A method for measuring optical parameters of transparent films and its application

By combining a spectrometer and a single-wavelength ellipsometer with adaptive particle swarm optimization and simulated annealing algorithms, high-precision measurement of optical parameters of transparent thin films was achieved, solving the problems of high equipment cost and low accuracy in existing technologies. This method is suitable for measuring the optical parameters of transparent and weakly absorbing thin films.

CN114894442BInactive Publication Date: 2024-11-15SOUTH CHINA NORMAL UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210409580.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-04-19
Publication Date
2024-11-15
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Existing technologies for measuring the optical parameters of transparent thin films suffer from high equipment costs, complex measurement processes, and low accuracy. In particular, the measurement accuracy decreases when the refractive index of the thin film is close to that of the substrate.

Method used

The spectral reflectance of the transparent film was measured using a spectrometer, and the refractive index and thickness were measured using a single-wavelength ellipsometer. The thickness was corrected using an adaptive particle swarm optimization algorithm, and the refractive index distribution was inverted using a simulated annealing algorithm. Finally, noise reduction and smoothing were performed.

Benefits of technology

A low-cost, simple, and high-precision method is provided, suitable for measuring the optical parameters of transparent and weakly absorbing films, with high accuracy and independent of the film's dispersion type.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114894442B_ABST
    Figure CN114894442B_ABST
Patent Text Reader

Abstract

The present invention discloses a method for measuring optical parameters of a transparent film and its application. The present invention measures the spectral reflectance of the transparent film by using a spectrometer, and measures the refractive index and thickness of the transparent film by using a single-wavelength ellipsometer; uses an adaptive particle swarm algorithm to correct the measured thickness; uses a simulated annealing algorithm to invert the refractive index of the film within a set wavelength range to obtain the refractive index distribution of the transparent film within the set wavelength; performs noise reduction and smoothing on the inverted refractive index to obtain an accurate refractive index dispersion distribution curve. The present invention can measure and solve the thickness of the film and its refractive index dispersion distribution without assuming the dispersion type of the film to be measured, and can be used for the measurement of transparent films or weakly absorbing films; and the required measurement equipment has low cost, a simple and convenient measurement process, and high measurement accuracy.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of thin film optical measurement, and in particular to a method for measuring optical parameters of a transparent thin film and its application. Background Art

[0002] With the rapid development of optoelectronics and microelectronics, optical coating has become an important means to improve the performance of various optical components. Knowing the refractive index and thickness of optical thin films is a prerequisite for the design of optical sensor devices. Therefore, in order to better improve the performance of optical components, accurately measuring optical parameters such as film thickness and complex refractive index has important application significance.

[0003] There are many methods used to measure optical constants such as film thickness and refractive index. In recent years, the commonly used methods include: ellipsometry, photometry, full spectrum fitting, etc. Among them, ellipsometry has the advantages of high sensitivity, high precision, and good repeatability, but it not only requires relatively complex and expensive experimental measurement equipment, but also relies on the known refractive index dispersion model of the sample material for the processing of measurement data, and the calculation process is relatively complicated. In addition, when the refractive index of the film is close to that of the substrate, the measurement accuracy of this method will be reduced; the envelope method based on photometry is the most commonly used method, but it requires a thicker film to produce at least 4 to 5 interference extremes, which is only suitable for thicker film parameters; the full spectrum fitting method has the characteristics of wide measurement application range, high precision, simple testing, non-harsh measurement conditions, non-destructiveness, and simultaneous measurement of optical constants and thickness of the film, so it has been widely studied and applied. However, the traditional full spectrum fitting method also requires the use of the sample refractive index dispersion model, and the selection of different dispersion models has a great influence on the results. Therefore, it is necessary to obtain a method for measuring the optical parameters of transparent films with low measurement equipment cost, convenient measurement process, and high measurement accuracy. Summary of the invention

[0004] The primary purpose of the present invention is to overcome the deficiencies of the prior art and provide a method for measuring optical parameters of a transparent film.

[0005] Another object of the present invention is to provide an application of the above method for measuring optical parameters of a transparent film.

[0006] The object of the present invention is achieved by the following technical solution: A method for measuring optical parameters of a transparent film comprises the following steps:

[0007] (1) The spectral reflectance of the transparent film is measured by a spectrometer, and the refractive index and thickness of the transparent film are measured by a single-wavelength ellipsometer;

[0008] (2) Adaptive particle swarm algorithm is used to correct the film thickness measured by single-wavelength ellipsometer;

[0009] (3) using a simulated annealing algorithm to invert the refractive index of the film within a set wavelength range to obtain the refractive index distribution of the transparent film within the set wavelength;

[0010] (4) The refractive index obtained by inversion is subjected to noise reduction and smoothing to obtain an accurate refractive index dispersion distribution curve.

[0011] The step (1) of measuring the spectral reflectance of the transparent film by using a spectrometer comprises the following steps:

[0012] The substrate and film-substrate reflection spectra are measured by a spectrometer, and the spectral reflectance distribution R(λ) of the transparent or weakly absorbing film to be measured within the set wavelength range is calculated using formula (4):

[0013]

[0014] Among them I f (λ) is the measured film-substrate reflectance spectrum, I s (λ) is the measured reflectance spectrum of the bare substrate, η3(λ) is the complex refractive index of the substrate, and n1 is the refractive index of air.

[0015] The integration time of the spectrometer is preferably set to 1.1 ms.

[0016] The refractive index and thickness of the transparent film in step (1) are preferably measured at a wavelength of 632.8 nm.

[0017] The refractive index and thickness of the transparent film in step (1) are preferably measured at least 3 times; more preferably 10 times.

[0018] The steps of the adaptive particle swarm algorithm described in step (2) are as follows:

[0019] Step 1: Determine the parameters

[0020] Firstly, the number of reflectance points N that the particle swarm algorithm needs to fit simultaneously is determined, and the wavelength difference between adjacent counting points is determined; secondly, the single-wavelength ellipsometer measurement point is selected as the base point C(λ0), and the adjacent N counting points C(λ0), A1(λ1), A2(λ2), …, A including the base point C are selected as the base point C(λ0). i (λ j ),…,A N-1 (λ N-1 ) as the reflectivity to be fitted, and the corresponding value is calculated by step (1); then, point A1, A2, ..., A N-1 The refractive index n(λ j )(j=1,2,…,N-1) and the thickness of the film d xAs unknown quantities, there are a total of N unknown quantities; finally, the single-wavelength ellipsometer measures the value n c (λ0), d c As the initial value, the upper and lower limits of the unknown disturbance in the inversion process are set at n c (λ0) and d c In the neighborhood of

[0021] Step 2: Calculation using adaptive particle swarm algorithm

[0022] S1. Set initial parameters: set the maximum and minimum values ​​of ω, c1 and c2, set the boundary values ​​of the search space and search speed, and set the maximum number of iterations k max , input the reflectivity of the point to be calculated;

[0023] S2. Initialize the population: Initialize the initial position and initial velocity of the population, and initialize the historical optimal position P of each particle in the population best,i , record the optimal position of the population G best ;

[0024] S3, change ω, c1 and c2 according to equations (7), (8) and (9), update the speed and position of each particle according to equations (5) and (6), and complete one iteration;

[0025] v i (k+1)=ωv i (k)+c1a1(P best ,i(k)-x i (k))+c2a2(G best -x i (k)) (5);

[0026] x i (k+1)=x i (k)+v i (k+1) (6);

[0027] Where k is the number of iterations, a1 and a2 are random numbers between 0 and 1; ω is the inertia weight coefficient; c1 represents the self-cognition factor of each particle, and c2 represents the social cognition factor of the entire population;

[0028] x i (k) is the position of the i-th particle in the solution space. i (k) is the velocity of the ith particle, indicating the magnitude and direction of the displacement of the ith particle in the solution space; P best,i is the historical optimal position of the i-th particle, G best is the historical optimal position of the entire population.

[0029]

[0030] c 1max and c 1min are the maximum and minimum values ​​of the self-perception factor, respectively;

[0031] S4. Calculate the mass of each particle's new position according to formula (10), and update the particle's own historical optimal position P according to the mass best,i and the population optimal position G best ;

[0032]

[0033] Where R th (λ j ) is the wavelength λ j The theoretical calculated value of reflectivity at the counting point, R exp (λ j ) is the wavelength λ j The experimental measured value of reflectivity at the counting point, δ R is the measurement error of the spectrometer reflectance; R th (λ j ) is n(λ j ) and thickness d x Substitute the value of into equations (1) and (2) to obtain; the smaller the value of the F function, the better the fitting quality;

[0034]

[0035] Where r1 and r2 are the Fresnel reflection coefficients of the interface between medium 1-2 and medium 2-3, respectively. 12 and r 23 Model;

[0036]

[0037] n1 is the refractive index of medium 1, n2 is the refractive index of medium 2;

[0038] S5. Determine whether the maximum number of iterations has been reached. If so, output the optimal particle solution. Otherwise, return to step S3. The value of the output optimal particle solution is the corrected film thickness.

[0039] The larger the value of N in the first step is, the more accurate the film thickness obtained by inversion is, but the longer it takes; it is preferably set to 5-10; more preferably 7.

[0040] The wavelength difference in the first step is preferably 6 to 10 nm, and more preferably 8 nm.

[0041] The N counting points described in the first step are preferably obtained by extending from the base point C to the adjacent two ends.

[0042] The upper and lower limits of the neighborhood described in the first step are set according to the measurement error of the single-wavelength ellipsometer used; the preferred neighborhood is ±0.2.

[0043] Step S3 1max =c 2max =2.5,c 1min =c 2min =1.25.

[0044] The inversion described in step (3) is to determine the wavelength difference Δλ of adjacent counting points of the inversion data according to the sampling interval and wavelength resolution of the spectrometer, and to divide the inversion process into two parts with the single-wavelength ellipsometer measurement point, i.e., the base point C(λ0), as the dividing point, and to perform numerical inversion using an adaptive particle swarm algorithm for the two segments with wavelengths greater than and less than the wavelength of the base point C(λ0).

[0045] The Δλ is not less than the sampling interval and wavelength resolution of the spectrometer.

[0046] The steps of the simulated annealing algorithm described in step (3) are as follows:

[0047] P1. Initialize the initial optimal solution and its internal energy: The λ c The refractive index at is n(λ0) as the initial solution X pre , calculate the internal energy of the solution as the internal energy of the initial optimal solution E(x best );

[0048] The internal energy E is calculated by formula (11):

[0049] E=|R th (λ)-R exp (λ)| (11)

[0050] Where R th is the theoretical calculated value of reflectivity, calculated by formula (1), R exp is the experimental measurement value of reflectivity;

[0051] If there is obvious high-frequency noise in the experimental measurement value curve, it needs to be low-pass filtered before calculation;

[0052] P2. Initialize simulated annealing parameters: set the initial annealing temperature T0, the cutoff temperature T end , cooling coefficient α;

[0053] P3, reset the Markov chain length M;

[0054] P4, at the initial X pre A random value is randomly selected from the neighborhood of as the new solution X when the wavelength is λ+Δλ next , calculate the internal energy E(Xnext );

[0055] P5. Calculate ΔE=E(X next )-E(X best ), if ΔE<0, the new solution is accepted, otherwise the Metropolis criterion (12) is used to determine whether to accept the new solution;

[0056]

[0057] Among them, k represents the number of iterations, E(x best ) represents the internal energy of the optimal solution, E(x new ) represents the internal energy of the new solution, T i Represents the current temperature;

[0058] P6, determine whether the number of iterations k reaches the preset Markov chain length M, if not, return to step P4;

[0059] P7, determine whether the temperature has reached the preset cut-off temperature T end If the condition is met, the loop is exited and the optimal solution of the current wavelength is output. Otherwise, the temperature is lowered by the annealing function T=T×α and then the step P3 is returned.

[0060] P8. Determine whether the current wavelength reaches the cutoff wavelength. If the condition is met, output the optimal solution for each wavelength. Otherwise, set λ=λ+Δλ and return to step P2.

[0061] The steps of denoising and smoothing the refractive index dispersion curve obtained by inversion in step (4) are as follows:

[0062] A. Perform Fourier transform on the refractive index curve obtained by inversion, perform inverse Fourier transform on the frequency domain after low-pass filtering, and reconstruct the signal after noise reduction;

[0063] B. Plot the reconstructed signal to obtain the final smooth film refractive index inversion curve.

[0064] The method for measuring the optical parameters of a transparent film is applied to measuring the thickness and / or refractive index of a transparent film or a weakly absorbing film.

[0065] Principle of the present invention:

[0066] When the incident light propagates on the upper and lower surfaces of the optical film, there will be an optical path difference between the beams. If the optical path difference is within the coherence range of the light source, the reflected light of the optical film will interfere and form an interference reflection spectrum. Since the reflection interference spectrum is related to the parameters of the optical film itself, the thickness and refractive index of the optical film can be obtained by measuring and analyzing the reflection interference spectrum.

[0067] Assume that the film sample is a transparent or weakly absorbing film (medium 2, refractive index n2) on a substrate (medium 3, complex refractive index η3 = n3-ik3), and the air surrounding the structure is medium 1, with a refractive index n1 = 1. Under normal incidence and ignoring the reflection of the lower surface of the substrate, the optical reflectivity of the isotropic uniform film is given by the following formula:

[0068]

[0069] Where r1 and r2 are the Fresnel reflection coefficients of the interface between medium 1-2 and medium 2-3, respectively. 12 and r 23 Model.

[0070]

[0071] When the complex refractive index η3 of the substrate is known, the reflectivity of the transparent film is a function of the optical parameters d and n2(λ) of the film, so the thickness and refractive index of the film can be obtained by solving the following equations:

[0072] R(λ)=f(d,n2(λ)) (3).

[0073] Since the above equation is a complex transcendental equation including multiple variables, it is difficult to obtain an analytical solution, and a numerical approximation inversion method is required to solve the optical parameters of the film. Assuming that the spectral reflectance of the transparent film at N wavelengths can be measured simultaneously, considering the refractive index dispersion of the film, N real equations can be obtained from equation (3). The parameters to be solved include the film refractive index and film thickness at N wavelengths, a total of N+1 parameters to be solved, which is greater than the number of equations. If the unknown parameters are greater than the number of equations, the equation will have multiple solutions. In order to solve this problem, a single-wavelength ellipsometer is used to measure the refractive index and thickness of the film at a certain wavelength (such as 632.8nm) as the initial condition of the inversion process, and then combined with the spectrometer to measure the spectral reflectance distribution of the film, the thickness and refractive index dispersion or distribution of the unknown transparent film can be obtained through data inversion fitting.

[0074] The idea of ​​the measurement method of the present invention is shown in Figure 1 .

[0075] The method of the present invention involves four main steps: 1. measuring the spectral reflectance of a transparent film with a spectrometer and measuring the refractive index and thickness of the film with a single-wavelength ellipsometer; 2. correcting the film thickness measured by the ellipsometer; 3. inverting the refractive index of the film within a set wavelength range; 4. performing noise reduction and smoothing on the preliminary refractive index dispersion curve obtained by inversion to obtain an accurate refractive index dispersion distribution curve.

[0076] The flowchart of the specific algorithm of the present invention is shown in Figure 2 The specific implementation methods of each step are as follows:

[0077] In the first step, a spectrometer is used to measure the substrate and film-substrate reflection spectra respectively, and the spectral reflectance distribution R(λ) of the transparent or weakly absorbing film to be measured within a set wavelength range is calculated using formula (4).

[0078]

[0079] Among them I f (λ) is the measured film-substrate reflectance spectrum, I s (λ) is the measured reflectance spectrum of the bare substrate, η3(λ) is the complex refractive index of the substrate, and n1 is the refractive index of air.

[0080] In the second step, the adaptive particle swarm algorithm is used to correct the thickness measured by the ellipsometer.

[0081] First, determine the number of reflectivity points N that the particle swarm algorithm needs to fit simultaneously (the larger the value, the more accurate the film thickness obtained by inversion, but the longer it takes), and determine the wavelength difference between adjacent counting points. Select the ellipsometer measurement point as the base point C (λ0 = 632.8nm), and include the adjacent N counting points C (λ0 = 632.8nm), A1 (λ1), A2 (λ2), ..., A i (λ j ),…,,A N-1 (λ N-1 ) is taken as the quantity to be fitted, and the corresponding value is obtained by the first step. N-1 The refractive index n(λ j )(j=1,2,…,N-1) and the thickness of the film d x As unknown quantities, there are a total of N unknown quantities. c (λ0), d c As the initial value, the upper and lower limits of the unknown disturbance in the inversion process are set at n c (λ0) and d c In the neighborhood of , an adaptive particle swarm algorithm is used for inversion.

[0082] In the calculation process of the adaptive particle swarm algorithm, the movement of particles is carried out in the solution space, which is an N-dimensional space determined by the number of unknown quantities N. For example, after a certain movement, the position of the i-th particle is x i (k)=(n(λ1),n(λ2),.,n(λ N-1 ),d x ). The movement of each particle in each iteration consists of three parts, namely, the inheritance of the previous speed, self-learning, and information interaction of the population, and its expression is:

[0083] v i(k+1)=ωv i (k)+c1a1(P best,i (k)-x i (k))+c2a2(G best -x i (k)) (5);

[0084] x i (k+1)=x i (k)+v i (k+1) (6);

[0085] Where k is the number of iterations, a1 and a2 are random numbers between 0 and 1; ω is the inertia weight coefficient; c1 represents the self-cognition factor of each particle, and c2 represents the social cognition factor of the entire population;

[0086] x i (k) is the position of the ith particle in the solution space. vi(k) is the velocity of the ith particle, indicating the magnitude and direction of the displacement of the ith particle in the solution space; P best,i is the historical optimal position of the i-th particle, G best is the historical optimal position of the entire population. The present invention adopts the following parameter change strategy:

[0087] ω=-(30k / k max -15) / (1+|30k / k max -15|)×0.3025+0.675 (7);

[0088] c1=c 1max -k(c 1max -c 1min ) / k max (8);

[0089] c2=c 2min -k(c 2min -c 2max ) / k max (9);

[0090] k is the number of iterations, c is 1max and c 1min are the maximum and minimum values ​​of the self-cognition factor, c 1max =c 2max =2.5, c 1min =c 2min =1.25.

[0091] The quality evaluation function used here is:

[0092]

[0093] Where R th (λ j ) is the wavelength λ j The theoretical calculated value of reflectivity at the counting point, R exp (λ j ) is the wavelength λ j λ j The experimental measured value of reflectivity at the counting point, δ R is the measurement error of the spectrometer reflectance. R th (λ j ) is n(λ j λ j ) and thickness d x Substitute the value of into equations (1) and (2) to obtain. The smaller the value of the F function, the better the fitting quality.

[0094] The specific steps of the adaptive particle swarm algorithm used in the present invention are as follows:

[0095] S1. Set initial parameters. Set the maximum and minimum values ​​of ω, c1 and c2, set the boundary values ​​of the search space and search speed, and set the maximum number of iterations k. max , input the reflectivity of the point to be calculated;

[0096] S2. Initialize the population. Initialize the population size, initial position and initial velocity, and initialize the historical optimal position P of each particle in the population. best , record the optimal position of the population G best ;

[0097] S3, change ω, c1 and c2 according to equations (7), (8) and (9), update the speed and position of each particle according to equations (5) and (6), and complete one iteration;

[0098] S4. Calculate the mass of each particle's new position according to formula (10), and update the particle's own historical optimal position P according to the mass best,i and the population optimal position G best ;

[0099] S5. Determine whether the maximum number of iterations has been reached. If so, output the optimal particle solution. Otherwise, return to step S3.

[0100] The final output value is the corrected film thickness.

[0101] In the third step, the refractive index of the film is inverted within the set wavelength range using the simulated annealing algorithm.

[0102] First, the wavelength difference Δ of adjacent counting points of the inversion data is determined according to the sampling interval and wavelength resolution of the spectrometer, and then the refractive index within the set wavelength range is inverted using the simulated annealing algorithm. Taking the single-wavelength ellipsometer measurement point, i.e., base point C(λ0), as the dividing point, the inversion process is divided into two parts, and numerical inversion is performed on the two sections with wavelengths greater than and less than the base point C(λ0) wavelength.

[0103] First, the portion with a wavelength greater than the base point C(λ0) is inverted. Using the single-wavelength λ0 ellipsometer measurement value n(λ0) as the initial value, in the neighborhood of n(λ0), combined with the optimization algorithm, the refractive index n(λ0+Δλ) at the wavelength of λ0+Δλ is inverted; then, using n(λ0+Δλ) as the initial value, the refractive index n(λ0+2Δλ) at the wavelength of λ0+2Δλ is inverted; the above process is repeated until the refractive index at the maximum wavelength is obtained. Then, the inversion direction is changed, and the portion with a wavelength less than the base point C(λ0) is inverted using the same method as above until the refractive index at the minimum wavelength is obtained. After the inversion is completed, the preliminary refractive index distribution of the film in the selected wavelength range can be obtained.

[0104] The inversion quality (i.e., internal energy) of each measurement point in the inversion process of the simulated annealing algorithm is evaluated using a flat evaluation function, where R th is the theoretical calculated value of reflectivity, calculated by formula (1), R exp is the experimental measurement value of reflectivity. If there is obvious high-frequency noise in the experimental measurement value curve of reflectivity, it needs to be low-pass filtered before calculation:

[0105] E=|R th (λ)-R exp (λ)| (11)

[0106] The Metropolis criterion is the core criterion of simulated annealing, which is used to calculate the probability of accepting a new solution. The expression of this criterion is as follows:

[0107]

[0108] Among them, k represents the number of iterations, E(x best ) represents the internal energy of the optimal solution, E(x new ) represents the internal energy of the new solution, T i Represents the current temperature.

[0109] The specific steps of simulated annealing inversion of film refractive index are:

[0110] P1. Initialize the initial optimal solution and its internal energy. c The refractive index at is n(λ0) as the initial solution X pre, calculate the internal energy of the solution as the internal energy of the initial optimal solution E(x best )

[0111] P2. Initialize simulated annealing parameters. Set the initial annealing temperature T0, the cut-off temperature T end , temperature reduction coefficient α;

[0112] P3. Reset the Markov chain length M;

[0113] P4, in X pre A random value is randomly selected from the neighborhood of as the new solution X when the wavelength is λ+Δλ next , calculate the internal energy E(X next );

[0114] P5. Calculate ΔE=E(X next )-E(X best ), if ΔE<0, the new solution is accepted, otherwise the Metropolis criterion (12) is used to determine whether to accept the new solution;

[0115] P6, determine whether the number of iterations k reaches the preset Markov chain length M, if not, return to step P4;

[0116] P7, determine whether the temperature has reached the preset cut-off temperature T end If the condition is met, the loop is exited and the optimal solution of the current wavelength is output. Otherwise, the temperature is lowered by the annealing function T=T×α and then the step P3 is returned.

[0117] P8. Determine whether the current wavelength reaches the cutoff wavelength. If the condition is met, output the optimal solution for each wavelength. Otherwise, set λ=λ+Δλ and return to step P2.

[0118] The fourth step is to perform noise reduction and smoothing on the inverted refractive index dispersion curve to obtain the refractive index dispersion distribution curve of the film. First, perform Fourier transform on the refractive index curve, perform low-pass filtering on the frequency domain, and then perform inverse Fourier transform to reconstruct the noise-reduced signal. Then, plot the reconstructed signal to obtain the final smooth film refractive index inversion curve.

[0119] Compared with the prior art, the present invention has the following advantages and effects:

[0120] (1) The method provided by the present invention is not only suitable for measuring the thickness of transparent films and their refractive index dispersion, but can also be used to measure the parameters of weakly absorbing films;

[0121] (2) The method provided by the present invention can measure and solve the parameters of the film without assuming the dispersion type of the film to be measured;

[0122] (3) The method provided by the present invention requires low-cost measuring equipment, a simple and convenient measuring process, does not require expensive experimental measuring equipment, and has high measuring accuracy;

[0123] (4) The method provided by the present invention has strong adaptability and can be used to measure the thickness of thin films and their refractive index dispersion on the surface of transparent substrates. BRIEF DESCRIPTION OF THE DRAWINGS

[0124] Figure 1 is a flow chart of the method of the present invention.

[0125] Figure 2 It is a flowchart of the single-point bidirectional expansion algorithm of the present invention.

[0126] Figure 3 is the reference refractive index dispersion diagram of the sample film.

[0127] Figure 4 It is the reflectance spectrum of sample film-substrate and bare substrate.

[0128] Figure 5 It is the spectral reflectance curve of the thin film.

[0129] Figure 6 It is the refractive index dispersion curve after noise reduction and reconstruction and the reference refractive index dispersion curve. DETAILED DESCRIPTION

[0130] The present invention is further described in detail below in conjunction with embodiments and drawings, but the embodiments of the present invention are not limited thereto.

[0131] Example 1

[0132] The silicon dioxide film (SiO2 / Si) on the surface of the silicon substrate (purchased from Wuhan Yiguang Technology Co., Ltd.) was used as the experimental sample for the experiment, and the measured values ​​of the optical parameters of the sample film and substrate by Wuhan Yiguang Technology Co., Ltd. were used as reference values ​​(the average measured thickness of the film was 103.62nm, and the refractive index dispersion was as follows Figure 3 The spectrometer model for measuring the spectral reflectance of the film is: AvaSpec-2048-USB2 fiber optic spectrometer. The sampling interval of the spectrometer is 0.595nm.

[0133] The experimental measurement and numerical inversion process includes the following steps:

[0134] (1) The refractive index and initial thickness of the sample film at a specific wavelength (632.8 nm) were measured using a single-wavelength ellipsometer (model: L116S300 reflective single-wavelength Stokes ellipsometer). The average value of ten measurements was taken as the initial refractive index value of the inversion process.

[0135] (2) Measure the thin film sample, adjust the sample surface to be perpendicular to the incident light, set the spectrometer integration time (the spectrometer integration time is set to 1.1 ms), and use a fiber optic spectrometer to measure the reflectance spectrum of the sample film-substrate I f And the reflectance spectrum of the substrate I s . Figure 4 Reflection spectra of sample film-substrate and bare substrate.

[0136] (3) Set up the program to input the refractive index measurement value of the ellipsometer and the reflection spectrum measured by the fiber optic spectrometer.

[0137] (4) The program automatically calculates the sample reflectance according to formula (4). Figure 5 is the spectral reflectance R of the film exp (λ).

[0138] (5) Enter the adaptive particle swarm algorithm to correct the thickness of the sample to be measured. The program automatically selects the ellipsometer measurement point as the base point C (λ0 = 632.8nm), sets the wavelength difference between adjacent counting points to 8nm, and takes the reflectivity of the adjacent N (experimental setting N = 7) counting points A0 (C) (λ0 = 632.8nm), A1 (λ1 = 608.8nm), A2 (λ2 = 616.8nm), A3 (λ3 = 624.8nm), A4 (λ4 = 640.8nm), A5 (λ5 = 648.8nm), A6 (λ6 = 656.8nm) including the base point C as the quantity to be fitted, and the corresponding values ​​are calculated in the previous step. The refractive index n (λ i )(i=1, 2, ..., 6) and the thickness d of the film x As the unknown quantity, the value n is measured by single-wavelength ellipsometer c (λ0=632.8nm), d c As the initial value, the upper and lower limits of the unknown disturbance in the inversion process are set at n c (λ0)±0.2 and d c Within the range of ±10nm (the upper and lower limits are related to the measurement error of the single-wavelength ellipsometer used. Due to the dispersion in a certain wavelength range, the upper and lower limits of the refractive index can be appropriately set larger), the adaptive particle swarm algorithm is used for inversion:

[0139] S1. Set initial parameters. Set c 1max =c 2max =2.5, c 1min =c 2min = 1.25, set the boundary value of N search space and search speed, and set the upper and lower limits of the first N-1(6) dimensions of the search space to n c (λ0)±0.2, the N(7)th dimension is set to d c±10nm Maximum number of iterations k max =5000, input the reflectivity of counting points A0(C) (λ0=632.8nm), A1 (λ1=608.8nm), A2 (λ2=616.8nm), A3 (λ3=624.8nm), A4 (λ4=640.8nm), A5 (λ5=648.8nm), A6 (λ6=656.8nm). The search speed is also an N-dimensional vector, the minimum value of each dimension is 0, the maximum value of the first N-1 dimensions is set to 0.001, and the maximum value of the Nth dimension is set to 0.1nm;

[0140] S2. Initialize the population. Initialize the population size, initial position and initial velocity. The initial population size is 200. Initialize the initial position of each particle, for example, the initial position of the i-th particle is x i (0)=(n(λ1),n(λ2),...,n(λ N-1 ), d x ), the value of each dimension of the vector is a random value of the corresponding dimension in the search space. The initial velocity of each particle is a random value within the upper and lower limits of the velocity. The initial position of each particle is recorded as the historical optimal position P of each particle in the population best,i , calculate the mass of each particle's position according to formula (10), and record the particle position corresponding to the minimum value as the optimal position of the population G best ;

[0141] S3, change ω, c1 and c2 according to equations (7), (8) and (9), update the speed and position of each particle according to equations (5) and (6), and complete one iteration;

[0142] S4. Calculate the mass of each particle's new position according to formula (10), and update the particle's own historical optimal position P according to the mass best,i and the population optimal position G best ;

[0143] S5. Determine whether the maximum number of iterations has been reached. If so, output the optimal particle solution. Otherwise, return to step S3.

[0144] The value of the quality evaluation function (10) is minimized, and finally only the inversion result dx of the optimal solution thickness is output as the correction value of the ellipsometer measurement value.

[0145] (6) Enter the simulated annealing algorithm program to calculate the refractive index. Set the initial annealing temperature T0 = 10 5 ℃, cut-off temperature T end =10 -1 ℃, cooling coefficient α=0.95, Markov chain length M=300, set the wavelength difference between adjacent measurement points to Δλ=2nm. xThe reflectivity R is input into the simulated annealing program for numerical inversion, and finally the film refractive index dispersion curve with noise is output. The specific steps of simulated annealing inversion of film refractive index are:

[0146] P1. Initialize the initial optimal solution and its internal energy. c The refractive index at is n(λ0) as the initial solution X pre , calculate the internal energy of the solution as the internal energy of the initial optimal solution E(x best );

[0147] P2. Initialize simulated annealing parameters. Set the initial annealing temperature T0 = 10 5 ℃, cut-off temperature T end =10 -1 ℃, cooling coefficient α=0.95;

[0148] P3, reset the Markov chain length M = 300;

[0149] P4, in X pre A random value is randomly selected as the new solution X when the wavelength is λ+Δλ in the neighborhood of n(λ0)±0.1 (i.e., the upper and lower limits of the neighborhood are related to the measurement error of the single-wavelength ellipsometer used. A small error can reduce the variation range). next , calculate the internal energy E(X next );

[0150] P5. Calculate ΔE=E(X next )-E(X best ), if ΔE<0, the new solution is accepted, otherwise the Metropolis criterion (12) is used to determine whether to accept the new solution;

[0151] P6, determine whether the number of iterations k reaches the preset Markov chain length M, if not, return to step P4;

[0152] P7, determine whether the temperature has reached the preset cut-off temperature T end If the condition is met, the loop is exited and the optimal solution of the current wavelength is output. Otherwise, the temperature is lowered by the annealing function T=T×α and then the step P3 is returned.

[0153] P8. Determine whether the current wavelength reaches the cutoff wavelength. If the condition is met, output the optimal solution for each wavelength. Otherwise, set λ=λ+Δλ and return to step P2.

[0154] (7) The noisy refractive index dispersion curve is filtered and smoothed to output a reconstructed dispersion curve, as follows: first, the refractive index curve is Fourier transformed, and the frequency domain is low-pass filtered and then inverse Fourier transformed to reconstruct a noise-reduced signal; then, the reconstructed signal is plotted to obtain a final smooth film refractive index inversion curve. Figure 6 It is the refractive index dispersion curve after noise reduction and reconstruction and the reference refractive index dispersion curve.

[0155] The average thickness of the silicon dioxide film measured by the single-wavelength ellipsometer is 104.1nm. The thickness corrected by the adaptive simulated annealing method in the present invention is 103.48nm, which is only 0.32nm different from the thickness reference value of 103.62nm measured by Wuhan Yiguang Technology Co., Ltd., and the relative error is 0.135%. Taking the measured value of the refractive index of the sample film by Wuhan Yiguang Technology Co., Ltd. as the reference value, the average relative error between the measured refractive index and the refractive index reference value is calculated by formula (13):

[0156]

[0157] Where m is the number of points for refractive index calculation, n is 2exp is the measured value of the film refractive index, n 2ref is the reference value of the film refractive index. The average relative error of the film refractive index after the final noise reduction is calculated to be 0.092%. The film was measured five times in succession, and the average relative errors of each experimental result were 0.0923%, 0.0963%, 0.0969%, 0.1536%, and 0.0955%, respectively, with a variance of 0.0241%. This shows that the film refractive index dispersion curve inverted by the method of the present invention has a small error, high accuracy, and good repeatability.

[0158] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be equivalent replacement methods and are included in the protection scope of the present invention.

Claims

1. A method for measuring optical parameters of a transparent film, characterized in that The steps include: (1) The spectral reflectance of the transparent film is measured by a spectrometer, and the refractive index and thickness of the transparent film are measured by a single-wavelength ellipsometer; (2) Adaptive particle swarm algorithm is used to correct the film thickness measured by single-wavelength ellipsometer; (3) using a simulated annealing algorithm to invert the refractive index of the film within a set wavelength range to obtain the refractive index distribution of the transparent film within the set wavelength; (4) De-noising and smoothing the inverted refractive index to obtain an accurate refractive index dispersion distribution curve; The steps of the adaptive particle swarm algorithm described in step (2) are as follows: Step 1: Determine the parameters Firstly, the number of reflectance points N that the particle swarm algorithm needs to fit simultaneously is determined, and the wavelength difference between adjacent counting points is determined; secondly, the single-wavelength ellipsometer measurement point is selected as the base point C(λ0), and the adjacent N counting points C(λ0), A1(λ1), A2(λ2), ..., A including the base point C are selected as the base point C(λ0). i (λ j ), ..., A N-1 (λ N-1 ) as the reflectivity to be fitted, and the corresponding value is calculated by step (1); then, point A1, A2, ..., A N-1 The refractive index n(λ j )(j=1, 2, ..., N-1) and the thickness d of the film x As unknown quantities, there are a total of N unknown quantities; finally, the single-wavelength ellipsometer measures the value n c (λ0), d c As the initial value, the upper and lower limits of the unknown disturbance in the inversion process are set at n c (λ0) and d c In the neighborhood of Step 2: Calculation using adaptive particle swarm algorithm S1. Set initial parameters: set the maximum and minimum values ​​of ω, c1 and c2, set the boundary values ​​of the search space and search speed, and set the maximum number of iterations k max , input the reflectivity of the point to be calculated; S2. Initialize the population: Initialize the initial position and initial velocity of the population, and initialize the historical optimal position P of each particle in the population best,i , record the optimal position of the population G best ; S3, change ω, c1 and c2 according to equations (7), (8) and (9), update the speed and position of each particle according to equations (5) and (6), and complete one iteration; V i (k+1)=ωv i (k)+c1a1(P best,i (k)-x i (k))+c2a2(G best -x i (k)) (5); x i (k+1)=x i (k)+v i (k+1) (6); Where k is the number of iterations, a1 and a2 are random numbers in the range of 0 to 1; ω is the inertia weight coefficient; c1 represents the self-cognition factor of each particle, and c2 represents the social cognition factor of the entire population; x i (k) is the position of the ith particle in the solution space. i (k) is the velocity of the ith particle, indicating the magnitude and direction of the displacement of the ith particle in the solution space; P best,i is the historical optimal position of the i-th particle, G best is the historical optimal position of the entire population; ω-(30k / k max -15) / (1+|30k / k max -15|)×0.3025+0.675 (7); c1=c 1max -k(c 1max -c 1min ) / k max (8); c2=c 2min -k(c 2min -c 2max ) / k max (9); k is the number of iterations, c is 1max and c 1min are the maximum and minimum values ​​of the self-perception factor, respectively; S4. Calculate the mass of each particle's new position according to formula (10), and update the particle's own historical optimal position P according to the mass best,i and the population optimal position G best ; Where R th (λ j ) is the wavelength λ j The theoretical calculated value of reflectivity at the counting point, R exp (λ j ) is the wavelength λ j The experimental measured value of reflectivity at the counting point, δ R is the measurement error of the spectrometer reflectance; R th (λ j ) is n(λ j ) and thickness d x Substitute the value of into equations (1) and (2) to obtain; the smaller the value of the F function, the better the fitting quality; Where r1 and r2 are the Fresnel reflection coefficients of the interface between medium 1-2 and medium 2-3, respectively. 12 and r 23 Model; n1 is the refractive index of medium 1, n2 is the refractive index of medium 2; S5, judging whether the maximum number of iterations has been reached, and if so, outputting the optimal particle solution, otherwise returning to step S3; the value of the output optimal particle solution is the corrected film thickness; N described in the first step is set to 5 to 10; The wavelength difference described in the first step is 6 to 10 nm; The N counting points described in the first step are obtained by extending from the base point C to the adjacent two ends; Step S3 1max =c 2max =2.5, c 1min =c 2min =1.25; The inversion described in step (3) is to determine the wavelength difference Δλ of adjacent counting points of the inversion data according to the sampling interval and wavelength resolution of the spectrometer, and to divide the inversion process into two parts with the single-wavelength ellipsometer measurement point, i.e., the base point C(λ0), as the dividing point, and to perform numerical inversion on the two sections with wavelengths greater than and less than the wavelength of the base point C(λ0) respectively; the thickness correction value and the reflectivity are input into the simulated annealing program for numerical inversion; The steps of the simulated annealing algorithm described in step (3) are as follows: P1. Initialize the initial optimal solution and its internal energy: The λ c The refractive index at is n(λ0) as the initial solution X pre , calculate the internal energy of the solution as the internal energy of the initial optimal solution E(x best ); The internal energy E is calculated by formula (11): E=|R th (L)-R exp (l)| (11); Where R th is the theoretical calculated value of reflectivity, calculated by formula (1), R exp is the experimental measurement value of reflectivity; If there is obvious high-frequency noise in the experimental measurement value curve, it needs to be low-pass filtered before calculation; P2. Initialize simulated annealing parameters: set the initial annealing temperature T0, the cutoff temperature T end , temperature reduction coefficient α; P3, reset the Markov chain length M; P4, at the initial X pre A random value is randomly selected from the neighborhood of as the new solution X when the wavelength is λ+Δλ next , calculate the internal energy E(X next ); P5. Calculate ΔE=E(X next )-E(X best ), if ΔE<0, the new solution is accepted, otherwise the Metropolis criterion (12) is used to determine whether to accept the new solution; Among them, k represents the number of iterations, E(x best ) represents the internal energy of the optimal solution, E(x new ) represents the internal energy of the new solution, T i Represents the current temperature; P6, determine whether the number of iterations k reaches the preset Markov chain length M, if not, return to step P4; P7, determine whether the temperature has reached the preset cut-off temperature T end If the condition is met, the loop is exited and the optimal solution of the current wavelength is output. Otherwise, the temperature is lowered by the annealing function T=T×α and then the step P3 is returned. P8. Determine whether the current wavelength reaches the cutoff wavelength. If the condition is met, output the optimal solution for each wavelength. Otherwise, set λ=λ+Δλ and return to step P2.

2. The method for measuring optical parameters of a transparent film according to claim 1, characterized in that: The steps of measuring the spectral reflectance of the transparent film using a spectrometer in step (1) are as follows: The substrate and film-substrate reflection spectra are measured by a spectrometer, and the spectral reflectance distribution R(λ) of the transparent or weakly absorbing film to be measured within the set wavelength range is calculated using formula (4): Among them I f (λ) is the measured film-substrate reflectance spectrum, I s (λ) is the measured reflectance spectrum of the bare substrate, η3(λ) is the complex refractive index of the substrate, and n1 is the refractive index of air.

3. The method for measuring optical parameters of a transparent film according to claim 1, characterized in that: The integration time of the spectrometer is set to 1.1 ms; The refractive index and thickness of the transparent film in step (1) are measured at a wavelength of 632.8 nm; The refractive index and thickness of the transparent film in step (1) are measured at least three times.

4. The method for measuring optical parameters of a transparent film according to claim 1, characterized in that: The steps of denoising and smoothing the refractive index dispersion curve obtained by inversion in step (4) are as follows: A. Perform Fourier transform on the refractive index curve obtained by inversion, perform inverse Fourier transform on the frequency domain after low-pass filtering, and reconstruct the signal after noise reduction; B. Plot the reconstructed signal to obtain the final smooth film refractive index inversion curve.

5. Use of the method for measuring optical parameters of a transparent film according to any one of claims 1 to 4 in measuring the thickness and / or refractive index of a transparent film or a weakly absorbing film.