White light interferometric focus detection system and demodulation method

Abstract

The present disclosure provides a white light interferometric focusing system and a demodulation method, the system comprising: a white light source; a template with a pattern prepared thereon; a substrate placed on a substrate table; a spectrometer for collecting a white light interference signal between the template and the substrate through the pattern; and a control system for calculating a gap value between the template and the substrate by using a Fourier transform demodulation method, a cross-correlation calculation demodulation method or a model transfer demodulation method according to the interference signal. The present disclosure uses the focusing system to adopt different demodulation methods according to different shapes of the reflection spectrum in a larger distance measurement range, and the gap demodulation method eliminates the influence of the patterned surface on the focusing spectrum and breaks through the limitation of the template pattern layout caused by the focusing window.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the field of optical detection, in particular to a white light interference focusing system and a demodulation method. BACKGROUND

[0002] Currently, the methods for focusing mainly include: capacitive sensor, surface plasmon resonance (SPR) sensing method, interference space phase imaging (ISPI) detection method and white light interference detection method. The capacitive sensor is not suitable for measuring the exposure process because it needs to embed a capacitor in the template substrate and is affected by thermal drift during the exposure process; the SPR sensing method needs to make a multi-layer film structure on the template, which is difficult to process and difficult to implement; the ISPI detection method has a detection range of more than 1 μm, which cannot achieve nanometer precision detection.

[0003] White light interference detection has the advantages of wide measurement range, high precision and absolute measurement, and is widely used in distance measurement of the exposure process. Traditional white light interference measurement needs to allocate a focusing window in the template pattern area, which seriously affects the template pattern layout and increases the difficulty of circuit design. SUMMARY

[0004] (I) Technical problems to be solved

[0005] To solve the above problems, the present disclosure provides a white light interference focusing system and a demodulation method to solve the technical problems of limited detection range of traditional methods and easy to affect the template pattern layout.

[0006] (II) Technical solutions

[0007] In one aspect, the present disclosure provides a white light interference focusing system, comprising: a white light source; a template having a pattern prepared thereon; a substrate placed on a wafer holder; a spectrometer for collecting a white light interference signal between the template and the substrate through the pattern; and a control system for calculating a gap value between the template and the substrate using a Fourier transform demodulation method, a cross-correlation calculation demodulation method or a model transfer demodulation method according to the interference signal.

[0008] Another aspect of the present disclosure provides a method for gap demodulation according to the aforementioned white light interference focus detection system, comprising: S1, collecting a white light interference signal between a template and a substrate to obtain a sample spectrum; S2, performing peak detection on the sample spectrum and performing spectrum preprocessing according to the result of the peak detection; S3, calculating a reflection spectrum according to a reference spectrum collected in advance and the sample spectrum preprocessed in S2; S4, determining a demodulation method according to the shape of the reflection spectrum to calculate a gap value between the template and the substrate; wherein the demodulation method comprises one of a Fourier transform demodulation method, a cross-correlation calculation demodulation method and a model transfer demodulation method.

[0009] Further, the determination of the demodulation method in S4 comprises: judging a gap estimation value between the template and the substrate according to the shape of the reflection spectrum; when the gap estimation value is greater than a first threshold, adopting the Fourier transform demodulation method to calculate the gap value; when the gap estimation value is less than or equal to the first threshold and greater than or equal to a second threshold, adopting the cross-correlation calculation demodulation method to calculate the gap value; when the gap estimation value is less than the second threshold, adopting the model transfer demodulation method to calculate the gap value.

[0010] Further, the first threshold ranges from 20 μm to 40 μm; and the second threshold ranges from 1 μm to 5 μm.

[0011] Further, the calculation of the gap value by the Fourier transform demodulation method comprises: S411, decomposing the reflection spectrum obtained in S3 into a spectral profile and an interference signal spectrum; S412, performing Fourier transform on the interference signal spectrum to obtain a frequency domain signal; S413, performing band-pass filtering according to the frequency domain signal; S414, performing inverse Fourier transform on the filtered signal to obtain an analytical signal; S415, performing complex logarithm operation on the analytical signal to obtain the phase of the sample spectrum; S416, selecting the phase at any two wavelengths, calculating the optical path difference according to the phase difference of the phase at the two wavelengths to obtain the gap value corresponding to the sample spectrum.

[0012] Further, the calculation of the gap value by the cross-correlation calculation demodulation method comprises: S421, initializing a film layer model and illumination parameters, determining a gap range and a gap interval; S422, performing simulation according to the film layer model by using rigorous coupled wave analysis to calculate simulation reflection spectra under different gaps; S423, performing normalization processing on the reflection spectrum obtained in S3; S424, performing cross-correlation operation on the simulation reflection spectrum in S422 and the normalized reflection spectrum in S3; S425, determining the optical path difference with the largest correlation coefficient to obtain the gap value corresponding to the sample spectrum.

[0013] Further, the calculation of the gap value by the model transfer demodulation method comprises a model transfer step and a cross-correlation demodulation step.

[0014] Further, the model transfer step comprises: S431, initializing particle swarm optimization parameters and neural network parameters, the neural network parameters comprising the number of hidden layers, the number of hidden layer nodes, the activation function, the randomly generated weight and the offset vector; S432, calling the activation function, and outputting the reflectance spectrum calculated by the neural network; S433, calculating the fitness of the particle swarm, and updating the particle swarm if the fitness is less than a set threshold; S434, updating until the fitness is not less than the set threshold, and outputting the optimized neural network parameters; and S435, performing spectrum model transfer by using the neural network parameters obtained in S435, and outputting the reflectance spectrum. Further, S4 comprises: adjusting the adhesion between the polymer film and the transmission metal film layer by changing the viscosity and the curing degree of the high molecular polymer in the high molecular polymer solution, so that the transmission metal film layer is completely peeled off from the polymer film, and the photosensitive film layer and the reflection metal film layer are not damaged.

[0015] Further, the cross-correlation demodulation step comprises: S436, performing normalization processing on the reflectance spectrum obtained in S3; S437, performing cross-correlation operation on the reflectance spectrum obtained in S435 and the reflectance spectrum subjected to the normalization processing in S436; and S438, determining the optical path difference with the largest correlation coefficient to obtain the gap value corresponding to the sample spectrum.

[0016] Another aspect of the present disclosure provides a device for gap demodulation by using a white light interference focusing system, comprising: an acquisition module configured to acquire a white light interference signal between a template and a substrate to obtain a sample spectrum; a detection module configured to perform peak value detection on the sample spectrum and perform spectrum preprocessing according to the result of the peak value detection; a calculation module configured to calculate a reflectance spectrum according to a reference spectrum acquired in advance and the sample spectrum subjected to the preprocessing in S2; and a determination module configured to determine a demodulation method for calculating a gap value between the template and the substrate according to the shape of the reflectance spectrum, wherein the demodulation method comprises one of a Fourier transform demodulation method, a cross-correlation calculation demodulation method and a model transfer demodulation method.

[0017] (III) Beneficial Effects

[0018] The white light interference focusing system and the gap demodulation method for penetrating a patterned surface according to the present disclosure utilize the white light interference focusing technology, and in a larger distance measurement range, according to the different shapes of the reflectance spectrum, one of the Fourier transform demodulation method, the cross-correlation calculation demodulation method and the model transfer demodulation method is used to calculate the gap value between the template and the substrate. The gap demodulation method guarantees the measurement accuracy and improves the measurement efficiency. Meanwhile, the light emitted by the light source penetrates the pattern on the template, is reflected by the substrate and then returns to the spectrometer for image acquisition, thereby eliminating the influence of the patterned surface on the focusing spectrum and breaking through the limitation of the template pattern layout caused by the focusing window. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 A structural schematic diagram of a white light interference focus detection system according to an embodiment of the present disclosure is shown schematically;

[0020] Figure 2 A cross-sectional schematic diagram of a white light interference focus detection system according to an embodiment of the present disclosure is shown schematically;

[0021] Figure 3 A flowchart of a gap demodulation method according to a white light interference focus detection system according to an embodiment of the present disclosure is shown schematically;

[0022] Figure 4 A flowchart of a gap demodulation method according to a white light interference focus detection system according to an embodiment of the present disclosure is shown schematically;

[0023] Figure 5 A reflection spectrum diagram of a gap of 40 μm according to an embodiment of the present disclosure is shown schematically;

[0024] Figure 6 A flowchart of a Fourier transform demodulation method according to an embodiment of the present disclosure is shown schematically;

[0025] Figure 7 A reflection spectrum diagram according to an embodiment of the present disclosure is shown schematically;

[0026] Figure 8 A reflection spectrum diagram according to an embodiment of the present disclosure is shown schematically;

[0027] Figure 9 A frequency spectrum diagram according to an embodiment of the present disclosure is shown schematically;

[0028] Figure 10 A reflection spectrum diagram of a gap of 10 μm according to an embodiment of the present disclosure is shown schematically;

[0029] Figure 11 A flowchart of a cross-correlation calculation demodulation method according to an embodiment of the present disclosure is shown schematically;

[0030] Figure 12 A reflection spectrum diagram of a gap of 0.01 μm according to an embodiment of the present disclosure is shown schematically;

[0031] Figure 13 A flowchart of a model transfer demodulation method according to an embodiment of the present disclosure is shown schematically;

[0032] Explanation of reference numerals:

[0033] 1, white light source; 2, template; 3, pattern; 4, substrate; 5, wafer table; 6, workpiece table; 7, spectrometer; 8, control system. DETAILED DESCRIPTION

[0034] In order to make the purpose, technical solutions and advantages of the present disclosure clearer, the present disclosure is further described in detail below with reference to specific embodiments and drawings.

[0035] The terms used herein are merely used to describe specific embodiments, and are not intended to limit the present disclosure. The terms "include", "contain" and the like used herein indicate the existence of the described features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0036] It should be noted that if the directionality indication is involved in the embodiments of the present disclosure, the directionality indication is only used to explain the relative position relationship, motion condition and the like between the components in a certain specific posture, and if the specific posture changes, the directionality indication also changes accordingly.

[0037] The present disclosure provides a white light interference focusing system, please see Figure 1 , comprising: a white light source 1; a template 2 having a pattern 3 prepared thereon; a substrate 4 placed on a wafer holder 5; a spectrometer 7 for collecting a white light interference signal between the template 2 and the substrate 4 through the pattern 3; and a control system 8 for calculating a gap value between the template 2 and the substrate 4 by using a Fourier transform demodulation method, a cross-correlation calculation demodulation method or a model transfer demodulation method according to the interference signal.

[0038] Figure 2 For the white light interference focusing system and its corresponding sectional view, the white light generated by the white light source 1 is used as the incident light source, passes through the template 2, and the incident light passes through the template 2 and the pattern 3 to reach the substrate 4, and then is reflected through the template 2 and is collected by the spectrometer 7. The wafer holder 5 is fixed on the workpiece table 6, and the control system 8 adjusts the posture and position of the substrate 4 by controlling the workpiece table 6. The interference signal between the template 2 and the substrate 4 is received by the spectrometer 7, and the signal received by the spectrometer 7 is analyzed by the determination module in the control system 8 to obtain the gap value between the template 2 and the substrate 4. The present disclosure includes a spectrum collection hardware system composed of a white light source 1, an optical fiber, a lens and a spectrometer, and a software system composed of a Fourier transform demodulation method, a cross-correlation calculation demodulation method and a model transfer demodulation method.

[0039] According to the different shapes of the reflected spectrum, one of the Fourier transform demodulation method, the cross-correlation calculation demodulation method and the model transfer demodulation method is used to calculate the gap value between the template 2 and the substrate 4. This gap demodulation method ensures the accuracy of the measurement, improves the efficiency of the measurement, and at the same time eliminates the influence of the pattern layer on the template on the focusing spectrum, and breaks through the limitation of the template pattern layout caused by the focusing window.

[0040] The present disclosure also provides a method for gap demodulation according to the aforementioned white light interference focus detection system, please refer to Figure 3 , comprising: S1, collecting the white light interference signal between the template 2 and the substrate 4 to obtain a sample spectrum; S2, performing peak detection on the sample spectrum, and performing spectrum preprocessing according to the result of the peak detection; S3, calculating a reflection spectrum according to the pre-acquired reference spectrum and the sample spectrum after the preprocessing in S2; S4, determining a demodulation method according to the shape of the reflection spectrum to calculate the gap value between the template 2 and the substrate 4; wherein the demodulation method comprises one of a Fourier transform demodulation method, a cross-correlation calculation demodulation method and a model transfer demodulation method.

[0041] First, the spectral acquisition is performed, the wavelength response range of the spectrometer 7 is for example 300nm to 900nm, the sample spectrum is obtained, then the peak detection is performed, the spectrum preprocessing is performed according to the result of the peak detection, including noise elimination and baseline elimination (the purpose of the baseline elimination is to correct the offset and the rotation direction of the sample spectrum in the Y direction), then the reflection spectrum is obtained through the reference spectrum and the sample spectrum after the preprocessing. Finally, according to the different shapes of the reflection spectrum, different demodulation methods are taken to process the spectrum gap demodulation.

[0042] It should be noted that the system initialization needs to be performed before step S1, and the system initialization includes background spectrum acquisition and reference spectrum acquisition. Before the template 2 and the substrate 4 are installed, the control system 8 controls the spectrometer 7 to acquire the background spectrum, which is saved in the control system 7 and used to eliminate the influence of environmental light and other stray light on the spectrum acquired during the test; then the template 2 and the substrate 4 are installed, the white light source 1 is turned on through the control system 8, the workpiece table 6 is controlled to move upward to a certain distance between the template 2 and the substrate 4, and the spectrometer 7 is controlled to acquire the reference light. In the embodiment, when the wavelength response range of the spectrometer 7 is 300nm to 900nm, the acquisition position of the reference light is about 250μm between the template 2 and the substrate 4, the spectrometer 7 is controlled to acquire the reference light, and the reference light is saved in the control system 8.

[0043] On the basis of the above embodiment, as shown in Figure 4 , the determination of the demodulation method in S4 comprises: judging the gap estimation value between the template 2 and the substrate 4 according to the shape of the reflection spectrum; when the gap estimation value is greater than a first threshold value, the Fourier transform demodulation method is used to calculate the gap value; when the gap estimation value is less than or equal to the first threshold value and greater than or equal to a second threshold value, the cross-correlation calculation demodulation method is used to calculate the gap value; when the gap estimation value is less than the second threshold value, the model transfer demodulation method is used to calculate the gap value.

[0044] When the gap estimation value is greater than the first threshold value, the collected spectrum is a dense curve, and the Fourier transform demodulation method is used to demodulate the speed, and if the cross-correlation calculation demodulation method is used, the spectrum peak may be dense, which may cause matching error. When the gap estimation value is less than or equal to the first threshold value and greater than the second threshold value, the cross-correlation calculation demodulation method has the advantages of high precision and controllable resolution. When the gap estimation value is less than or equal to the second threshold value, the pattern of the template 2 will affect the collected spectrum, and if the cross-correlation calculation demodulation method is still directly used, the demodulation error will occur, and the demodulated gap value is not the actual gap value, so the model transfer demodulation method is used for spectrum demodulation.

[0045] On the basis of the above embodiment, the range of the first threshold value is 20-40 μm, and the range of the second threshold value is 1-5 μm.

[0046] For example, the first threshold value is 20 μm, the second threshold value is 1 μm, when the gap is greater than 20 μm, the Fourier transform demodulation method is used for spectrum demodulation, when the gap is between 1 μm and 20 μm, the cross-correlation calculation demodulation method is used for spectrum demodulation, and when the gap is less than 1 μm, the model transfer demodulation method is used for spectrum demodulation. In fact, the specific values of the first threshold value and the second threshold value are related to the collected sample spectrum, and the first threshold value and the second threshold value can be selected according to the collected sample spectrum. In the case of the film layer model and the illumination parameter in the embodiment, when the gap is greater than 20 μm, the collected sample spectrum is a dense curve, when the gap is between 1 μm and 20 μm, the collected sample spectrum is a sparse curve, and when the gap is less than 1 μm, the collected sample spectrum is a monotonous curve as shown in Figure 12 .

[0047] On the basis of the above embodiment, the Fourier transform demodulation method for calculating the gap value includes: S411, decomposing the reflection spectrum obtained in S3 into a spectrum profile and an interference signal spectrum; S412, performing Fourier transform on the interference signal spectrum to obtain a frequency domain signal; S413, performing band-pass filtering according to the frequency domain signal; S414, performing inverse Fourier transform on the filtered signal to obtain an analytical signal; S415, performing complex logarithm operation on the analytical signal to obtain the phase of the sample spectrum; S416, selecting the phases at two arbitrary wavelengths, calculating the optical path difference according to the phase difference of the phases at the two wavelengths, and obtaining the gap value corresponding to the sample spectrum.

[0048] According to the above embodiment, when the gap is greater than 20 μm, the collected spectrum is a dense curve, as shown in Figure 5 , which is the reflection spectrum of the gap of 40 μm, at this time, the Fourier transform demodulation method is used for spectrum demodulation. Figure 6 is a flowchart of the Fourier transform demodulation method.

[0049] S411, in order to improve the contrast of the signal, the reflection spectrum (as shown in Figure 7 ) is decomposed into a spectral profile and an interference signal spectrum. After S411, the original reflection spectrum becomes the interference signal spectrum (as shown in Figure 8 ), the peak intensity is more uniform, and the spectral calculation is easier.

[0050] S412-S414, after the reflection spectrum is decomposed, the interference signal spectrum is subjected to Fourier transform, the interference signal spectrum is mapped to the frequency domain, band-pass filtering is performed in the frequency domain, the frequency between 0.01-0.2 (this range can be selected as needed) is taken, and the intensity of other frequencies is set to 0; after the band-pass filtering, the Fourier inverse transform is performed on the frequency spectrum (as shown in Figure 9 ), and the spectrum obtained after the Fourier inverse transform is still a function of wavelength.

[0051] S415, taking the imaginary part of the result of the Fourier inverse transform, the phase of the white light interference spectrum is obtained.

[0052] S416, selecting the phases at any two wavelengths, calculating the optical path difference according to the phase difference to obtain the gap value corresponding to the sample spectrum. The phase is a function of wavelength, and the calculation formula of the gap value L obtained from the phase is:

[0053]

[0054] wherein λ2 and λ1 are any two wavelengths in the Fourier inverse transform spectrum, respectively, and φ2 and φ1 are their corresponding phases.

[0055] On the basis of the above embodiment, the gap value is calculated by using the cross-correlation calculation demodulation method, including: S421, initializing the film layer model and the illumination parameters, determining the gap range and the gap interval; S422, using the rigorous coupled wave analysis to simulate according to the film layer model, and calculating the simulation reflection spectrum under different gaps; S423, normalizing the reflection spectrum obtained in S3; S424, performing cross-correlation operation on the simulation reflection spectrum in S422 and the reflection spectrum normalized in S3; S425, determining the optical path difference with the largest correlation coefficient to obtain the gap value corresponding to the sample spectrum.

[0056] According to the above embodiment, when the gap is between 1 μm and 20 μm, the cross-correlation calculation demodulation method is used for spectral demodulation, Figure 10 is the reflection spectrum under a 10 μm gap.

[0057] Figure 11 is a flowchart of the cross-correlation calculation demodulation method, and the specific process is as follows:

[0058] S421, first, the initialization film layer model and illumination parameters, film layer model includes film layer thickness, dielectric constant, etc., the illumination parameters at least include light intensity, incident angle. For example, the simulation film layer model from top to bottom are template (the upper is mask substrate, the lower is pattern layer), air gap and substrate, the film layer model is not unique, those skilled in the art can select the required model according to the actual situation; in this embodiment, the environmental thickness in the film layer model is infinite, the thickness of the pattern layer is 10nm, and the dielectric constant is a function of wavelength.

[0059] The gap range and gap interval required for initialization are selected as needed. Since the response range of the spectrometer 7 selected in the above embodiment is 300-900nm, the gap range is selected as 0-20μm (the sample spectrum under the gap range in the 300-900nm band is sparsely distributed in peaks). The gap range is the measurable gap range represented by the subsequent simulation of the reflection spectrum model library; the gap interval is the interval value between the adjacent two gaps (the distance between the template and the substrate) in the reflection spectrum model library, representing the resolution of the subsequent gap measurement. For example, if the reflection spectrum model library obtained by simulation has a large number of reflection spectra corresponding to different gaps d, and the gap interval is 1nm, then the gaps are d-2nm, d-1nm, d, d+1nm, d+2nm, …, and the reflection spectra corresponding to the gaps changing by 1nm can be obtained. If the gap interval is 2nm, then the gaps are d-4nm, d-2nm, d, d+2nm, d+4nm, …, and the reflection spectra corresponding to the gaps changing by 2nm can be obtained.

[0060] S422, using Rigorous Coupled Wave Analysis (RCWA) to simulate the above film layer model, calculating the simulated reflection spectrum, the wavelength range is from 300nm to 900nm, according to the above embodiment, the gap interval in the reflection path model library, that is, the wavelength resolution of the spectrum is 0.4nm, and the gap range is 0-20μm.

[0061] S423, performing normalization processing on the collected spectrum. The spectrum normalization processing formula is:

[0062]

[0063] is the normalized spectrum, which is a function of wavelength.

[0064] S424-S425, the simulated white light interference signal in S422 and the actual sampled white light interference signal are cross-correlated, and the optical path difference with the maximum correlation coefficient is the current measured gap value.

[0065] The cross-correlation calculation formula is:

[0066]

[0067] Where x is the current collected white light interferometric spectrum, y d is the simulated spectrum with a simulated optical path difference of d.

[0068] On the basis of the above embodiment, the gap value is calculated by using the model transfer demodulation method, which includes a model transfer step and a cross-correlation demodulation step.

[0069] According to the above embodiment, when the gap is less than 1 pm, the model transfer demodulation method is used for spectral demodulation, Figure 12 is the reflected spectrum under a gap of 0.01 pm. The model transfer demodulation method includes model transfer and cross-correlation demodulation. A machine learning framework based on heuristic optimization is used for model transfer. Taking a feedforward neural network as an example, the conversion from patterned interferometric spectrum to thin film interferometric spectrum is realized, thereby realizing high-precision white light interferometric spectrum gap detection.

[0070] On the basis of the above embodiment, the model transfer step includes: S431, initializing particle swarm optimization parameters and neural network parameters, the neural network parameters including the number of hidden layers, the number of hidden layer nodes, the activation function, the randomly generated weight and the offset vector; S432, calling the activation function, outputting the reflected spectrum calculated by the neural network; S433, calculating the fitness of the particle swarm, and updating the particle swarm if the fitness is less than a set threshold; S434, updating until the fitness is not less than the set threshold, and outputting the optimized neural network parameters; S435, performing spectrum model transfer by using the neural network parameters obtained in S435, and outputting the reflected spectrum.

[0071] Figure 13 is a flowchart of the model transfer demodulation method, and the specific process is as follows:

[0072] S431, first, initialize the particle swarm parameters and the neural network parameters. The particle swarm parameters include the variables to be optimized, the optimization range of the variables, the maximum number of iterations, the population size, and the variable update. The neural network parameters are initialized, including the number of hidden layers of the neural network, the number of hidden layer nodes, the activation function, the randomly generated weight w, and the offset vector b.

[0073] S432, calculate the hidden layer value, call the activation function, and output the reflected spectrum calculated by the neural network. The number of hidden layer nodes L of the feedforward neural network is determined according to the randomly generated weight vector w and the offset vector b, which are functions of the wavelength l. Thus, the value of the hidden layer is:

[0074] y L = w * x + b

[0075] According to the activation function h(x) = exp(-r||x-a||), where r and a are activation function parameters. The hidden layer y L is input, and the output of the feedforward neural network is:

[0076] f = βh(y L )

[0077] f = βh(w*x+b)

[0078] where β is the output weight.

[0079] The final training target of the feedforward neural network is:

[0080]

[0081] where C is a regularization parameter, and y is the actual spectrum.

[0082] S433, calculate the fitness g of the particle swarm, when the optimal particle fitness is less than the set threshold G, update the particle swarm, update each particle position. The population fitness calculation formula is as follows:

[0083] Fitness = ||f-f act || 2

[0084] where f is the current neural network output spectrum, and f act is the actual spectrum.

[0085] S434, update the neural network output, and the formula for updating the particle swarm is:

[0086] [w, b] next = [w, b] + ηv + c1r1([w, b] pbest - [w, b]) + c2r1([w, b] gbest - [w, b])

[0087] where [w, b] pbest is the best position of the current particle, and [w, b] gbest is the best position of all particles. The speed v of the current particle is randomly generated, where c1 and c2 are acceleration factors, r1 and r2 are uniform distributions of random numbers r ∈ [0, 1], and η is the inertia factor, which is updated as follows:

[0088]

[0089] where It is the maximum number of iterations, it is the current number of iterations, η ini is the initial inertia weight, and η endis the final inertia weight. Thus, the weight factor linearly decreases. And S433 and S434 are repeated.

[0090] S435, output the optimized neural network parameters [w, b], and use the neural network parameters to perform spectrum model transfer to output the white light interference spectrum. The forward neural network output calculation formula is shown in S432.

[0091] On the basis of the above embodiment, the cross-correlation demodulation step includes: S436, performing normalization processing on the reflection spectrum obtained in S3; S437, performing cross-correlation operation on the reflection spectrum obtained in S435 and the reflection spectrum after normalization processing in S436; and S438, determining the optical path difference with the largest correlation coefficient to obtain the gap value corresponding to the sample spectrum.

[0092] S436, performing normalization processing on the collected interference spectrum that passes through the patterned surface; and S423 is the same as described above.

[0093] S437-S438, using cross-correlation operation to perform gap demodulation, output the optical path difference, and then calculate the gap value corresponding to the sample spectrum; and S425 is the same as described above.

[0094] Under the nanoscale small gap, the gap value is calculated by mapping the interference spectrum of the patterned region to the thin film interference spectrum and then performing test spectrum and model library matching. The model transfer demodulation method eliminates the influence of the patterned surface on the focusing spectrum and breaks through the limitation of the template pattern layout caused by the focusing window.

[0095] Based on the above-mentioned method of gap demodulation using a white light interference focusing system, the disclosure also provides a device for gap demodulation using a white light interference focusing system.

[0096] The device for gap demodulation using a white light interference focusing system of this embodiment includes an acquisition module, a detection module, a calculation module, and a determination module.

[0097] The acquisition module is configured to acquire the white light interference signal between the template 2 and the substrate 4 to obtain a sample spectrum. In an embodiment, the acquisition module can be configured to perform the operation S1 described above, and details are not repeated here.

[0098] The detection module is configured to perform peak detection on the sample spectrum and perform spectrum preprocessing according to the peak detection result. In an embodiment, the detection module can be configured to perform the operation S2 described above, and details are not repeated here.

[0099] The calculation module is configured to calculate a reflection spectrum according to a pre-acquired reference spectrum and the sample spectrum after preprocessing in S2. In an embodiment, the calculation module can be configured to perform the operation S3 described above, and details are not repeated here.

[0100] a determining module configured to determine a demodulation method to calculate the gap value between the template 2 and the substrate 4 according to the shape of the reflection spectrum; wherein the demodulation method comprises one of a Fourier transform demodulation method, a cross-correlation calculation demodulation method and a model transfer demodulation method; in an embodiment, the determining module can be configured to perform the operation S4 described above, which will not be repeated here.

[0101] According to embodiments of the present disclosure, any of the plurality of modules of the collecting module, the detecting module, the calculating module and the determining module can be combined in one module, or any of the modules can be split into a plurality of modules. Alternatively, at least part of the functions of one or more of the modules can be combined with at least part of the functions of other modules, and implemented in one module. According to embodiments of the present disclosure, at least one of the collecting module, the detecting module, the calculating module and the determining module can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on chip, a system on board, a system on package, an application specific integrated circuit (ASIC), or any other reasonable way of integrating or packaging a circuit, etc. hardware or firmware, or in any one of software, hardware and firmware or in any appropriate combination of several of them. Alternatively, at least one of the collecting module, the detecting module, the calculating module and the determining module can be at least partially implemented as a computer program module which, when executed, can perform the corresponding functions.

[0102] The white light interferometric focusing system and the gap demodulation method of the present disclosure are based on white light interferometric focusing technology, and different demodulation methods are used under different gaps, i.e. the accuracy of the measurement is ensured and the efficiency of the measurement is improved, especially under a small gap at the nanometer level, the model transfer method is used for spectral mapping to realize the conversion of the spectrum collected through the patterned surface and the thin film interference spectrum, and the simulation test results show that the model transfer method can provide stable spectral mapping and realize nanometer precision focusing. Specifically, the present disclosure builds a white light interferometric focusing detection system based on model transfer, which can realize large measurement travel and high-precision white light interferometric gap detection through a patterned surface; the Fourier transform demodulation method is used to realize spectral demodulation of 120 μm travel, and the resolution is 1 μm under a hundred-micron travel; the cross-correlation calculation demodulation method is used for gap demodulation below 20 microns; below 1 μm, different transfer models are trained according to different patterns to realize 1 nanometer resolution gap demodulation.

[0103] The above-described specific embodiments further illustrate the purpose, technical solutions and beneficial effects of the present disclosure, and it should be understood that the above-described specific embodiments are merely specific embodiments of the present disclosure and are not intended to limit the present disclosure. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A method for gap demodulation using a white light interferometric focusing system, characterized in that, The white light interferometric focusing system includes: White light source (1); Template (2), on which a pattern (3) is prepared; The substrate (4) is placed on the substrate support (5); A spectrometer (7) is used to collect white light interference signals between the template (2) and the substrate (4) through the pattern (3); The control system (8) is used to calculate the gap value between the template (2) and the substrate (4) based on the interference signal using the Fourier transform demodulation method, the cross-correlation calculation demodulation method or the model transfer demodulation method; The method includes: S1, collect the white light interference signal between the template (2) and the substrate (4) to obtain the sample spectrum; S2, perform peak detection on the sample spectrum, and perform spectral preprocessing based on the peak detection results; S3, calculate the reflectance spectrum based on the pre-acquired reference spectrum and the sample spectrum after preprocessing in S2; S4, based on the shape of the reflection spectrum, determine the demodulation method to calculate the gap value between the template (2) and the substrate (4); wherein, the demodulation method includes one of the Fourier transform demodulation method, cross-correlation calculation demodulation method and model transfer demodulation method; The demodulation method determined in S4 includes: judging the estimated gap between the template (2) and the substrate (4) based on the shape of the reflection spectrum; when the estimated gap is less than a second threshold, calculating the gap value using the model transfer demodulation method; The calculation of the gap value using the model transfer demodulation method includes a model transfer step and a cross-correlation demodulation step. The model transfer steps include: S431, Initialize particle swarm optimization parameters and neural network parameters, wherein the neural network parameters include the number of hidden layers, the number of hidden layer nodes, the activation function, randomly generated weights, and the offset vector; S432, Callback activation function, outputs the reflectance spectrum calculated by the neural network; S433, calculate the fitness of the particle swarm; if the fitness is less than a set threshold, update the particle swarm. S434, Update until the fitness is not less than the set threshold, and output the optimized neural network parameters; S435, using the neural network parameters obtained in S434 to transfer the spectral model and output the reflectance spectrum; The cross-correlation demodulation step includes: S436, Normalize the reflectance spectrum obtained in S3; S437, perform cross-correlation calculation on the reflectance spectrum obtained in S435 and the reflectance spectrum after normalization in S436; S438, determine the optical path difference with the largest correlation coefficient, and obtain the gap value corresponding to the sample spectrum.

2. The method for gap demodulation according to claim 1, characterized in that, The method for determining demodulation in S4 also includes: When the estimated gap value is greater than the first threshold, the Fourier transform demodulation method is used to calculate the gap value. When the estimated gap value is less than or equal to the first threshold and greater than or equal to the second threshold, the cross-correlation calculation and demodulation method is used to calculate the gap value.

3. The method for gap demodulation according to claim 2, characterized in that, The first threshold ranges from 20μm to 40μm; the second threshold ranges from 1μm to 5μm.

4. The method for gap demodulation according to claim 2, characterized in that, The calculation of the gap value using the Fourier transform demodulation method includes: S411, decompose the reflection spectrum obtained in S3 into spectral profile and interference signal spectrum; S412, Perform a Fourier transform on the spectrum of the interference signal to obtain a frequency domain signal; S413, perform bandpass filtering based on the frequency domain signal; S414 performs an inverse Fourier transform on the filtered signal to obtain the analytical signal; S415, Perform a complex logarithmic operation on the analytical signal to obtain the phase of the sample spectrum; S416, Select any two phases at any wavelength, calculate the optical path difference based on the phase difference between the two phases at the two wavelengths, and obtain the gap value corresponding to the sample spectrum.

5. The method for gap demodulation according to claim 2, characterized in that, The calculation of the gap value using the cross-correlation calculation and demodulation method includes: S421, Initialize the film model and illumination parameters, and determine the gap range and gap interval; S422, using rigorous coupled-wave analysis to simulate the film model, the simulated reflection spectrum under different gaps is calculated; S423, normalize the reflectance spectrum obtained in S3; S424, perform cross-correlation calculation between the simulated reflectance spectrum in S422 and the normalized reflectance spectrum in S3; S425, determine the optical path difference with the largest correlation coefficient, and obtain the gap value corresponding to the sample spectrum.

6. A device for gap demodulation using a white light interferometric focusing system, characterized in that, The white light interferometric focusing system includes: White light source (1); Template (2), on which a pattern (3) is prepared; The substrate (4) is placed on the substrate support (5); A spectrometer (7) is used to collect white light interference signals between the template (2) and the substrate (4) through the pattern (3); The control system (8) is used to calculate the gap value between the template (2) and the substrate (4) based on the interference signal using the Fourier transform demodulation method, the cross-correlation calculation demodulation method or the model transfer demodulation method; The device includes: The acquisition module is used to acquire the white light interference signal between the template (2) and the substrate (4) to obtain the sample spectrum; The detection module is used to perform peak detection on the sample spectrum and perform spectral preprocessing based on the peak detection results; The calculation module is used to calculate the reflectance spectrum based on the pre-acquired reference spectrum and the sample spectrum after S2 preprocessing; The determination module is used to determine the gap value between the template (2) and the substrate (4) based on the shape of the reflection spectrum; wherein the demodulation method includes one of the Fourier transform demodulation method, cross-correlation calculation demodulation method and model transfer demodulation method; The method for determining demodulation includes: determining the estimated gap between the template (2) and the substrate (4) based on the shape of the reflection spectrum; when the estimated gap is less than a second threshold, calculating the gap value using the model transfer demodulation method; The calculation of the gap value using the model transfer demodulation method includes a model transfer step and a cross-correlation demodulation step. The model transfer steps include: S431, Initialize particle swarm optimization parameters and neural network parameters, wherein the neural network parameters include the number of hidden layers, the number of hidden layer nodes, the activation function, randomly generated weights, and the offset vector; S432, Callback activation function, outputs the reflectance spectrum calculated by the neural network; S433, calculate the fitness of the particle swarm; if the fitness is less than a set threshold, update the particle swarm. S434, Update until the fitness is not less than the set threshold, and output the optimized neural network parameters; S435, using the neural network parameters obtained in S434 to transfer the spectral model and output the reflectance spectrum; The cross-correlation demodulation step includes: S436, Normalize the reflectance spectrum obtained in S3; S437, perform cross-correlation calculation on the reflectance spectrum obtained in S435 and the reflectance spectrum after normalization in S436; S438, determine the optical path difference with the largest correlation coefficient, and obtain the gap value corresponding to the sample spectrum.

Images