Measurement data analysis method

Abstract

A preferred embodiment relates to a measurement method used, for example, for recess analysis in semiconductor manufacturing, and includes steps of generating a sample image using atomic force microscope (AFM) data of a sample having an array of two-dimensional periodic features and calculating the periodicity of the features. The method identifies periodic peaks to determine a feature period and a lattice angle, constructs a lattice mask registered in the image, and performs an alignment calculation. The mask is offset to perform the alignment calculation in order to optimize the cost.

Description

Technical Field

[0001] [Cross - reference to Related Applications] This application claims priority under 35 USC § 1.119(e) to U.S. Provisional Patent Application No. 63 / 352,120, filed on June 14, 2022. The subject matter of the corresponding application is incorporated herein by reference in its entirety.

[0002] Preferred embodiments relate to the field of atomic force microscopy (AFM) data analysis. In particular, they relate to the analysis of spatial and topographical data of sample features, such as the detection of periodic features in a lattice. Preferred embodiments are particularly useful in measurements of high - throughput applications, such as, for example, the analysis of recesses in semiconductor manufacturing processes.

Background Art

[0003] A scanning probe microscope, such as an atomic force microscope (AFM), is a device in which a probe having a tip interacts with the surface of a sample with an appropriate force to characterize the surface down to the atomic scale. Generally, by introducing the probe onto the surface of the sample and providing a relative scanning movement between the tip and the sample, surface characteristic data can be collected in a specific region of the sample to generate a corresponding map of the sample.

[0004] Overall, such devices can generate relative movement between the probe and the sample while measuring the topography or other surface characteristics of a part of the sample, as described in Patent Document 1 by Hansma et al., Patent Document 2 by Elings et al., and Patent Document 3 by Elings et al.

[0005] In a general configuration, the probe is often coupled to a vibration actuator or drive used to drive the probe at or near the resonant frequency of the cantilever. As another approach, the deflection, torsion, or other movement of the cantilever can be measured. Generally, the probe is often a microfabricated cantilever with an integrated tip.

[0006] Generally, under the control of an SPM controller, an electronic signal is supplied from an AC signal source, and the actuator or scanner drives the probe to vibrate. The probe-sample interaction is generally controlled via the feedback of the controller. In particular, the actuator may be coupled to the scanner and the probe, or may be formed integrally with the cantilever of the probe as part of the operating cantilever / probe.

[0007] AFM can be designed to operate in various modes such as contact mode and vibration mode. The operation is performed by moving one of the sample or the probe assembly vertically relative to the surface of the sample in response to the deflection of the cantilever of the probe assembly when scanned across the surface of the sample. The scan is generally performed in an "x-y" plane that is at least generally parallel to the surface of the sample, and the vertical movement is performed in the "z" direction that is perpendicular to the x-y plane. Many samples have roughness, curvature, and tilt that deviate from a plane, and the expression "at least generally parallel" may be used. In such a manner, data regarding the vertical movement is stored and may be used to construct an image of the surface of the sample corresponding to the sample characteristics to be measured (e.g., surface shape). In TappingMode (registered trademark) (TappingMode (registered trademark) is a trademark of the applicant), which is one of the AFM operation modes, the relevant cantilever of the probe vibrates at or near the resonant frequency of the cantilever. The feedback loop attempts to keep the amplitude of this vibration constant and minimize the "tracking force", i.e., the force due to the tip / sample interaction. Alternative feedback arrangements keep the phase or vibration frequency constant. Similar to the contact mode, such feedback signals are collected, stored, and used as data for characterizing the sample. It is clarified that the abbreviations for "SPM" and specific types of SPM may be used in this document to refer to a microscope apparatus or related technology (e.g., "atomic force microscope"). In a recent improvement of the ubiquitous TappingMode (registered trademark) AFM called Peak Force Tapping (registered trademark) (PFT), discussed in Patent Document 4, Patent Document 5, and Patent Document 6, which are explicitly incorporated by reference herein, the feedback is based on the force measured at each vibration cycle (referred to as the transient probe-sample interaction force).

[0008] Regardless of the operating mode, the AFM can obtain atomic-level resolution on various insulating or conductive surfaces in air, liquid, or vacuum, etc., using a piezoelectric scanner, an optical lever deflection detector, and an ultra-small cantilever fabricated by photolithography. Due to its resolution and versatility, the AFM is used as an important measuring device in various fields from semiconductor manufacturing to biological research.

[0009] In this regard, the AFM can be used in automated applications including high-precision manufacturing processes such as semiconductor manufacturing. The AFM has been proven useful in the semiconductor field because it can measure nanoscale surface features (e.g., topography) with high resolution.

[0010] There are various analyses that can be performed on the collected AFM data to confirm various characteristics of a specific sample. Depending on the sample under study and the purpose of use, the area of interest and the characteristics of the analysis used can vary. For example, in semiconductor manufacturing and wafer bonding processes, there may be copper pads surrounded by dielectrics. In such cases, planarity between the copper pads and the surrounding dielectrics is required. There are thresholds regarding how much dents and protrusions between the copper pads and the dielectric boundaries are allowed. Also, for the dielectric itself, planarity is required across the entire sample. Also, planarity is required for the copper pads themselves.

[0011] Flattening across the entire sample is the goal of the flattening process. However, it has been proven that determining whether the resulting sample meets specific threshold criteria for semiconductor functionality is a time-consuming and costly process. Existing AFM analyses such as Depth, CFA, and FinFET are too specialized for the application and are not useful in the field. Using conventional techniques, after the user manually identifies the areas where features are expected to exist, each area needs to be analyzed individually to understand the topography and spatial characteristics. Such a time-consuming process is not practical in a mass production environment.

[0012] Taking in-depth analysis as an example, this is a histogram-based analysis that returns Z information from an AFM image. CFA is an analysis that analyzes rectangular features within a rectangular grid and compares its sub-features. FinFet analysis is designed to measure the fin height, gate height, and gate-to-fin height for each fin in an image by comparing the image with a CAD clip of the same area. Useful data can be obtained through such existing analyses, but due to the nature of the problem, it is considered that the user will spend a considerable amount of time and effort setting up mathematical formulas, so it is not practical as a mass-production solution. Such analyses are carried out more intensively when the number of pads and their absolute positions are unknown.

[0013] Therefore, there is a need for a method that can quickly obtain meaningful data for comparing the spatial position and topography with respect to the features of a specific sample with minimal user intervention.

Prior Art Documents

Patent Documents

[0014]

Patent Document 1

Patent Document 2

Patent Document 3

Patent Document 4

Patent Document 5

Patent Document 6

Summary of the Invention

[0015] Preferred embodiments can overcome the shortcomings of current AFM data analysis techniques by enabling autonomous measurement of multiple key metrics for various regions of interest (ROIs) within a single AFM image without user intervention to specify the location of such ROIs. Previously, when customers wanted to secure similar data, they had to process the data manually, which required a lot of man-hours. According to the present invention, a user can obtain meaningful data regarding the quality of a sample within minutes. Since photolithography is an evolving field, the data required for quality assurance is also evolving. The present invention helps meet the needs of users who want to confirm the quality of a sample quickly and with minimal human intervention.

[0016] The present invention is particularly useful for the analysis of new bonding pads in a hybrid bonding process used in semiconductor manufacturing. The automatic pad detection function creates an easy-to-use environment that does not require prior knowledge about the number of pads and the absolute positions of the pads in the AFM image. In many cases, there are more than 20 pads in an AFM image. In the current high-volume production environment, it has become practical to accurately place measurement regions of interest (ROIs) for all pads included in one image over a number of images.

[0017] The automatic pad detection function is accurate and repeatable, ensuring that subsequent measurement ROIs are reliably placed for each pad for reproducible measurements. Such measurement of process steps requires sub-nanometer accuracy, which is only possible if the measurement ROIs are accurately placed for all measurement executions. If the measurement ROI for a pad is misaligned, a region slightly misaligned with respect to the pad will be measured. Due to the pad-specific topography, such misalignment generates a dataset with low reproducibility.

[0018] The present invention overcomes such problems and the aforementioned drawbacks of existing AFM analysis by analyzing AFM acquired data using various algorithms and returning information regarding the quality of the sample. The present invention combines a lattice detection algorithm and a new lattice alignment technique. After applying all the detection and alignment steps, the discovered lattice is used to analyze the localized depth, dispersion, inclination, etc. of each feature. Next, the quality of the sample can be determined using the returned data. This measurement method can be used not only for the contact fold of the sample and other topographical and spatial features, but also for the analysis of any feature within the periodic lattice.

[0019] According to a preferred embodiment, a measurement method for analyzing an AFM image includes the step of calculating the periodicity of the AFM image using fast Fourier transform autocorrelation. Next, this method includes the step of searching radially outward from the center of the image to find the peak of periodicity. Next, this method includes the step of quantifying the circular shell of the peak of periodicity in order to obtain the possible lattice period and angle. Further, the image can be downsampled in order to speed up the cost calculation. Next, a lattice mask can be constructed using the previously acquired lattice period and angle. Next, by overlaying the lattice mask on the image, the algorithm can distinguish between feature pixels and background pixels. Also, user input parameters may be applied to the alignment calculation. From here, the method can vary according to the data requested by the user.

[0020] On one side, the standard deviation of the background pixels is calculated and the value is set as the cost. Next, the cost can be recalculated by applying the offset of the lattice mask overlay. The cost is calculated for each offset in the range of 1.2 periods in order to include all alignment options. Finally, the offset that provides the minimum cost can be searched for and set for the final lattice alignment. This embodiment of using the standard deviation is likely to be used when the background is rough.

[0021] In a further aspect, the median between the background pixels and the feature pixels can be calculated and set as the cost. Next, an offset is applied to the grid mask overlay to recalculate the cost. The cost is calculated for each offset in the range of 1.2 periods to cover all alignment options. Finally, the offset that provides the maximum cost is searched for and set as the final grid alignment. In the above embodiments, the median is likely to be preferred when the background is smooth.

[0022] According to another aspect of the preferred embodiment, the method includes the steps of repeatedly calculating for at least two types of 2D model types including square, rectangle, hexagon, and oblique, and selecting the periodicity of the grid type that generates the smallest deviation between the model grid type and the acquired data.

[0023] In a further aspect of the preferred embodiment, the method further includes applying an adaptive flattening algorithm to the sample image.

[0024] In other embodiments, the measurement method includes generating an image of the sample using atomic force microscope (AFM) data and calculating the periodicity of the features of the image. Next, the method searches for at least one peak in the periodicity and obtains the feature period and the grid angle. Then, a grid mask template is constructed using the feature period and the grid angle, and the grid mask template is overlaid on the image. Then, an alignment calculation is performed to determine the cost, and an offset of the grid mask template is applied to the image to recalculate the cost. The application and recalculation steps are repeated to determine the alignment between the grid mask template and the image.

[0025] In other embodiments, an AFM for collecting sample data includes a probe that interacts with the surface of the sample and a controller that controls the probe-sample interaction to collect atomic force microscope (AFM) data of the sample having periodic features. The controller generates a sample image including feature pixels and background pixels using the AFM data and calculates the periodicity of the features. Further, the controller identifies peaks in the periodicity to determine the feature period and the lattice angle, and constructs a lattice mask template using the feature period and the lattice angle. Next, the lattice mask template is overlaid on the image, and the controller performs an alignment calculation to determine a cost. An offset of the lattice mask template is applied to the image and the cost is recalculated. The application and recalculation steps are repeated to determine the alignment between the lattice mask template and the image, which is particularly important in semiconductor manufacturing.

[0026] According to a further aspect of this embodiment, the controller performs an alignment step by at least one of: a) calculating a standard deviation of the background pixels and setting the standard deviation as a cost value; and b) calculating a median of the background pixels and the feature pixels and setting the median as a cost value. The controller can determine an offset of the lattice at which the cost value is minimized when the standard deviation is calculated, and can determine an offset of the lattice at which the cost value is maximized when the median is calculated.

[0027] Such features and other advantages of the present invention will become apparent to those skilled in the art from the following detailed description and the accompanying drawings. However, it should be understood that the detailed description and the specific embodiments are illustrative for the purpose of describing the preferred embodiments of the present invention and are not to be construed as limiting the present invention. Many changes and modifications can be made within the scope not departing from the spirit of the present invention, and it is clear that the present invention includes all such modifications.

Brief Description of the Drawings

[0028] Preferred embodiments of the present invention are shown in the accompanying drawings, where reference numerals may be used to refer to the same parts throughout.

Figure 1

Figure 2A

Figure 2B

Figure 2C

Figure 2D

Figure 3

Figure 4A

Figure 4B

Figure 4C

Figure 4D

Figure 4E

Figure 5

[0029] [Detailed Description of Preferred Embodiments] A preferred embodiment relates to a measurement method for analyzing spatial and topographical data of 2D elements / features within a lattice from raw atomic force microscope (AFM) data. The method described herein combines a lattice detection algorithm with a new lattice alignment technique. After applying all the detection and alignment steps, the acquired lattice can be used to analyze local depths, dispersions, inclinations, etc. of each feature. This invention helps meet the requirement of quickly verifying the quality of a sample with minimal user intervention.

[0030] First, referring to FIG. 1, a scanning probe microscope apparatus 150 (e.g., AFM) according to a preferred embodiment is illustrated. In this embodiment, a probe 152 including a tip 154 extending from the distal end of a cantilever 155 is fixed by a probe holder (not shown) supported by a piezoelectric tube scanner 156. The scanner 156 may be a “Z” or vertical scanner that reacts to sample characteristics in a closed-loop control system and can position the tip 154 relative to the sample 158 during AFM imaging. The tube scanner 156 is coupled to an XY scanner 160 (preferably, a piezoelectric tube) used to raster the probe tip 154 over the surface of the sample 158 during operation of the AFM. In particular, a scanned sample can also be used instead. For example, a mechanical Z stage 162 can be provided to provide a large Z movement between the tip 154 and the sample 158 while starting AFM image acquisition to connect the tip 154 and the sample 158.

[0031] Sample 158 is attached to an XY stage 164 that provides a coarse XY motion for positioning the probe 152 primarily in the region of interest of sample 158. The XY stage controller 166 controls the stage 164 to position the probe / sample in the relevant region of interest. However, the stage 164 may be configured to provide a relative scanning motion (e.g., raster) between the chip 154 and the sample 158 at a selected scanning speed. The controller 166 also arranges for an image scan in the region of interest in response to the AFM controller 174. The controllers 166, 174 are operated by a computer 180.

[0032] During operation, after the chip 154 is engaged with the sample 158, a high-speed scan of the sample is initiated with the XY scanner 160 in the AFM operating mode (e.g., PFT mode) as described above. As the chip 154 interacts with the surface of the sample 158, the probe 152 deflects, and this deflection is measured by the optical beam bounce deflection detection device 168. The corresponding device 168 includes a laser 170 that irradiates the beam L towards the photodetector 172 behind the cantilever 155 for high-speed processing of the deflection signal, and the photodetector transmits the deflection signal, for example, to the DSP 176 of the AFM controller 174 to process the deflection signal at high speed.

[0033] The AFM controller 174 continuously determines a control signal according to the AFM operating mode and transmits the signal to the piezoelectric tube scanner 156 to maintain the Z position of the probe 152 relative to the sample 158, and more specifically, to maintain the deflection of the probe at the feedback set point.

[0034] Referring to FIGS. 2A to 2D, a series of shapes showing the progress of the analysis of the AFM data according to this measurement method can be confirmed. In FIG. 2A, the raw AFM data is shown in image 200. This raw AFM data is generated by the scanning probe microscope apparatus and method described above. The data includes features 202 and background 204. Next, FIG. 2B is a diagram 206 showing the step of detecting periodicity in the raw AFM data. Here, it is preferable to search for peaks and periodicity from the raw AFM data through fast Fourier transform (FFT) autocorrelation. The image is correlated with the image itself, and the peak indicates a point 208 that is symmetric with the image itself. Also, FIG. 2C shows a periodicity ring 210 discovered from the data. Each point 208 shown in FIG. 2C indicates a peak of periodicity. The distance positions of these peaks with respect to the center and with respect to each other, and the two-dimensional angle (2D-angle) are quantified and used to form different lattices from each other (as will be described in detail below). FIG. 2D shows a schematic diagram 212 of a lattice that can be generated for use as a mask to extract sample information in the region of feature 202 adjacent to background 204.

[0035] Referring now to FIG. 3, a simplified flowchart of the measurement method 300 of the present invention is illustrated. At step 302, raw AFM surface data is collected from a sample. Next, at step 304, the periodicity of the image is calculated using a fast Fourier transform (FFT). At step 306, a search is made radially outward from the center of the image to find a periodic ring (see FIG. 2C). In the case of a hexagonal lattice, four sets may be used to find the ring. When using different lattices (rectangular, triangular, octagonal, etc.), other sets can be searched. Such a periodic ring provides information regarding the relative peak position of the distance from the center, and also provides information regarding the distance and angle of the peaks relative to each other. Also, a plurality of periodic rings are discovered while moving outward from the center. In the next step 308, a circular shell of the periodic peaks is quantified to obtain a possible lattice period and lattice angle. At step 310, the image is downsampled for faster cost calculation. In the next step 312, a lattice mask is constructed using the previously obtained lattice period and angle. A plurality of lattices as shown in FIG. 2D are generated.

[0036] At step 314, the lattice mask is overlaid on the image, enabling the algorithm to distinguish between feature pixels and background pixels. Here, the mask matrix is added / multiplied with the image matrix to extract the feature pixels. The mask 212 (FIG. 2D) divides the image into pixels of black and white regions. The white region pixels indicate sample features, and the black region pixels indicate the background. In the next step 316, user input parameters for alignment calculation are applied. The user may select one of standard deviation calculation or median calculation. The standard deviation is likely to be applied when the background is rough, and the median is likely to be applied when the background is smooth.

[0037] The next step changes according to the parameter selected by the user. When the user selects the standard deviation, the standard deviation of the background pixels (black regions) is calculated, and the corresponding standard deviation value is set as the cost in step 318. Next, in step 322, the offset of the grid mask overlay is applied and the cost is recalculated. The cost is preferably calculated in the range of 1.2 cycles for each offset so as to cover all alignment options. Since this is a brute-force search for the area of one unit cell, all possible offsets can be tested. Finally, in step 324, the offset that provides the minimum cost is searched for and set as the final grid alignment.

[0038] This method changes when the user selects the median as the input parameter. In this case, the next step after step 316 is step 320, and the difference in the median between the background pixels (black regions) and the feature pixels (white regions) shown in FIG. 2D is calculated. The difference in the median between the background pixel 204 and the feature pixel 202 is calculated and set as the cost. Next, in step 322, the offset of the grid mask overlay is applied to recalculate the cost in the same way as in the case of the standard deviation. The cost is calculated for each offset in the range of 1.2 cycles in order to cover all alignment options. Since this is a brute-force search for the area of one unit cell, all possible offsets are tested. Finally, in step 326, the offset that provides the maximum cost value is searched for and set as the final grid alignment.

[0039] If the cost is calculated appropriately, the final grid alignment is determined and the design of the feature is set. For example, if it is confirmed that the feature is an array of a series of concentric rectangles, pixels corresponding to the entire area of the rectangle can be extracted from the AFM image, and specific pixels corresponding to specific parts of the feature can be selectively analyzed.

[0040] When the 2D lattice type is unknown, it is clarified that the periodicity of the lattice type with the smallest deviation between the model lattice and the acquired data can be selected by repeating possible mode 2D lattice types such as square, rectangle, hexagon, oblique (see https: / / en.wikipedia.org / wiki / Bravais_lattice). Here, the smallest deviation corresponds to the least alignment cost.

[0041] Next, referring to FIGS. 4A to 4E, a series of images showing the process of analyzing the AFM data according to the present measurement method are illustrated. FIG. 4A shows the raw AFM data. The image 400 shown in FIG. 4A is generated in step 302 of the measurement method described above. The data includes features 402 and background 404. Defects 406 may be present in the sample and the result data. Next, FIG. 4B is an AFM data image after applying "adaptive flatten" to remove the tilt of the image. FIG. 4C is an image showing the periodicity map obtained by fast Fourier transform autocorrelation. The point 408 of the image indicates the peak of periodicity. FIG. 4C corresponds to step 304 of the measurement method described above. FIG. 4D is an image showing a ring of periodicity 410. FIG. 4D corresponds to step 306 of the measurement method described above. Further, FIG. 4E is an image showing the lattice mask template 412 generated using the distribution of the peaks. This mask is generated in step 312 above. Next, the corresponding lattice mask template 412 is overlaid on the AFM image. In this process, since the mask matrix is multiplied by the image matrix, only specific pixels are analyzed. This corresponds to step 314 described above.

[0042] Referring to FIG. 5, a hexagonal lattice 500 showing features of interest and differences in height therebetween is illustrated. Here, a hexagonal lattice rather than a square lattice is shown. The dark rectangle 502 indicates the position of the feature of interest 502 (e.g., 402 in FIG. 4A) found by the present method. Once the feature of interest 502 is identified, it can be analyzed to quantify the height distribution among the features 502. The shading 504 on the feature 502 indicates the difference in height between the features 502. FIG. 5 shows the results obtained in step 324 or step 326 described above. The unshaded rectangular portion in the center of FIG. 5 indicates that the corresponding portion of the sample lacks features or may hardly be printed on the wafer. When the wafer is printed, since it is always periodic, theoretically, the feature 502 should exist in all squares of the image. This information regarding the localized depth, variance, slope height, etc. of each feature 502 is very important for the quality and function of the sample.

[0043] The preferred embodiments are particularly useful in semiconductor manufacturing. For example, the recess analysis can perform crucial measurements in the IC manufacturing process of bonding two semiconductor wafers having patterned surfaces to each other. Such bonding between wafers requires very precise topographical information of the wafer surface of post - chemical - mechanical polishing (CMP) including metal pads surrounded by dielectric materials. To enhance the effect of bonding, the surface must be extremely flat. The recess analysis calculates the difference in height (dishing) of the metal pads with respect to the surrounding dielectrics, the local slope of the dielectric material adjacent to the metal pads, and the global flatness across the entire field of view.

[0044] IC manufacturers can make important process decisions based on the ratio of out - of - specification roughness and slope regions through the recess analysis results.

[0045] While the best mode contemplated by the inventors for carrying out the present invention has been disclosed above, the embodiments of the present invention are not limited thereto. It will be apparent that various additional, modifications, and rearrangements of the features of the present invention can be made without departing from the spirit and scope of the basic inventive concept.

Claims

1. A measurement method comprising: generating a sample image including feature pixels and background pixels using atomic force microscope (AFM) data of a sample including an array of periodic features; calculating the periodicity of the features; identifying peaks of the periodicity to determine a feature period and a lattice angle; constructing a lattice mask template using the feature period and the lattice angle; overlaying the lattice mask template onto the image; performing an alignment calculation to determine a cost; applying an offset of the lattice mask template to the image and recalculating the cost; repeating the applying and recalculating steps to determine an alignment between the lattice mask template and the image; A measurement method comprising the above steps.

2. The step of performing the alignment calculation includes at least one of: a) calculating a standard deviation of the background pixels and setting the standard deviation as a cost value; b) calculating a median value of the background pixels and the feature pixels and setting the median value as a cost value. The measurement method according to Claim 1.

3. The measurement method according to Claim 2, further comprising determining an offset of the lattice at which the cost value is minimized when calculating the standard deviation, and determining an offset of the lattice at which the cost value is maximized when calculating the median value.

4. The measurement method according to Claim 1, further comprising extracting data regarding the features after applying the alignment.

5. The data corresponds to at least one of feature characteristics including height, depth, shape, uniformity, dispersion, and inclination. The measurement method according to Claim 4.

6. The measurement method according to Claim 5, further comprising comparing at least one feature characteristic with a known model to determine the feature characteristic.

7. The comparing step is used for recess analysis in semiconductor manufacturing. The measurement method according to Claim 6.

8. The features are two-dimensional periodic features, and the step of identifying peaks of the periodicity proceeds in a radial outward direction starting from the center of the sample image. The measurement method according to Claim 1.

9. Repeatedly calculating for at least two two-dimensional model types including a square, a rectangle, a hexagon, and a hypotenuse. Selecting the periodicity of the lattice type that generates the smallest deviation between the model lattice type and the acquired data; The measurement method according to claim 8, further comprising.

10. The measurement method according to claim 1, wherein the step of calculating the periodicity is performed using a fast Fourier transform algorithm.

11. The measurement method according to claim 1, wherein the lattice mask template is hexagonal.

12. The measurement method according to claim 1, further comprising applying an adaptive flattening algorithm to the sample image.

13. A measurement method, comprising: Generating a sample image using atomic force microscope (AFM) data; Calculating the periodicity of the features of the image; Searching for at least one peak with the periodicity; Determining the characteristic period and the lattice angle; Constructing a lattice mask template using the characteristic period and the lattice angle; Overlaying the lattice mask template on the image; Performing an alignment calculation to determine the cost; Applying the offset of the lattice mask template to the image and recalculating the cost; Repeating the step of applying and recalculating the cost to determine the alignment between the lattice mask template and the image; A measurement method comprising.

14. The measurement method according to claim 13, wherein the cost is calculated with respect to the total area of one unit cell.

15. The measurement method according to claim 13, further comprising downsampling the image to calculate the cost at high speed.

16. The measurement method according to claim 13, wherein the step of searching for at least one peak with the periodicity proceeds in a radial outward direction starting from the center of the sample image.

17. The measurement method according to claim 13, wherein the step of calculating the periodicity is performed using a fast Fourier transform (FFT) algorithm.

18. An AFM for collecting data of a sample atomic force microscope (AFM), comprising: A probe that interacts with the surface of the sample; A controller that controls the probe-sample interaction and collects atomic force microscope (AFM) data of a sample having an array of periodic features; Including, The controller is Generate a sample image with feature pixels and background pixels using AFM data, calculate the periodicity of the features, identify the peaks of periodicity to determine the feature period and the lattice angle, construct a lattice mask template using the feature period and the lattice angle, overlay the lattice mask template on the image, perform an alignment calculation to determine the cost, apply the offset of the lattice mask template to the image and recalculate the cost, Repeat the application and recalculation to determine the alignment between the lattice mask template and the image, AFM.

19. The controller performs the alignment step by at least one of: a) calculating the standard deviation of the background pixels and setting the standard deviation as a cost value; b) calculating the median of the background pixels and the feature pixels and setting the median as a cost value. The AFM according to claim 18.

20. The controller according to claim 19, wherein when calculating the standard deviation, it determines the offset of the lattice where the cost value is minimized, and when calculating the median, it determines the offset of the lattice where the cost value is maximized.

Images