TSOM microstructure linewidth measurement method based on interference fringe focusing

Abstract

The present invention discloses a TSOM microstructure linewidth measurement method based on interference fringe focusing, which belongs to the field of optical detection. The steps include: using a low-coherence interference optical path to perform over-focus scanning on a microstructure linewidth sample to obtain an image sequence with surface information of the object to be measured; determining the interference signal envelope peak index in the image sequence; establishing an over-focus scanning optical microscopy image of the object to be measured with this position as the focusing plane; using a time-domain finite difference algorithm and angular spectrum theory to simulate low-coherence interference over-focus scanning scattered light field images of different linewidth models to establish a standard size linewidth model database; and extracting the linewidth of the structure to be measured in combination with a library matching algorithm. The present invention breaks through the diffraction limit of the low-coherence interference measurement system, improves the lateral resolution of the system, and realizes super-resolution measurement of linewidth; at the same time, the focusing plane is accurately positioned by interference fringes, reducing the over-focus scanning linewidth measurement error introduced by the system focal depth, and ensuring the effectiveness of the measurement method.

Description

Technical Field

[0001] The present invention belongs to the field of optical detection, and in particular relates to a TSOM microstructure line width measurement method based on interference fringe focusing. Background Art

[0002] Step-type microstructure components are widely used in semiconductors, micro-electromechanical systems, aerospace, and modern optics. Their morphological characteristic parameters such as line width, depth, and sidewall angle significantly affect the quality and performance of the components. Among them, the measurement of line width parameters has always been the focus and difficulty in the field of precision measurement. Traditional line width measurement often uses scanning probe microscopes, scanning electron microscopes, and atomic force microscopes. Such methods are often not suitable for online detection in industrial scenarios due to their high cost and limitations such as damage caused by contact with the sample surface. With the development of non-destructive measurement technology based on optical methods, low-coherence microinterferometry has gradually become the mainstream method for measuring microstructure line width, but its lateral measurement accuracy is limited by the lateral resolution of the system and can only reach the micron level. Attota et al. proposed a model-based microstructure line width measurement method, namely through-focus scanning optical microscopy (TSOM). This method scans near the focus along the optical axis and calculates and compares the defocused images of the microstructure sample to achieve super-resolution line width measurement. Although the TSOM method does not require precise focusing during the scanning process, it still stacks the intensity distributions of positive and negative defocused images centered on the focal plane when generating a two-dimensional TSOM image. Due to the influence of the optical system's depth of focus, the scanned image achieves the "clearest" standard within a certain axial scanning range. Current TSOM image stacking methods still stack positive and negative defocused images centered on the focal plane of the TSOM image, as subjectively evaluated by the human eye. This makes it difficult to determine the exact location of the focal plane within the focal depth range during actual measurement, affecting the accuracy of the measurement results. Summary of the Invention

[0003] In response to the above-mentioned defects or improvement needs of the prior art, the present invention provides a TSOM microstructure line width measurement method based on interference fringe focusing, which aims to solve the problem of decreased TSOM line width measurement accuracy due to the difficulty in determining the focusing plane within the system focal depth when stacking TSOM images.

[0004] The technical solution to achieve the purpose of the present invention is: a TSOM microstructure line width measurement method based on interference fringe focusing, the steps are as follows:

[0005] Step 1: Perform over-focus scanning on the sample to be tested by a low-coherence micro-interferometer to obtain a low-coherence interference image sequence containing surface information of the sample to be tested.

[0006] Step 2: Based on the low-coherence interference image sequence, the intensity distribution of the scattered light field of the sample to be tested and the low-coherence interference signal are extracted along the scanning height, and the low-coherence interference signal envelope peak index is calculated using the HDVSI algorithm.

[0007] Step 3: Using the scanning height where the low-coherence interference signal envelope peak index is located as the focusing plane, stack the scattered light field intensity distribution to construct a low-coherence interference over-focus scanning optical microscopy image of the sample to be tested.

[0008] Step 4: remove background signals from the low-coherence interference overfocus scanning optical microscopy image, retain the scattered light fields at different defocus positions of the sample to be measured, and obtain the low-coherence interference overfocus scanning optical microscopy image as the measurement image.

[0009] Step 5: Use the finite-difference time-domain algorithm and angular spectrum theory to simulate the low-coherence interference over-focus scanning scattered light field images of different linewidth models as simulation images, and then establish a standard size linewidth model database.

[0010] Step 6: Input the measured image into the standard size line width model database, and extract the line width of the sample to be measured based on the library matching algorithm.

[0011] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects:

[0012] (1) TSOM super-resolution measurement of microstructure line width is affected by the depth of focus, with an error of up to 1.5 μm. The present invention can effectively locate the focal plane of stacked TSOM images, reduce the error caused by the depth of focus, and improve the accuracy of TSOM super-resolution measurement of microstructure line width.

[0013] (2) Through high-precision numerical simulation, a model can be established to simulate the interference light field distribution of the sample to be measured under different conditions. In the absence of samples, a standard size line width model database can be quickly and accurately established, which expands the measurement accuracy and reduces the measurement cost.

[0014] (3) It can be directly applied on a commercial low-coherence micro-interferometer without changing the optical path structure. It is simple and easy to operate, has low cost, and can achieve online measurement. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 Flowchart of the TSOM microstructure line width measurement method based on interference fringe focusing.

[0016] Figure 2 Schematic diagram of selecting the region of interest on the sample to be tested.

[0017] Figure 3 Schematic diagram of the interference light field intensity distribution in the region of interest.

[0018] Figure 4 Schematic diagram of low-coherence interference signal.

[0019] Figure 5 Schematic diagram of low-coherence interferometric over-focus scanning optical microscopy image.

[0020] Figure 6 Schematic diagram of the measurement image after removing the background signal.

[0021] Figure 7 Schematic diagram of the simulation image.

[0022] Figure 8 Schematic diagram of the results of measuring microstructure line width using the library matching method. DETAILED DESCRIPTION

[0023] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0024] To address the problem of difficulty in determining the focal plane due to the influence of focal depth during traditional TSOM image stacking, the present invention utilizes low-coherence interference fringes and adopts the HDVSI algorithm to locate the peak of the low-coherence interference signal envelope, thereby accurately locating the scanning height where the focusing plane is located. Low-coherence interference over-focus scanning optical microscopy images are stacked with this scanning height as the center, effectively improving the accuracy of measuring the line width of the sample to be measured by the library matching method.

[0025] Combine Figure 1 , a TSOM microstructure linewidth measurement method based on interference fringe focusing, the steps are as follows:

[0026] S1. Use a low-coherence microscopic interferometer with a scanning step of λ / 8 to vertically scan the sample to be tested, and obtain a low-coherence interferometric image sequence containing surface information of the sample to be tested. Where λ is the central wavelength of the light source.

[0027] Furthermore, a low-coherence microscopic interferometer with a λ / 8 stepping function is used to vertically scan the sample to be measured, acquiring two image sequences {p1, p2}, where p1 = {p11, p12, …, p1n} and p2 = {p21, p22, …, p2n}, where 1, 2, …, n correspond to the number of images (frames) in each sequence. Image sequence p1 is a low-coherence microscopic interferometer image containing surface information of the sample to be measured, while image sequence p2 is a low-coherence microscopic interferometer image of a smooth, clean silicon surface. p1-p2 can be used as the measurement image after background noise is removed.

[0028] S2. Calculate the following for each image in the low coherence interference image sequence: Figure 2 The intensity distribution of the scattered light field in the area shown is as follows: Figure 3 As shown, and axial Figure 4 The low coherence interference signal is shown. Using the HDV SI algorithm, the low coherence interference signal envelope peak index h is obtained.

[0029] Furthermore, the HDVSI algorithm is used to obtain the low coherence interference signal envelope peak index h.

[0030] The low coherence interference expression for a single pixel in the image can be expressed as:

[0031]

[0032] Where I0 is the background light intensity, γ represents the fringe contrast, Represents the amplitude information of the pixel point, Represents the phase information of the pixel; h0 is the surface height, h step is the scanning step, l c is the coherence length of the light source, λ is the central wavelength of the light source, is the initial phase of the sample, I i (i) represents the light intensity value on the i-th image.

[0033] The HDVSI algorithm is as follows:

[0034] Use the centroid method to roughly locate the low-coherence interference signal envelope peak index to obtain the rough positioning index h1. The formula of the centroid method is as follows:

[0035]

[0036] Where M i is the contrast of the i-th interference pattern during the scanning process, and i is the interference pattern number. The contrast algorithm formula is as follows:

[0037]

[0038] Where, I i is the light intensity value of the i-th interference image;

[0039] The orthogonal demodulation four-step phase shift algorithm is used to precisely locate the low-coherence interference signal envelope peak index to obtain the precise positioning index h2. The formula of the orthogonal demodulation four-step phase shift algorithm is as follows:

[0040]

[0041] Where ΔΦ0 is the phase shift;

[0042] Calculate the index intermediate variable h r1 :

[0043]

[0044] In the formula, round is the function rounding operation;

[0045] Calculate the index intermediate variable h r2 :

[0046]

[0047] Calculate and compare the absolute value of the difference between the two index intermediate variables and h1-h2, and the smaller value is used as the final index intermediate variable h r .

[0048] Low coherence interference signal envelope peak index h:

[0049] h=h r +h2

[0050] S3, with the scanning height h as the center, the scattered light field intensity distribution calculated in the low coherence interference image sequence is stacked according to the spatial position of each scan, and the following is obtained: Figure 5 The low coherence interference over-focus scanning optical microscopy image of the sample to be tested is shown.

[0051] Furthermore, the scattered light field intensity distribution calculated from the stacked image sequence according to the spatial position of each scan is obtained with the scanning height h as the center, and the steps are as follows:

[0052] Calculate the index h of the center plane of the low-coherence interference over-focus scanning optical microscopy image round :

[0053] h round =ro un d(h)

[0054] With index h round The low coherence interference image of is taken as the central plane. In the low coherence interference image sequence, a low coherence interference image is measured every 7 times phase shift. The average value along the groove direction is taken as the scattering intensity of the point. The scattered light field intensity distribution in the line width direction of each low coherence interference image is calculated.

[0055] The scattered light field intensity distribution curves are stacked according to their respective scanning heights, with 20 curves stacked in the positive defocus direction and the negative defocus direction, for a total of 41 curves forming a two-dimensional TSOM image, which is the low-coherence interference through-focus scanning optical microscopy image of the sample to be tested.

[0056] S4, remove the background signal and retain the scattered light field of the sample at different defocus positions to obtain the following Figure 6 Measurement images shown.

[0057] S5. Use the time-domain finite difference algorithm and angular spectrum theory to simulate the low-coherence interference over-focus scanning scattered light field images of different line width models, and obtain the following Figure 7The simulation image shown is used to establish a standard size line width model database.

[0058] Furthermore, the finite-difference time-domain algorithm and angular spectrum theory are used to simulate the scattered light field images of low-coherence interference over-focus scanning of different linewidth models to establish a standard size linewidth model database. The steps are as follows:

[0059] Set simulation parameters such as simulation model (groove width, groove depth), FDTD simulation area, light source parameters, monitor, etc.

[0060] The point scanning-based simulation method simulates the near-field distribution of the reflected light field corresponding to each pixel area of ​​the image sensor.

[0061] The sample to be tested is moved vertically according to the preset step length to scan the sample to be tested and its background.

[0062] Using angular spectrum theory, the two data sets are far-field transmitted to obtain the light field distribution in the imaging plane.

[0063] Using the 3D-PSF model of low-coherence micro-interference, the interference light field is calculated on the imaging plane and the background noise is removed to obtain the following: Figure 7 The simulation image is shown.

[0064] Determine whether all sizes are simulated. If not, change the simulation model and re-execute the simulation. If all sizes are simulated, obtain the standard size line width model database.

[0065] S6. Input the measured image into the standard size line width model database, based on the library matching algorithm, e.g. Figure 8 As shown, the line width of the sample to be tested is extracted.

[0066] Furthermore, the library matching algorithm is as follows:

[0067] The over-focus scanning scattered light field images of different line width models in the standard size line width model database are simulation images, with a size of M×N matrix, where M ranges from 0 to 2048 and N ranges from 0 to 2048;

[0068] Performing bicubic interpolation on the measurement image obtained in step S4, that is, adjusting the measurement image to an M×N matrix;

[0069] The structural similarity SSIM between each simulated image in the standard size line width model database and the resized measured image is calculated using the following formula:

[0070]

[0071] Among them, x is the simulated image, y is the measured image, μ x is the local mean of the simulated image x, μy is the local mean of the measured image y, σ x is the standard deviation of the simulated image x, σ y is the standard deviation of the measured image y, σ xy is the cross-covariance of images x and y, C1=(0.01×L x ) 2 , L x is the dynamic range of the simulated image x, C2=(0.03×L y ) 2 , L y To measure the dynamic range of image y;

[0072] like Figure 8 As shown, by evaluating the SSIM value between the standard size line width model database and the low coherence interference overfocus scanning optical microscopy image, the maximum SSIM value is used as the best matching target, and the line width of the corresponding model in the standard size line width model database is used as the line width of the structure to be measured.

[0073] Those skilled in the art will readily appreciate that the above examples are merely preferred embodiments of the present invention and are not intended to limit the present invention. The present invention is not limited to the above examples. It should be noted that various modifications and improvements may be made without departing from the principles of the present invention, and these modifications and improvements should also be considered within the scope of protection of the present invention.

Claims

1. A TSOM microstructure linewidth measurement method based on interference fringe focusing, characterized in that: Here are the steps: Step 1: Perform over-focus scanning on the sample to be tested by a low-coherence micro-interferometer to obtain a low-coherence interference image sequence containing surface information of the sample to be tested; Step 2: Based on the low-coherence interference image sequence, the intensity distribution of the scattered light field of the sample to be tested and the low-coherence interference signal are extracted along the scanning height, and the low-coherence interference signal envelope peak index is calculated using the HDVSI algorithm; Among them, HDVSI is used to calculate the low coherence interference signal envelope peak index h, which specifically includes: Use the centroid method to roughly locate the low-coherence interference signal envelope peak index to obtain the rough positioning index h1. The formula of the centroid method is as follows: Where M i is the contrast of the i-th interference pattern during the scanning process, i is the interference image number; The contrast algorithm formula is as follows: Where, I i is the light intensity value of the i-th interference image; The orthogonal demodulation four-step phase shift algorithm is used to precisely locate the low-coherence interference signal envelope peak index to obtain the precise positioning index h2. The formula of the orthogonal demodulation four-step phase shift algorithm is as follows: Where ΔΦ0 is the phase shift; Calculate the index intermediate variable h r1 : In the formula, round is the function rounding operation; Calculate the index intermediate variable h r2 : Calculate and compare the absolute value of the difference between the two index intermediate variables and h1-h2, and the smaller value is used as the final index intermediate variable h r ; Low coherence interference signal envelope peak index h: h=h r +h2 Step 3: Using the scanning height where the low-coherence interference signal envelope peak index is located as the focusing plane, stacking the scattered light field intensity distribution to construct a low-coherence interference over-focus scanning optical microscopy image of the sample to be tested; The stacked scattered light field intensity distribution specifically includes: Calculate the index h of the center plane of the low-coherence interference over-focus scanning optical microscopy image round : h round =round(h) With index h round The low coherence interference image of is taken as the central plane, and a low coherence interference image is taken at every 7-fold phase shift interval in the low coherence interference image sequence, and the intensity distribution of the scattered light field is calculated; According to the respective scanning heights, 20 scattered light field intensity distribution curves are stacked in the positive defocus direction and the negative defocus direction, and a total of 41 curves form a two-dimensional TSOM image, which is the low-coherence interference through-focus scanning optical microscopy image of the sample to be tested; Step 4: removing background signals from the low-coherence interference over-focus scanning optical microscopy image, retaining scattered light fields at different defocus positions of the sample to be measured, and obtaining a low-coherence interference over-focus scanning optical microscopy image as a measurement image; Step 5: Using the finite-difference time-domain algorithm and angular spectrum theory to simulate low-coherence interference over-focus scanning scattered light field images of different linewidth models as simulation images, and then establishing a standard size linewidth model database; Step 6: Input the measured image into the standard size line width model database, and extract the line width of the sample to be measured based on the library matching algorithm.

2. The TSOM microstructure linewidth measurement method based on interference fringe focusing according to claim 1, characterized in that: The sample to be tested adopts a silicon substrate, and a groove with high depth-width ratio characteristics is etched on the silicon substrate.

3. The TSOM microstructure linewidth measurement method based on interference fringe focusing according to claim 1, characterized in that: In step 1, the sample to be tested is subjected to over-focus scanning by a low-coherence micro-interferometer, specifically including: The illumination light emitted by the low-coherence light source is incident on the surface of the sample to be tested and the reference plane through a spectroscope. After being modulated and scattered by the sample to be tested, it interferes with the reference light reflected back by the reference plane, forming a low-coherence interference image containing the surface information of the sample to be tested on the image detector; According to the set moving distance, the sample to be tested is defocused, focused, and defocused layer by layer by vertically moving the piezoelectric ceramic translation stage, where the interference objective lens of the low-coherence micro-interferometer is located on the piezoelectric ceramic translation stage; By vertically moving the piezoelectric ceramic translation stage, a low-coherence interference image sequence containing the surface information of the sample to be measured is obtained.

4. The TSOM microstructure linewidth measurement method based on interference fringe focusing according to claim 3, characterized in that: The moving distance is set to 0-20 μm, and sequential vertical scanning is performed according to a phase shift of π / 2. Each time a scan is performed, a low-coherence interference image is recorded by the image detector.

5. The TSOM microstructure linewidth measurement method based on interference fringe focusing according to claim 3, characterized in that: The image detector is a CMOS detector or a CCD detector.

6. The TSOM microstructure linewidth measurement method based on interference fringe focusing according to claim 1, characterized in that: In step 5, the different line width model refers to a line width trench simulation model having a different line width from the sample to be tested, the same side wall angle, and the same depth.

7. The TSOM microstructure linewidth measurement method based on interference fringe focusing according to claim 1, characterized in that: In step 5, the standard size line width model database includes line widths ranging from 0.5 to 5 μm.

8. The TSOM microstructure linewidth measurement method based on interference fringe focusing according to claim 1, characterized in that: In step 6, the measured image is input into the standard size line width model database, and the line width of the sample to be measured is extracted based on the library matching algorithm, as follows: Step 6-1, the low-coherence interference over-focus scanning scattered light field images of different line width models in the standard size line width model database are simulated images, the size of which is an M×N matrix, where M ranges from 0 to 2048 and N ranges from 0 to 2048; Step 6-2: Perform bicubic interpolation on the measured image obtained in step 4 to adjust the matrix size to M×N; Step 6-3: Calculate the structural similarity SSIM between each simulated image and the resized measured image in the standard size line width model database. The calculation formula is as follows: Among them, x is the simulated image, y is the measured image, μ x is the local mean of the simulated image x, μ y is the local mean of the measured image y, σ x is the standard deviation of the simulated image x, σ y is the standard deviation of the measured image y, σ xy is the cross-covariance of images x and y, C1=(0.01×L x ) 2 , L x is the dynamic range of the simulated image x, C2=(0.03×L y ) 2 , L y To measure the dynamic range of image y; By evaluating the SSIM value between the standard size line width model database and the low coherence interference overfocus scanning optical microscopy image, the maximum SSIM value is used as the best matching target, and the line width of the corresponding model in the standard size line width model database is used as the line width of the sample to be tested.

Images