rows of data to obtain the synergy index between the any two rows of data.

[0030] Preferably, the improving the differential coefficient of the PID control according to the synergy factor of the etching matrix includes:

[0031] Preset the adjustment factor; obtain the normalized value of the synergy factor of the etching matrix; calculate the product of the adjustment factor, the normalized value, and the differential coefficient of the PID control; and use the product as the differential coefficient of the improved PID control.

[0032] The present invention has at least the following beneficial effects:

[0033] The present invention addresses the problems of unstable etching and prone to errors caused by the heterogeneity of the ceramic base and the chain changes in the process environment during the etching step of ceramic-based thin-film circuit preparation. The present invention improves the PID control algorithm. First, a microscopic multi-scale local analysis is performed to achieve high etching precision. Then, a partitioned FFT analysis is performed to analyze the heterogeneity of the electroplated film surface during etching. The heterogeneity reflected by the data is determined based on the changes in its period and the differences in the width of the frequency. Finally, a window analysis is performed on the data to determine the final synergy factor based on the chain changes between different data. This factor enables the PID algorithm to calculate the heterogeneity differences reflected by different data while reflecting the chain changes caused by the synergistic effect of the data. The PID algorithm is then improved based on the results to provide real-time feedback on the oscillation and instability of the ion beam energy caused by the heterogeneity of the electroplated film surface and the chain changes in parameters, based on the synergistic effect and heterogeneity. This enhances the robustness and practicality of the algorithm, improves the accuracy of ion beam energy control, and ensures the integrated quality of new ceramic-based thin-film circuit products. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0035] Figure 1 A flow chart of a method for integrating a ceramic-based thin film circuit product provided by the present invention;

[0036] Figure 2 Flowchart for obtaining the differential coefficient of the improved PID control. DETAILED DESCRIPTION

[0037] To further illustrate the technical means and effectiveness of the present invention in achieving its intended objectives, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effectiveness of a ceramic-based thin-film circuit product integration method proposed in accordance with the present invention. In the following description, references to different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0038] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0039] The specific scheme of the integration method of a ceramic-based thin film circuit product provided by the present invention is described in detail below with reference to the accompanying drawings.

[0040] An embodiment of the present invention provides a method for integrating a ceramic-based thin-film circuit product.

[0041] Specifically, a method for integrating a ceramic-based thin film circuit product is provided as follows. Figure 1 , the method comprises the following steps:

[0042] Step S001: Acquire various types of etching data, and use the various types of etching data as rows of an etching matrix.

[0043] This embodiment uses a semi-additive process with adhesive electroplating to replace the traditional subtractive process, and at the same time introduces a temporary protective layer during the etching stage, which not only significantly improves the accuracy of the circuit, but also effectively protects the conductor layer during the etching of the seed layer and the bottom layer. Afterwards, advanced process methods such as layered integration, electron beam evaporation, magnetron sputtering, chemical plating and electroplating are used to form a nickel-chromium-gold composite film. Subsequently, through precise process steps such as photolithography and selective electroplating, a conductive band film or a stop band film is formed according to product requirements. Then, based on the circuit pattern to be etched, a photolithography mask template is made for etching path planning, and the electroplated film is etched using a dry physical ion sputtering method.

[0044] After etching begins, etching data is collected, including but not limited to material change data, surface elevation difference data, temperature data, vacuum data, ion beam movement distance data, material change data at the corresponding position of the ion beam, surface roughness data at the corresponding position of the ion beam, ion beam energy data, ion beam current density data, etc.

[0045] In this embodiment, uniform data collection is performed at time intervals t, and Z-score normalization is performed on the collected various types of etching data. Z-score is a well-known technology in the art and will not be described in detail here. The etching matrix K is obtained, and the expression is:

[0046] ,

[0047] Where, is the data obtained when the m-th type of data is collected for the n-th time. In this embodiment, the value of m is 9.

[0048] It should be noted that the acquisition cycle is from the starting end to the branch end, the branch end is the node where at least two different paths appear on the etching path, and the next etching position after passing the node is the starting end, and the data of the next cycle of the path is collected, that is, within a collection cycle, the path passed by the ion beam has only one line segment, and no branch point appears.

[0049] At this point, the etching matrix is ​​obtained.

[0050] Step S002: Detrended fluctuation analysis is used to obtain the fractal dimension sequence of each row of data in the etching matrix; the element distribution characteristics of the fractal dimension sequence of each row of data are analyzed, and combined with the mutation point detection algorithm, the data trend within the local range of the mutation point data is analyzed to obtain the anomaly index of each row of data; the frequency domain signal of each row of data is obtained, and the stability parameter of each row of data is determined based on the change of the frequency distribution range and amplitude in the frequency domain signal of each row of data, and the local heterogeneity evaluation parameter of each row of data is obtained in combination with the said anomaly index.

[0051] Etching ceramic thin-film integrated circuits involves complex physical and chemical changes, which can affect circuit performance and surface quality. For example, subtle local variations in the surface quality of the electroplated layer during the etching process can lead to uneven erosion, affecting circuit performance. Temperature fluctuations during the etching process can also cause thermal effects in the circuit, leading to thermal expansion or contraction of the circuit material. Due to the precision of etching the electroplated layer, data fluctuations caused by these complex changes are short-lived and difficult to capture.

[0052] This embodiment uses the i-th row of data in the etching matrix as an example for analysis. The i-th row of data is used as input, and multifractal detrended fluctuation analysis (MFDFA) is used. The output is a fractal dimension curve. The fractal dimension curve is a curve that shows how the fractal dimension changes with the change of scale. The characteristics of this curve can be used to understand the multi-scale fractal structure of data, thereby more deeply analyzing the complexity of data under microscopic conditions. The fractal dimension curve of the i-th row of data is analyzed, and M data points are uniformly sampled from the fractal dimension curve to form a fractal dimension sequence. In this embodiment, M is 100; the Pettitt mutation point detection algorithm is used to mark the mutation points in the fractal dimension sequence, and the number of mutation points in the fractal dimension curve is saved as A. With the a-th mutation point as the center, the set including the a-th mutation point, the h data points before the a-th mutation point, and the h data points after the a-th mutation point is taken as the local data sequence of the a-th mutation point. In this embodiment, h is ,in, is the number of scales of the MFDFA algorithm, is a rounding-up function. It should be noted that if there are less than h data points before or after a mutation point, all the data points before or after it are taken as part of the local data sequence of the mutation point. A linear fit based on the gradient descent method is performed on the local data sequence of the a-th mutation point, and the slope of the fitted line is finally obtained to construct the anomaly index of the i-th row of data.

[0053] First, the local data eigenvalue of each mutation point is calculated based on the degree of change in the elements in the local data sequence at each mutation point and the data increase or decrease trend. Specifically, for each row of data in the fractal dimension sequence, the slope of the straight line obtained by fitting the local data sequence at each mutation point is forward-fused with the element mean of the first-order difference sequence of the local data sequence at the corresponding mutation point to obtain the local data eigenvalue of each mutation point in the fractal dimension sequence of each row of data. In this embodiment, the forward fusion of multiple variables uses a multiplication calculation method.

[0054] Afterwards, the abnormality index of each row of data is determined by combining the local data eigenvalues ​​of all mutation points in each row of data, the distribution characteristics of the fractal dimension sequence of each row of data, and the number of mutation points: the discrete degree of the elements in the fractal dimension sequence of each row of data is obtained; the mean of the local data eigenvalues ​​of all mutation points corresponding to each row of data is calculated; and the result of forward fusion of the discrete degree and the mean of the local data eigenvalue corresponding to each row of data is used as the abnormality index of each row of data.

[0055] In this embodiment, the degree of discreteness of sequence elements is calculated using a variance method; and the forward fusion results between multiple variables are calculated using a multiplication method.

[0056] It should be understood that the anomaly index represents the variation characteristics and trends of each row in a short-term local state. First, the greater the degree of dispersion of the fractal dimension sequence elements in each row of data, the more significant the relative difference fluctuations in each row of data. That is, during etching, the data types in that row exhibit significant fluctuations. In this case, the greater the constraint on local variation, the larger the anomaly index. Among them, the local data characteristics reflect the local variation characteristics of each row of data. By calculating the average of the slope of the fitted line of the local data sequence at the a-th mutation point and the product of the first-order difference, the local fluctuation of the i-th row of data is further analyzed as a whole. The slope of the fitted line composed of the sequence of local range data at the mutation point reflects the trend of the fractal dimension. A larger value indicates an increasing trend of the fractal dimension within the mutation range. Combined with the first-order difference sequence of the data within the window, it reflects the variation of the fractal dimension at different scales. A larger product indicates more dramatic fluctuations in the local growth condition. That is, during etching, the properties of the conductive film undergo drastic changes under subtle ion beam movement, and the larger the anomaly index.

[0057] The fractal dimension of the i-th row of data is obtained using the MFDFA algorithm, and the fluctuations in the local state of the etching data at different scales are reflected through the detrended multi-scale analysis method. However, the etching process can cause changes in the quality and smoothness of the circuit board surface, resulting in oscillations, ripples, or other periodic changes on the surface. Surface structures of different scales, such as microscopic textures or macroscopic fluctuations, are easily formed during the etching process, leading to a dependence of the etching efficiency on the microscopic surface morphology changes. When the surface morphology changes are relatively subtle or even unchanged, the etching efficiency is high. When the surface morphology changes significantly, the etching efficiency is also affected. However, due to the high precision of etching, the collected data may contain non-stationary or transient signals, which makes it more difficult to observe its periodicity from the collected data. However, in the frequency domain, these transient signals may be more clearly presented as frequency components, making them easier to analyze and understand. The peak and trough partitioning can be used to highlight or isolate specific frequency components in the frequency domain signal, which can more intuitively and specifically display the signal characteristics.

[0058] Use Fast Fourier Transform (FFT) on the i-th row data to convert it from time domain signal to frequency domain signal, record the number and size of peaks and troughs in the frequency domain signal, and assume that the frequency domain signal of the i-th row data contains peaks, and the size of the x-th peak is , the corresponding frequency is , suppose the obtained frequency domain signal contains troughs, and the size of the y-th trough is , the corresponding frequency is , obtain the local heterogeneity evaluation parameters of the i-th row data.

[0059] First, based on the changes in the frequency distribution range and amplitude in the frequency domain signal of the i-th row of data, the stability parameters for evaluating the changes in the frequency domain of the i-th row of data are determined: the absolute value of the amplitude difference between the peaks and troughs of the same position sequence is recorded as the first difference; the absolute value of the frequency difference between the peaks and troughs of the same position sequence is recorded as the second difference; the average value of the ratio between the first difference and the second difference under all positions is recorded as the stability parameter of each row of data in the frequency domain.

[0060] In this embodiment, the stability parameter of the i-th row data in the frequency domain is recorded as , its formula form is:

[0061] ; Where, the stability parameter of the i-th row data in the frequency domain, is the minimum function, 、 are the number of peaks and troughs of the frequency domain signal of the i-th row of data, 、 are the amplitude difference between the pth peak and trough, and the frequency difference between the pth peak and trough in the frequency domain signal of the i-th row of data.

[0062] Afterwards, the product of the anomaly index of each row of data and the stability parameter is used as the local heterogeneity evaluation parameter of each row of data.

[0063] It should be understood that the local heterogeneity evaluation parameter reflects the heterogeneity of the electroplated film during the etching process in the data. By partitioning the p-th peak and trough of the frequency domain signal, when When the value is larger, it means that in the etching process, there are obvious peaks and valleys in the frequency domain of the i-th row data, that is, the i-th row data has more obvious periodic changes in the p-th frequency region, and there is a stronger periodic structure. The stronger the periodic changes in the etching data, the higher the heterogeneity on the electroplating surface. The bigger the value, the smaller the value; , which represents the difference in specific frequencies corresponding to the p-th peak and trough in the frequency domain signal. In fact, this difference represents the width difference on the frequency axis. In frequency domain analysis, the width of a frequency region can reflect the bandwidth or frequency range of the signal in the frequency region. For the peaks and troughs in the frequency domain signal, the corresponding frequency difference represents the frequency distribution range of the signal in this frequency region. If it is larger, it means that in the p-th frequency region, the frequency difference between the peaks and troughs is larger, that is, the signal in this frequency region covers a wider frequency range, reflecting that the interval of the p-th period change is wider, that is, the occurrence of the p-th period change is more stable rather than drastic. The smaller.

[0064] S003: With each data in each row of the etching matrix as the center, preset the initial length of the window, and evenly increase it towards both ends with a preset step size. According to the consistency of the data changes in the window after the window corresponding to each data in any two rows of the etching matrix is ​​increased, the local consistency parameter between the arbitrary two rows of data is determined; the synergy factor of the etching matrix is ​​determined by comprehensively considering the distribution similarity, local consistency parameters and local heterogeneity evaluation parameters between different rows of data.

[0065] In etching data, due to the microscopic conditions of ion etching, various parameters are extremely sensitive to changes in the environment or other parameters. Ion etching is a complex physical process, and there are correlations between various parameters. One change may cause coordinated changes in other parameters. Suppose the energy of the ion beam is adjusted and increased during the etching process. Increasing the energy of the ion beam will cause more energy to be converted into heat, which will increase the temperature of the etching environment. High-energy ions bombarding the material surface will cause evaporation of the substance, increase the number of gas molecules in the chamber, and affect the vacuum degree. Adjusting the energy of the ion beam will affect the performance of the ion source, resulting in changes in the ion beam current density. High-energy ions will cause changes in the microstructure of the material surface, resulting in an increase in surface roughness, and so on.

[0066] First, calculate the local heterogeneity evaluation parameters of all rows and normalize their Norm norm to form a heterogeneity index sequence. For the i-th row data and the j-th row data in the etching matrix, calculate their Pearson correlation coefficient , then build a gradually larger window with the qth data in the i-th row as the center, and the initial scale of the window is In this embodiment, I takes a value of 3, the window increases evenly toward both ends, the step size of each end is 2, the maximum number of increases is U, and U takes a value of 3. The data in the window is analyzed to construct the synergy factor of the etching matrix.

[0067] First, the local consistency parameter between two rows of data is determined based on the consistency of the data changes in the window after the window taken by each data in different rows of data is increased: the ratio between the window length increased each time and the maximum window increase length is calculated; the negative correlation mapping result of the difference between the extreme values ​​of the data in the window after each increase of the window centered on each data in any two rows of data is obtained; the ratio is used as the weight of the negative correlation mapping result corresponding to each increase of the any two rows of data, and the sum is obtained to obtain the local consistency parameter between the any two rows of data.

[0068] In this embodiment, the local consistency parameter between the i-th row data and the j-th row data is recorded as , its formula is Where, is the local consistency parameter between the i-th row data and the j-th row data, n is the number of columns of the etching matrix, u is the number of times the window is increased, and U is the maximum number of times the window is increased. To increase the window weight at u times, The size of is the ratio of the window length after increasing u times to the window length after increasing U times; 、 It is the range of the data in the window when the window centered on the qth data in the i-th row and the j-th row is increased u times. exp( ) represents an exponential function with a natural constant as the base.

[0069] Afterwards, the local heterogeneity evaluation parameters of each row of data and the distribution similarity between different rows of data are combined to determine the synergy factor of the etching matrix: the similarity measure between any two rows of data is obtained, and after forward fusion with the local consistency parameter, it is divided by the absolute value of the difference between the local heterogeneity evaluation parameters between the any two rows of data to obtain the synergy index between the any two rows of data; the average of the synergy indexes obtained from all pairwise combinations of row data in the etching matrix is ​​used as the synergy factor of the etching matrix.

[0070] In this embodiment, the synergy factor of the etching matrix is ​​recorded as , its formula form is: Where, is the synergy factor of the etching matrix, m represents the number of rows of the etching matrix, is the similarity measure between the data in the i-th row and the data in the j-th row. In this embodiment, the Pearson correlation coefficient is used for calculation; is the absolute value of the difference between the local heterogeneity evaluation parameters of the data in the i-th row and the j-th row, is the local consistency parameter between the i-th row data and the j-th row data, It should be noted that, in order to prevent the denominator from returning to zero, the adjustment parameter is set to 0.01 in this embodiment.

[0071] It should be understood that the synergy factor of the etching matrix reflects the synergy and heterogeneity between the etching data. By analyzing the fluctuation difference between the windows centered on data q between the i-th row and the j-th row of data when the window is increased u times, the greater the fluctuation difference, the smaller the similarity between the two rows of data from a local perspective. The smaller it is; then the window size is calculated based on the proportion of the window size at its maximum window size. The weight is the number of increases and the larger the window, the greater the weight, because the larger the window, the more local features it contains, and the greater the weight is given to it. After multiplying the two, the number of window increases and the center of the window are summed to obtain the comprehensive local consistency parameter between the i-th row data and the j-th row data. , the larger its value, the greater the local synergy between the data in row i and row j. The bigger.

[0072] At the same time, combined with the absolute value of the difference in local heterogeneity evaluation parameters between the i-th row data and the j-th row data, the local heterogeneity evaluation parameters of the two rows of data represent the heterogeneity between the i-th row data and the j-th row data. The larger the absolute value of the difference, the more different the electroplating surface heterogeneity reflected by the i-th row data and the j-th row data, and the worse the synergy between the two rows. The smaller the value, the smaller the Pearson correlation coefficient represents the linear correlation between two rows of data. The bigger. Overall By constraining local synergy through overall correlation, the heterogeneity differences reflected between different rows are further introduced to represent the comprehensive synergy of different data in the etching process and their degree of mutual correlation, so that the synergistic impact of etching can be analyzed from all data.

[0073] S004: Improve the differential coefficient of PID control according to the synergy factor of the etching matrix; etch the surface of the electroplated film in combination with the improved differential coefficient of PID control, and complete the wire bonding of the semiconductor chip through thin film hybrid integration.

[0074] The deviation between the actual measured ion beam energy and the rated etching ion beam energy is used as input, and a PID controller is used to control the ion beam energy emitted by the ion beam source. However, due to the accuracy requirements of the dry physical ion sputtering method, subtle material changes on the surface of the electroplated film and the synergistic effects between various data can have a significant impact on the etching effect when controlling the ion beam. Therefore, the differential coefficient in the PID control is improved:

[0075] ,

[0076] Where, is the improved differential coefficient, is the synergy factor of the etching matrix, kd is the differential coefficient of PID control, is the adjustment factor. In this embodiment, the value is 2. The implementer can adjust it according to the actual accuracy requirements. is the sigmoid function. The process of obtaining the differential coefficient of the improved PID control is as follows: Figure 2 shown.

[0077] The improved differential coefficient is brought into the PID algorithm for use, and the collected data is analyzed in real time, so that a synergistic factor can be calculated in real time during etching to adjust the differential coefficient, so that it can provide real-time feedback based on the ion beam energy oscillation or instability caused by the surface heterogeneity of the electroplated film or the synergistic effect generated during electroplating. That is, the feedback of the error change rate of the differential part is amplified (or reduced) by the synergistic factor, thereby achieving precise control of the ion beam energy.

[0078] The ion beam energy is controlled by a PID algorithm to etch the surface of the electroplated film. The etched substrate is then attached to the tube base using eutectic welding or slurry bonding to connect the different components. Finally, through thin film hybrid integration, one or more semiconductor chips or other chip components are integrated onto the convex ceramic substrate, and wire bonding of the semiconductor chips is completed, achieving a high level of integration. This completes the integration of new ceramic-based thin-film circuit products.

[0079] In summary, the embodiments of the present invention improve the PID control algorithm to address the problems of unstable etching and prone to errors caused by the heterogeneity of the ceramic base and the chain changes in the process environment during the etching step of ceramic-based thin-film circuit preparation. First, a microscopic multi-scale local analysis is performed to achieve high etching precision. Then, a partitioned FFT analysis is performed to analyze the heterogeneity of the electroplated film surface during etching. The heterogeneity reflected by the data is determined based on the change in its period and the difference in the width of the frequency. Finally, a window analysis is performed on the data to obtain the final synergy factor based on the chain changes between different data. This factor can not only reflect the chain changes caused by the synergistic effect of the data, but also calculate the heterogeneity differences reflected by different data. Then, based on the results, the PID algorithm is improved to provide real-time feedback on the oscillation and instability of the ion beam energy caused by the heterogeneity of the electroplated film surface and the chain changes in parameters according to the synergistic effect and heterogeneity. This enhances the robustness and practicality of the algorithm, improves the control accuracy of the ion beam energy, and ensures the integrated quality of new ceramic-based thin-film circuit products.

[0080] It should be noted that the order in which the embodiments of the present invention are described above is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0081] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

[0082] The above-described embodiments are only used to illustrate the technical solutions of the present application, and not to limit them. Modifications to the technical solutions described in the aforementioned embodiments, or equivalent replacements of some of the technical features therein, do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A method for integrating ceramic-based thin film circuit products, characterized in that: The method comprises the following steps: Obtaining various types of etching data, and using the various types of etching data as rows of an etching matrix; Detrended fluctuation analysis is used to obtain a fractal dimension sequence for each row of data in the etching matrix; the element distribution characteristics of the fractal dimension sequence for each row of data are analyzed, and combined with a mutation point detection algorithm, the data trend within the local range of the mutation point data is analyzed to obtain an anomaly index for each row of data; the frequency domain signal for each row of data is obtained, and based on the frequency distribution range and amplitude changes in the frequency domain signal for each row of data, the stability parameter for each row of data is determined, and combined with the anomaly index, a local heterogeneity assessment parameter for each row of data is obtained; Taking each data in each row of the etching matrix as the center, the initial length of the window is preset, and the window is evenly increased towards both ends with a preset step size. The local consistency parameter between any two rows of data in the etching matrix is ​​determined based on the consistency of the data changes in the window after the window corresponding to each data is increased; the synergy factor of the etching matrix is ​​determined by comprehensively considering the distribution similarity between different rows of data, the local consistency parameter, and the local heterogeneity evaluation parameter; The differential coefficient of PID control is improved according to the synergistic factor of the etching matrix; the surface of the electroplated film is etched in combination with the improved differential coefficient of PID control, and the wire bonding of the semiconductor chip is completed through thin film hybrid integration.

2. The method for integrating a ceramic-based thin film circuit product according to claim 1, wherein: The method of obtaining the fractal dimension sequence of each row of data in the etching matrix by using detrended fluctuation analysis includes: Each row of the etching matrix is ​​used as the input of the multifractal detrended fluctuation analysis, and the fractal dimension curve of each row of data is output; the fractal dimension curve is uniformly sampled to obtain the fractal dimension sequence.

3. The method for integrating a ceramic-based thin film circuit product according to claim 1, wherein: The abnormality index of each row of data is obtained as follows: Preset the local data sequence of each mutation point in the fractal dimension sequence of each row of data; Calculate the local data eigenvalue of each mutation point based on the degree of element change in the local data sequence of each mutation point and the data increase or decrease trend; Obtain the degree of discreteness of elements in the fractal dimension sequence of each row of data; calculate the mean of the local data eigenvalues ​​of all mutation points corresponding to each row of data; and forwardly fuse the degree of discreteness and the mean of the local data eigenvalues ​​corresponding to each row of data as the anomaly index of each row of data.

4. The method for integrating a ceramic-based thin film circuit product according to claim 3, wherein: The calculation of the local data characteristic value of each mutation point is specifically as follows: For the fractal dimension sequence of each row of data, the slope of the straight line obtained by fitting the local data sequence of each mutation point is forward fused with the element mean of the first-order difference sequence of the local data sequence corresponding to the mutation point to obtain the local data eigenvalue of each mutation point in the fractal dimension sequence of each row of data.

5. The method for integrating a ceramic-based thin film circuit product according to claim 1, wherein: The stability parameter of each row of data is determined as follows: Get all the peaks and troughs of the frequency domain signal for each row of data; The absolute value of the amplitude difference between the peaks and troughs of the same sequence is recorded as the first difference; the absolute value of the frequency difference between the peaks and troughs of the same sequence is recorded as the second difference; the average value of the ratio between the first difference and the second difference under all sequences is recorded as the stability parameter of each row of data in the frequency domain.

6. The method for integrating a ceramic-based thin film circuit product according to claim 3, wherein: The local heterogeneity evaluation parameter of each row of data is specifically the product of the local data eigenvalue and the stability parameter.

7. The method for integrating a ceramic-based thin film circuit product according to claim 1, wherein: Determining the local consistency parameter between any two rows of data includes: Calculate the ratio between the window length increased each time and the maximum window length increased; obtain the negative correlation mapping result of the difference between the extreme values ​​of the data in the window after each increase of the window centered on each data in any two rows of data; use the ratio as the weight of the negative correlation mapping result corresponding to each increase of the any two rows of data, and sum them to obtain the local consistency parameter between the any two rows of data.

8. The method for integrating a ceramic-based thin film circuit product according to claim 1, wherein: The specific process of determining the synergy factor of the etching matrix is ​​as follows: Obtaining the similarity measure and the difference of the local heterogeneity evaluation parameter between any two rows of data, and combining the local consistency parameter to obtain the synergy index between the any two rows of data; The average of the synergy indexes obtained from all pairwise combinations of row data in the etching matrix is ​​taken as the synergy factor of the etching matrix.

9. The method for integrating a ceramic-based thin film circuit product according to claim 8, wherein: The synergy index between the two rows of data is obtained as follows: For any two rows of data in the etching matrix, the similarity measure and the local consistency parameter are forward fused and then divided by the absolute value of the difference of the local heterogeneity evaluation parameter between the any two rows of data to obtain the synergy index between the any two rows of data.

10. The method for integrating a ceramic-based thin film circuit product according to claim 1, wherein: The improvement of the differential coefficient of PID control according to the synergy factor of the etching matrix includes: Preset the adjustment factor; obtain the normalized value of the synergy factor of the etching matrix; calculate the product of the adjustment factor, the normalized value, and the differential coefficient of the PID control; and use the product as the differential coefficient of the improved PID control.

Images