CN115020174A

CN115020174A - Method for measuring and monitoring actual pixel size of charged particle beam scanning imaging equipment

Info

Publication number: CN115020174A
Application number: CN202210677384.5A
Authority: CN
Inventors: 刘骊松
Original assignee: Shanghai Precision Measurement Semiconductor Technology Inc
Current assignee: Shanghai Precision Measurement Semiconductor Technology Inc
Priority date: 2022-06-15
Filing date: 2022-06-15
Publication date: 2022-09-06

Abstract

The invention discloses a method for measuring and monitoring the actual pixel size of charged particle beam scanning imaging equipment, which is used for a patterned wafer. The measuring method comprises the following steps: carrying out offline comprehensive measurement FM on the patterned wafer by using charged particle beam scanning imaging equipment to obtain the actual pixel size of each pixel to be measured; the QM is partially measured based on the comprehensive FM method and is used for measuring the actual pixel size of partial pixels to be measured on line by the equipment, and the speed is high and the time cost is low when the equipment does not occupy the machine; the monitoring method comprises a closed-loop process, which comprises an off-line comprehensive measurement FM method and an on-line partial measurement QM method, different measurement frequencies are provided, and an automatic adjusting and optimizing mechanism of the measurement frequencies can more accurately grasp the time for implementing the actual pixel size comprehensive measurement FM. The problem that measurement is not timely so that work of the equipment is affected can be avoided, and waste of machine time can be avoided.

Description

Method for measuring and monitoring actual pixel size of charged particle beam scanning imaging equipment

Technical Field

The invention relates to the technical field of semiconductor equipment, in particular to a method for measuring and monitoring the actual pixel size of charged particle beam scanning imaging equipment.

Background

In the front end of semiconductor lsi manufacturing process, an Electron beam Scanning and imaging device, such as a Scanning Electron Microscope (SEM) based device, is often used to detect defects or review semiconductor wafers (wafers). Such equipment includes electron Beam Inspection Equipment (EBI) or electron Beam Review Equipment (EBR), and Critical Dimension (CD) measurement equipment (CD-SEM), which are in principle electron Beam scanning imaging systems. The EBR apparatus is described below by way of example, but EBI and CD-SEM are not excluded. In the application of the equipment to the semiconductor wafer, an accurate method is needed for acquiring the imaged actual pixel size of the equipment, so as to accurately estimate the actual physical size of an object to be measured, such as a defect or the actual physical size of a morphology.

Referring to fig. 1, the EBR apparatus 100 generally includes a mechanical motion stage 110, which is movable and rotatable in X, Y, and Z directions, and on which an electrostatic Chuck (E-Chuck) 120 on which a Wafer (Wafer)111 is placed is provided. EBR equipment also typically includes an Optical microscopy imaging system (OM) 130, which has a lower magnification but a larger Field of View (FOV), and is typically used to assist in tasks such as primary wafer alignment. The EBR apparatus further includes a core task component 140, that is, an electron optical system 140, that is, the scanning electron microscope and SEM system, including a lens barrel, in which functional components such as electron emission, focusing, beam limiting, scanning, and biasing are included, and also including a circuit part for collecting and signal amplifying processing of the emitted electrons (mainly secondary electrons) on the surface of the wafer. The EBR apparatus further comprises a computer 150 on which software including a database is run that can be used to process and store data and display images.

Referring to fig. 2A and 2B, the electron beam 210 in the SEM system of the EBR apparatus is imaged with a focal plane 211, a focal depth 212, a z-direction which is the density of the electron beam 210, and an x-direction which is an x-coordinate direction on the wafer. In operation, the SEM system scans back and forth over a given area, with the sampling intervals Δ X and Δ Y in the X, Y directions over the wafer surface (and within its depth/range of focus), and the system needs to stay at each sampling point for a certain amount of time to accumulate enough emitted electrons (to achieve a certain signal-to-noise ratio). The obtained SEM image is theoretically roughly equivalent to the convolution of the beam spot shape of the electron beam (approximately gaussian distribution) and the wafer surface topography (including different materials and structures) when reaching the wafer surface. When the theoretical field of view/FOV of the SEM system (in most applications, the X and Y directions are set to be the same FOV, i.e., the foxx and FOVy are set to be the same FOV, but actually the foxx and FOVy are different, i.e., the Px and Py are different, and if the difference is too large, image deformation and other problems may be caused) and the number of pixels in the X and Y directions of the image are determined, the X and Y directions Pixel Size, i.e., the theoretical Pixel Size (NPS), is determined, and the theoretical Pixel Size is the theoretical field of view Size divided by the image Size (e.g., the image Size is 1024/2048 pixels wide/high), so the theoretical Pixel Size is Δ X and Δ Y. In actual use, operating parameters and conditions of SEM systems often drift, for example, due to 1) minor changes in the actual beam spot size of the system or the scan sampling interval as determined by the device control circuitry; 2) the theoretical pixel size and the actual pixel size are different in SEM systems due to changes in the working distance of the objective lens to the wafer surface (due to, for example, drift in the Z-direction of the mechanical motion stage or due to different types of wafer thickness changes, thus changing the actual sampling interval on the wafer). Thus, the theoretical pixel size and the actual pixel size are greatly different due to the superposition of various factors, so that the actual pixel size with higher accuracy is often required in practical application of the device.

Referring to fig. 3A and 3B, a method 1 for obtaining an actual pixel size commonly used in the prior art includes: a Patterned Wafer or sample comprising known dimensions is imaged and the actual pixel dimensions are then calculated. In method 1, a specially prepared sample 311 is typically placed on an electrostatic tray of a mechanical motion stage 310, and an image 313 taken of the sample 311 contains an object 312 of known size and known length, e.g., 5 μm. However, the defect is obvious, the thickness of the sample 311 cannot be guaranteed to be close to the wafer in actual work, so that the distance from the surface of the sample 311 to the lens barrel in the SEM system, namely the working distance, is different from the distance from the surface of the actual customer wafer to the lens barrel in the SEM system, which causes a large error, and in addition, the local material of the specially-made sample 311 is easily damaged after long-term multiple scanning.

A commonly used method 2 for obtaining an actual pixel size in the prior art is to use a special wafer with a graphic/object of a known size, but the disadvantages of the method are obvious, firstly, the cost of the special wafer is increased, the thickness of the special wafer cannot be ensured to be close to different wafers of customers in actual work, and the distance from the surface of the special wafer to a lens barrel in an SEM system, namely the working distance, is different from the distance from the surface of the actual wafer to the lens barrel in the SEM system; and more importantly, the measurement is non-online, and the sheet is required to be taken up and down during each use, so that the measurement is very inconvenient when a valuable machine is occupied, and is difficult to realize on an actual production line.

Therefore, there is a method 3 in the industry, that is, during the operation of the EBR apparatus, there is a method of directly using the customer Wafer (limited to the Patterned Wafer) to obtain the actual pixel size, which seems more convenient, and there is no problem of the height/operation distance difference in the above methods 1 and 2, and the procedure is roughly as follows:

fig. 4 shows a prior art method for determining a Matching position (Xm ', Ym') by acquiring images (two frames of images) before and after a wafer movement and performing template Matching ((PM) method for measuring an actual pixel size (not shown in the figure) wherein the coordinate axes of the mechanical motion stage coordinate system are X-axis and Y-axis, and a view field 402 is shown when the wafer 401 and the SEM system acquire the images, and the position of the view field 402 is fixed, the principle is that a first frame of image, i.e. a template image, is acquired first, the image therein is selected as a template 403, the template 403 is located under the image coordinate system (Xm, Ym), the displacements of the mechanical motion stage in X and Y directions are dXs and dYs, respectively, dXs and dYs are marked as 404 and 405 in the figure, a second frame of image is acquired as a target image, and template Matching is performed to determine the Matching position (Xm ', Ym'), the matching position (Xm ', Ym') is also a coordinate in the image coordinate system. The origin of the coordinates of the image coordinate system is, for example, the upper left corner, the lower left corner, or the center point of the image, but is not limited thereto. The actual pixel size is obtained by dividing dXs by (Xm '-Xm) and dYs by (Ym' -Ym).

A more detailed description of the method is given below, with reference to fig. 5A, the steps in method 3 are as follows:

1) determining a certain position on the wafer, collecting template images 510, and selecting one template 511 from the template images 510;

2) moving the mechanical moving platform, enabling the wafer and the mechanical moving platform to synchronously move, acquiring a target image 512 at another position, and recording relative displacement dXs and dYs of the mechanical moving platform; the moving position is limited to ensure that the target corresponding to the template is still in the target image (i.e. located in the field of view of the SEM system) under the condition that the mechanical motion platform has a known error range;

3) searching a template in the target image, namely performing template matching to obtain a matching position 513;

4) the distance dXm, dYm, i.e. dXm and dYm between the template 511 and the matching position 513 is obtained as the target displacement in the x direction and the y direction, specifically:

dXm＝Xm′-Xm

dYm＝Ym′-Ym

wherein, (Xm, Ym) is the position of the template in the template image (initial position), (Xm ', Ym') is the position in the target image where the template is matched (matching position);

5) obtaining actual pixel sizes Px and Py, specifically:

the actual displacements of the mechanical motion platform in the x direction and the y direction are dXs and dYs, that is, the coordinate position of the mechanical motion platform changes to (dXs, dYs). For example, a precision code reader/laser interferometer typically provided on the mechanical motion stage may give a more accurate actual position of the mechanical motion stage, which may differ slightly from the position the system commands the mechanical motion stage to reach, e.g., typically to within 0.5 μm.

Referring to fig. 5A, if the LMSEM image (hereinafter abbreviated as LM image, for example, the image with theoretical pixel size above 100nm may be regarded as LM) has a higher magnification, the method is as described above, i.e., the template image 520 is collected, the template 521 is selected from the template image, the wafer collection target image 522 is moved, and the template matching is performed to obtain the matching position 523. Referring to fig. 5B, the same method is applied to the HM SEM image (hereinafter abbreviated as HM image), and the template image 520 is collected, the template 521 is selected, and then the wafer collection target image 522 is moved to perform template matching to obtain the matching position 523. It can be seen that the method 3) seems to be an improvement over the methods 1) and 2), but the method still has the following problems that a user collects template images somewhere on the wafer to select a template for single template matching, the single stroke moving distance of a mechanical motion platform (Stage) is limited, and the relative error of the measurement result is large.

In addition to the above problems of accuracy, speed, and use cost of the actual pixel size measuring methods 1, 2, and 3, there is another more serious problem. That is, the actual pixel size is measured in the above-mentioned devices at irregular time in the industry, and there is a problem that the measurement timing is not proper: a. on the one hand, it may sometimes not be necessary to measure the actual pixel sizes because they do not change significantly, and unnecessary measurements take up equipment time and waste equipment resources because equipment on an IC production line is at a premium; b. on the other hand, measurement may be not timely, so that the actual pixel size drift/deviation causes inaccuracy of the work result of the semiconductor device, for example, the defect size of EBR/EBI or the IC feature size of CD-SEM is inaccurate, thereby reducing/damaging the device performance and causing serious consequences such as product quality reduction. These problems are caused by the lack of effective monitoring of the actual pixel size to grasp the timing of the actual pixel measurement, which ultimately leads to significant deficiencies in the prior art in terms of accuracy, reliability, and time cost, and thus needs to be greatly improved.

Disclosure of Invention

The invention aims to provide a method for measuring the actual pixel size of a charged particle beam scanning imaging device so as to solve the problem of high time cost of the measurement method in the prior art; another object of the present invention is to provide a method for monitoring the actual pixel size of a charged particle beam scanning imaging device, so as to accurately grasp the timing of the actual pixel size measurement, without delaying the timing of the actual pixel size measurement and wasting valuable time and resources.

To achieve the above object, an aspect of the present invention provides a method for measuring an actual pixel size of a charged particle beam scanning imaging device, including: carrying out offline comprehensive measurement FM on the patterned wafer by using charged particle beam scanning imaging equipment to obtain the actual pixel size of each pixel to be measured; and performing partial measurement QM based on the comprehensive measurement FM method, wherein the partial measurement QM is used for the equipment to measure the actual pixel size of partial pixels to be measured on line.

In order to achieve the above object, another aspect of the present invention provides a method for monitoring an actual pixel size of a charged particle beam scanning imaging device, including:

1411. after an electronic optical system in the equipment is maintained and calibrated, determining monitoring working parameters, wherein the monitoring working parameters comprise a second parameter Nf, a second threshold value Thfi, a first parameter Nq and a first threshold value Thqi, the second parameter Nf is greater than the first parameter Nq, and the second threshold value Thfi is greater than the first threshold value Thqi;

1412. executing the overall measurement FM, and simultaneously clearing the second counter nf and the first counter nq;

1413. comparing the actual pixel size measured in the comprehensive measurement FM with a second threshold value Thfi, returning to the step 1411 when all the actual pixel sizes measured in this time are outside the second threshold value Thfi, and otherwise, entering the next step;

1414. saving all actual pixel sizes measured in the current comprehensive measurement FM;

1415. when the equipment starts or continues to execute the work of the equipment, counting through the second counter nf and the first counter nq;

1416. when the first counter nq reaches the second parameter Nf, returning to step 1412, otherwise, entering step 1417;

1417, when the first counter Nq does not reach the first parameter Nq, returning to the step 1415, and enabling the equipment to perform the job as usual; otherwise go to step 1418;

1418. performing a QM based on said partial measurement while clearing the first counter nq;

1419. when the result of the actual pixel size is greater than or equal to a second threshold value Thfi, returning to the step 1411 to perform equipment calibration maintenance again; otherwise, when the monitored value is smaller than a first threshold value Thqi, returning to the step 1415, and continuing to execute the job work of the equipment; otherwise, when the monitored value is smaller than the second threshold value Thfi but greater than or equal to the first threshold value Thqi, the first parameter Nq and the second parameter Nf are changed, and then the process returns to step 1415.

In the method for measuring the QM partially according to the present invention, the actual pixel size of the partial pixel to be measured is measured online by using the device while the device is performing the job, and the method is fast and can be used for obtaining the actual pixel size of the partial pixel to be measured under the current condition without occupying space.

Another aspect of the present invention provides a monitoring method, which can accurately determine the timing of measuring the actual pixel size, without delaying the timing of measuring the actual pixel size and wasting valuable time and resources.

Drawings

FIG. 1 is a schematic diagram of relevant portions of an EBR apparatus of the prior art;

FIG. 2A is a schematic diagram of a prior art electron beam density profile in an SEM system;

FIG. 2B is a schematic illustration of the focal length and depth of focus of an electron beam in the prior art;

FIG. 3A is a schematic representation of a prior art custom sample placed on a mechanically moving platform;

FIG. 3B is a schematic representation of a prior art SEM image taken from a tailored sample;

FIG. 4 is a schematic diagram of a prior art method for obtaining an actual pixel size of an image;

FIG. 5A is a diagram illustrating template matching in a target image using a template at a low magnification LM according to the prior art;

FIG. 5B is a diagram illustrating template matching in a target image using a template at a high magnification HM according to the prior art;

FIG. 6A is a diagram illustrating a wafer alignment method according to the prior art;

FIG. 6B is a schematic diagram of another wafer alignment method in the prior art;

FIG. 7 is a schematic diagram of generating a template for measuring an actual pixel size in a template image under different theoretical magnifications corresponding to different pixels to be measured according to an embodiment of the present invention;

FIG. 8 is a diagram illustrating a multi-template chain-weighted template matching method in an embodiment of the present invention;

FIG. 9 is a schematic diagram of a template selection method in an image at high magnification HM according to an embodiment of the present invention;

FIG. 10A is a diagram illustrating a template selection method for partial QM measurement at low magnification LM according to an embodiment of the present invention;

FIG. 10B is a diagram illustrating a template selection method for partial QM measurement at high magnification HM according to an embodiment of the present invention;

FIG. 11 is a diagram of a template selection in a single frame image for partial measurement QM at high magnification HM;

fig. 12 is a schematic diagram of a method for acquiring a neighborhood 2 image using electron-optical system offset for partially measuring QM at actual pixel size measurement at a high magnification HM in an embodiment of the present invention;

FIG. 13 is a schematic illustration of an embodiment of the invention for acquiring different adjacent 2 images with electron optical system offset during actual pixel size measurement at high magnification HM;

FIG. 14 is a flowchart of an actual pixel size monitoring method in a charged particle beam scanning imaging device in an embodiment of the present invention;

fig. 15 is a flowchart of a method for measuring an actual pixel size in a charged particle beam scanning imaging device in an embodiment of the present invention.

Detailed Description

The technical solution in the embodiments of the present invention is clearly and completely described below with reference to the accompanying drawings. The examples are given solely for the purpose of illustration and are not intended to limit the scope of the invention.

A description will first be given regarding background knowledge.

Most of the work of the charged particle beam scanning imaging equipment on the IC production line relates to a Patterned Wafer (Patterned Wafer), and the measuring method and the monitoring method in the embodiment of the invention are also directed to the Patterned Wafer.

Taking a charged particle beam scanning imaging apparatus as an EBR apparatus for example, the EBR apparatus operation includes two main steps as with most semiconductor apparatuses. Firstly, a user establishes a work menu (Recipe) in advance, wherein the work menu comprises all steps of main tasks of equipment; and then, the equipment automatically and repeatedly executes the working menu on the same type of wafer during normal working.

Further assume that the above-described apparatus is in a normal state at the time of operation for determining the actual pixel size.

In practical use, the above-mentioned device usually has a series of pixels to be measured each time, for example, 6-7 pixels in total, the theoretical pixel size is from 500nm to 2nm, and the common practice is to measure the pixels in sequence from the theoretical pixel size large/View Field (FOV) large/small magnification ratio to the theoretical pixel size small/small View Field/large magnification ratio.

In this embodiment, the range of the magnification is determined according to the empirical value of the theoretical pixel size, for example, 500nm to 50nm may be regarded as the low magnification LM, and the magnification corresponding to 50nm and down to 2nm may be regarded as the high magnification HM, which may be adjusted according to the image situation in practice.

In the embodiments of the present invention, the known information (referred to hereinafter) included in the Wafer Alignment (WA) is utilized as much as possible, because WA is very important to the job of the equipment and is a prerequisite for its normal operation. It is therefore necessary here to first supplement some WA background. In general, in order to ensure the accuracy of the work of the semiconductor device, the WA is usually 3-grade (in few cases, 2-grade or 4-grade), after the wafer is mounted, the primary WA is usually an Optical image of an Optical Microscope (OM) with a larger Field of View (FOV), and the subsequent WA is an SEM image with a gradually larger magnification (FOV). In creating the wafer alignment Recipe, a preliminary WA is performed starting from a 1 or more level OM image at low to high magnification, and a successful post-transition to a SEM image for a 1 or more level WA (typically 2 levels, low magnification LM and high magnification HM) is performed so that the wafer orientation and position determination meets the required accuracy requirements. The template Matching (PM) method is the same as that in the background, and NCC or characteristic-based template Matching is also used, so that the similarity is between [0, 1 ]. The template is also selected as 511, 521. At each stage WA, referring to fig. 6A as an example, images are collected along the same row (or column) of multiple Die positions to perform template matching, so as to obtain matching

points

611, 612, 613, 614, 615, 616 in the figure, a point fitting straight line in which matching is successful (the similarity reaches a predetermined threshold, usually above 0.65) is taken to obtain a wafer orientation angle θ, and the wafer orientation angle θ is corrected (e.g., by rotating the mechanical motion stage) and then shifted to a higher stage wafer alignment until all stages WA are completed. And maintaining the successful matching position when the WA Recipe is created, and automatically executing the WA Recipe after the wafer is mounted when the Recipe is executed. Referring to fig. 6B, there are WA's with 4-5 points in the X and Y directions, i.e., matching points 620, and WA (including with OM images and with SEM images) at each level is thus performed until WA at the last level is completed.

In the embodiment of the present invention, two methods of Full Measurement (FM) and fast Measurement (QM) of an actual pixel size are distinguished, where the fast Measurement QM is also referred to as a partial Measurement QM and generally increases from a small amplification factor to a small amplification factor. The FM can provide high-precision measurement results, namely off-line (offline) measurement, and the QM is high in speed and can be used for on-line (inline) measurement without interrupting the work of the equipment.

As shown in fig. 15, an embodiment of the present invention provides a method for measuring an actual pixel size of a charged particle beam scanning imaging device, including: carrying out offline comprehensive measurement FM on the patterned wafer by using charged particle beam scanning imaging equipment to obtain the actual pixel size of each pixel to be measured; and performing partial measurement QM based on the comprehensive measurement FM method, wherein the partial measurement QM is used for the equipment to measure the actual pixel size of partial pixels to be measured on line.

After the actual pixel size of each pixel to be measured is obtained through the comprehensive measurement FM, the actual pixel size of the pixel to be measured needs to be measured again because the actual pixel size may change with time during the use of the device. In order to not delay the work of the equipment, the actual pixel size of part of pixels to be measured is measured on line by the equipment when the equipment executes the work of the equipment of the work of; on the other hand, the following description considers the measurement results of the full measurement FM and the partial measurement QM in combination, and provides a monitoring method.

The following describes the FM method for overall measurement of the actual pixel size in this embodiment.

In this embodiment, the method for comprehensively measuring FM includes:

acquiring template images on the wafer, extracting one or more templates from the template images, and storing the templates;

moving the wafer according to a preset direction and a preset distance, collecting a target image, performing template matching in the target image according to the templates to obtain matching positions which are greater than or equal to a similarity threshold, obtaining target displacement between each template and the corresponding matching position, and obtaining weighted target displacement by taking the similarity ratio corresponding to each matching position as a normalization weight;

taking the target image as a template image, performing next extraction template and template matching to obtain another weighted target displacement, continuing in the same way after the target image is completed, stopping image acquisition and template extraction until a preset stopping condition is met, and entering the next step;

the actual pixel size is obtained from the accumulated wafer motion distance and the (accumulated) weighted target displacement.

The basic idea is that template images matched with the templates are selected on the wafer, the templates are extracted from the template images, then the wafer is moved to acquire target images, the templates are used for template matching, and finally the measurement values Px and Py of the actual pixel size under the set theoretical magnification are obtained by combining the displacement of the mechanical motion platform. This is similar to the background art procedure except that both wafer movement and template matching are superior so that the measurement error of the result is lower and the accuracy is higher. As mentioned before, the full measurement FM needs to cover all the actual pixels to be measured, for example, the actual pixel size measurement involves 7 different magnifications. Referring to fig. 7, a template image 710 under theoretical magnification corresponding to a certain pixel to be measured has a template area location 711, and brightness, contrast, and uniqueness in the image/FOV all satisfy the template condition. Similar areas that may be used as templates include

areas

712, 713, 714 and more (not shown). Similarly, the theoretical magnification corresponding to the pixel to be measured is higher, the template image 730 is provided, and the region 732 is provided as the template for comprehensively measuring the FM. When the theoretical magnification corresponding to the pixel to be measured is higher, a template image 740 exists, wherein an area 742 can be used as a template for comprehensively measuring FM.

The template matching method referred to above includes similarity such as NCC (cross correlation method) or feature-based template matching. The features include corners extracted as Harris, FAST algorithm and post-processed, e.g. non-maximum suppression. The matching uses its corresponding feature vector/Descriptor (Descriptor), and the feature dense region can be converted into an image class template, matching uses similarity. In addition, template matching in current commercial/industrial image processing/machine vision software, including feature point-based template matching, is mature, and can be done. In addition, the user can also self-define the feature and the corresponding feature vector, and the dot product is used for matching, which belongs to the field of the prior art and is not described any more.

In this embodiment, the wafer is moved according to the predetermined distance, but due to the error of the mechanical motion platform, there may be a difference between the actual moving distance of the wafer, i.e. the wafer moving distance, and the predetermined distance.

The preset stop condition may be that a preset maximum matching time is reached, for example, the maximum matching time is 19 times; or the cumulative sum of the wafer movement distances may reach a preset value, for example, the total stroke of the mechanical movement platform reaches a preset maximum value; the template satisfying the preset extraction condition may also be unable to be obtained from the current image, for example, the preset extraction condition may be that brightness, contrast and uniqueness all satisfy corresponding threshold conditions; it is also possible to obtain a number of sub-templates that satisfies a threshold condition in the current image, for example, the number of sub-templates is insufficient or the number of sub-templates that have not been extracted is zero. Of course, the preset stop condition is not limited thereto, and for example, the preset stop condition may also be a combination of at least two of the above conditions, as long as one of them is satisfied, that is, the preset stop condition is considered to be satisfied.

In this embodiment, the specific steps for fully measuring FM are set forth below. Referring to FIG. 8, the electron optical system 140 (including the SEM system in this embodiment) first acquires a first starting template image 810, typically at least 1 (typically a plurality) of image type templates such as the previous region 722 (without excluding the previously mentioned feature type templates, although image type templates are preferred in this embodiment of the invention), selects a certain number of regions as templates, e.g., a template a, b, c, d, e, moves the wafer/mechanical motion stage, acquires the next frame of image, i.e., the target image 811, and searches the target image for a matching position a ', b ', c ', d ', e ' that meets a predetermined threshold condition using the templates a, b, c, d, e, and detects a new template in the target image 811 (the template condition for the image type is still the current brightness, contrast, or, The characteristics and uniqueness meet the requirements, and can be automatically extracted by a certain algorithm), for example, the region f' is used as a template, after a wafer is moved in a given direction to acquire a new target image 812, template matching is performed therein to obtain matching positions b ", d", e ", f" meeting a given threshold condition, and a new template, for example, g ", h", can be detected in the image, and is used for performing template matching in the new target image acquired after the wafer is moved to obtain target displacement of the next stage, although the actual work often cannot be done so far, because it cannot be guaranteed that the template matching is maintained by finding the new template, but only a single stroke/step length is reasonably planned, and because of adopting multiple templates, 5-6 times of such movement + matching can be guaranteed, the method can be called as a "multiple template weighted chain matching", rather than a single movement as in the prior art. The actual pixel size thus finally obtained is:

wherein the content of the first and second substances,

and are respectively the weighted average of the target displacements (X, Y direction) corresponding to all template matching (J in total) results that reach the established threshold condition in the ith (between the ith wafer position/image and the (i + 1) th wafer position/image) template (from the ith frame image) matching, (dXmj, dYmj) is the target displacement determined by the jth template matching in the J successful matches in the ith movement; then (DXmi, DYmi) is the i-th wafer single movement displacement used for the calculation of the actual pixel size. Where i is 1, … n, n is the wafer movement number, n +1 is the number of image acquisitions on the wafer, and the weight Wj is from the template matching score, and the weight Wj is:

where Sj is the similarity of a single template match that satisfies a given template match threshold condition (e.g., 0.60), with values between [0, 1 ]. The target displacements obtained for a single match in the X, Y directions are:

dXmj＝Xmj′-Xmj

dYmj＝Ymj′-Ymj

where (Xmj, Ymj) is the position of the jth template of the J templates in the template image, (Xmj ', Ymj') is its position in the target image acquired by the wafer moving to the next position.

In addition, since the movement of the mechanical motion stage is only related to the start and end positions, then (the molecular part of the above formula) there are:

where n ═ n + 1. The actual pixel size thus finally obtained can be written as:

since the actual situation includes errors, the formula for considering the errors is as follows:

the newly added part has the maximum error of the mechanical motion platform position degree as δ S (usually about 0.5 μ M), and the maximum error of single image template matching as δ M (usually between 0.025 pixel and 0.25 pixel, usually determined by various factors). The Px, Py error portion can be further separated, thus having:

since the template matching error δ M in the denominator is 3 orders of magnitude smaller than the single target displacement (Dxmi, DYmi), n δ M is negligible and δ S is fixed, looking at the error/second part of the above equation, i.e.:

it can be seen that the actual pixel size obtained by using the multi-template weighted chain matching in the embodiment of the invention is greatly reduced in comparison with the relative error of the single-template matching result in the prior art, and in addition, the total wafer movement times can be reduced by using the multi-template while still having higher statistical accuracy. Thus, there are:

in addition, when the magnification is rarely high, for example, the theoretical pixel size is around 2nm, the feature becomes sparse in the SEM image, and at this time, the image preprocessing including edge extraction, for example, the convolution of the image I with the gradient Δ G or Laplace ^ G of the Gaussian function G, i.e., Δ G ^ I, in which ^ represents the convolution operation, the size of the scale 2 σ of the Gaussian function changes (in the opposite direction) with the change in the size of the FOV, is empirically determined. Then, a single connected curve segment can be obtained by using a conventional thining method, and then a characteristic region, such as a corner portion, excluding an end point in the curve is obtained as a template, and weighted chain multi-template (weighted) matching is performed in the above manner to obtain an actual pixel size. For example, referring to fig. 9, there is provided an HM image of high magnification HM, which is an image after edge extraction, in which there are a plurality of

templates

902, 903 with features (2-directional edges/corners) that can be matched as templates. Since the matching is affected by how much the image looks irregular in the high magnification HM, the matching results of the individual templates can also be assisted by the distance of the connecting lines between them (e.g. 908 in fig. 9) and their relative orientation to confirm the correctness of the matching results. In addition, because the maximum error range of the mechanical motion platform is limited each time, the searching range of the template matching is limited, which is helpful for determining whether the matching is correct or not, and the requirement (similarity threshold) of the template matching can be relaxed (reduced). The image processing techniques involved are common in the art and are not described in detail herein.

When a template is extracted from the template image, a weighted chain matching method may also be used, and the weight at this time is 1. Therefore, the full scale FM measurement method provided in this embodiment uses a weighted chain matching method, which includes both the case of the single-template weighted chain matching method and the case of the multi-template weighted chain matching method.

Therefore, the overall FM measurement method in the embodiment of the invention covers all pixels to be measured, and due to the use of the weighted chain type matching method (for example, the multi-template weighted chain type matching method), compared with the prior art, the method has the advantages of lower relative error and higher precision, and the precision and the reliability of the method are greatly improved compared with the method in the prior art.

The following describes part of the QM measuring method in the embodiment of the present invention.

According to the idea of the embodiment of the present invention, the partial measurement QM is a fast actual pixel measurement method, and there is little or no additional time consumption, so that the device can also fast measure/evaluate the actual pixel size of the system while performing the work, and can be performed on-line, and can be used to further judge the reasonable time for comprehensively measuring the FM.

In this embodiment, the partial measurement QM method includes:

for any part of pixels to be detected, template images are collected on the wafer, and at least one pair of templates meeting preset template requirements are extracted from the template images; obtaining and storing the template distance between two templates in each pair of templates;

acquiring a target image on a wafer, performing template matching by using the template to obtain matching positions meeting preset target requirements, obtaining a target distance between two matching positions in each pair of matching positions, comparing the template distance with the corresponding target distance, and obtaining the actual pixel size by combining the corresponding actual pixel size obtained in the process of comprehensively measuring FM.

In the present embodiment, the measurement principle in QM is explained below. Recall that, as described in the background, the relationship between the pixel sizes Δ X, Δ Y and X, the field size foxx in the Y direction, and the FOVy is foxx ═ Nx Δ X, FOVy ═ Ny Δ Y, where Nx, Ny are X, and the pixel size in the Y direction is usually the same as Nx ═ Ny ═ N, that is, the SEM image width and height, foxx ═ N Δ X, FOVy ═ N Δ Y, and N is the number of pixels in the X/Y direction of the image. Thus, with Nx, Ny unchanged, the actual field size is proportional to the actual pixel size. So that changes in actual pixel size can be estimated as long as changes in distance between certain fixed objects/features in the image can be measured. Thus after the correct (within design criteria) pixel size is known, areas in the image with features (both corners or X, Y sides) are selected, e.g. in template image 1000 in fig. 10A there are

image areas

1001, 1002, 1003, 1004 as templates. There are several

independent feature points

1006, 1007, 1008 as described above in fig. 10A, which can also be used as templates in the case of image class template absence. The selection of the individual feature points is many, the same as in the previous full measurement FM.

In the present embodiment, with continued reference to fig. 10A, where partial measurement

QM template areas

1002, 1003, 1004, which are templates of image types, and

feature points

1006, 1007, 1008, which are the above-described feature class templates, can be defined, in this case of fig. 10A, an image type template can be constructed with sub-images in the periphery thereof. In this embodiment, the template requirements include template conditions (including brightness, contrast, and uniqueness requirements) and distance conditions, and a pair of templates (each template is used for 1 time) that satisfy the template conditions (the templates individually satisfy the requirements of characteristics, contrast, uniqueness, etc.) and satisfy the distance conditions (the distance between 1 and 1, i.e., the template distance, reaches or exceeds a predetermined template distance threshold) may be numbered and stored, for example, in a Recipe. The template distance threshold includes the minimum distance between the pair of templates (e.g., 1/5 where the distance between the centers of the pair of templates is greater than the minimum width and height of the image) and the distance between the respective regions and the edge of the image (e.g., 10 pixels). In a specific implementation, the feature point type template is usually supplemented in the case that the image class templates satisfying the predetermined condition are insufficient, otherwise, the template of the image class is usually selected preferentially. In addition, the dense region of the feature class templates can be converted into image class templates, and the region of the image containing the feature class templates (if the template condition is met) is used. It is assumed that this is the SEM image 1000 when the wafer is aligned with WA, and is taken as the template image when WA, and there is an area 1001 for the wafer to be aligned with WA template. The magnification of the SEM image 1000 is exactly the same as the theoretical magnification of the pixels under test in the partially measured QM, which allows the wafer to be directly aligned to the template image of WA. However, in general applications, the theoretical magnification of the image used for partial measurement QM is generally the same as the theoretical magnification of a pixel under test in the global measurement FM and different from that in wafer alignment WA, although it is not excluded that in some cases, the theoretical magnification of 1 or more actual pixels under test in the global measurement FM is the same as that in wafer alignment WA.

In this embodiment, the template requirement includes a template distance threshold value regarding the template distance, and the target requirement includes a target distance threshold value regarding the target distance. In this embodiment, the target distance threshold is the template distance threshold corresponding thereto plus a tolerance (e.g., a fixed amount such as 3 pixels, or a relative amount such as 5% of the template distance).

In the present embodiment, the template image selection and template matching method in the partial measurement QM are explained below. The information of each template in the QM template set is partially measured, some are image types, some are feature points (feature vector description), and 1-1 distance between each template and orientation angle of connecting lines are also important information, which are all information to be stored. Partial measurement QM template selection may be performed by capturing template images independently, e.g., by specifically finding areas on the wafer that contain rich template pairs (which may be referred to as image capture areas), or by referencing template image information in wafer alignment WA (which may be performed after establishing Recipe for wafer alignment WA), e.g., by directly selecting the position of the template in the template image at a certain level of wafer alignment WA, or its peripheral position. For example, the template image at the time of wafer alignment WA may be used as the template image, which typically includes a plurality of image class templates and independent feature point templates, if and only if all pixels to be measured include the same theoretical magnification and the same SEM image in wafer alignment WA, that is, if the magnification of the SEM image used for wafer alignment is the same as the theoretical magnification of the pixels to be measured; ii) acquiring a template image acquisition position on the wafer and acquiring the template image when the wafer is aligned with the WA, namely, aligning the WA template image acquisition position by using experience (because the characteristics are abundant and the template is well found in general); iii) based on the template image capture position on the wafer when the wafer is aligned WA, using the apparatus to bias the charged particle beam and capture the template image at the biased position, i.e. using the template image capture position when the wafer is aligned WA, the mechanical motion stage is not moved but the electron optical system 140 is used to bias capture nearby images, which typically include multiple image class templates and independent feature point templates. The biasing is prior art in current electron optical systems with the advantage that no mechanical movements are time consuming. The selection of the mode in practical application is determined according to practical requirements. In addition, a consistent method is required for creating and executing both a Recipe, for example, if a bias is used for creating a Recipe, then a bias is also used for executing the Recipe; iv) collecting the template image in an image collecting area in a Die (Die) with a number of template pairs larger than or equal to a preset number on the wafer; this is one of the options, but the embodiments of the present invention tend to make the best use of the information of the wafer alignment template image itself or its image capture area, since wafer alignment is the most important preparation for the equipment, and the requirements for templates are very high and close.

In this embodiment, the successfully matched target image is selected according to any one of the following methods:

obtaining a target image corresponding to the template image acquisition mode at a successfully matched position closest to the wafer center in the wafer alignment WA;

obtaining a target image corresponding to the template image acquisition mode at a successfully matched position with the highest matching similarity in the wafer alignment WA;

and acquiring a target image in an image acquisition area of the same crystal grain or an image acquisition area at the same position in other crystal grains (the image acquisition area of the same crystal grain) on the wafer.

The offset is additionally described. For a precisely calibrated electron optical system 140, the offset may help to substantially expand the stroke to n times, e.g., n > 10, compared to no offset, so that the range of choices around the perimeter is large and many electron optical systems 140 may be biased on the order of microns sufficient for SEM images involved in embodiments of the present invention.

In the present embodiment, the calculation steps in part of the measurement QM are described below as an example. With continued reference to FIG. 10A, where the distance between the

templates

1002 and 1003 is referred to as the template distance d23, the distance in the X, Y direction is dX23, dY23, and is saved after determination. In the case of the partial measurement QM measurement to be performed subsequently, the target image is captured to the above-mentioned certain capture position (for example, 611 or 612 in fig. 6A), and a pair of templates consisting of the saved

templates

1002 and 1003 is fitted thereto. The distances dX23 ', dY 23' between corresponding matching objects are obtained, for quickly estimating the change in actual pixel size,

where Px, Py is the previously saved correct (i.e., obtained from the most recent full measurement FM) actual pixel size, and Px ', Py' is the actual pixel size currently estimated using the partial measurement QM method. It can be seen that this method, while relatively simple, is suitable for the monitoring of the actual pixel size hereinafter.

When there are a plurality of templates in the template image and there are n pairs of templates with non-repeating spacing satisfying the predetermined condition, one pair with the farthest distance in the image, for example, the i-th and j-th templates, can be selected and written into a more general form, the distance dXij, dYIj in the template image, and then the distance dXij ', dYIj' in the target image have:

the matching of m pairs of templates (each template can be used only once) whose distance in the image reaches a predetermined threshold and whose similarity score reaches the predetermined threshold when matching in the target image may be selected, and the average of these distances may be obtained, and the results may be replaced with dXij, dYij, dXij ', dYij' in the above formula, thereby improving the reliability of the results. Still further, the matching of m pairs of templates in which the distances in the X and Y directions in the image both reach a predetermined threshold and the similarity score reaches a predetermined threshold when matching in the target image can be selected, and a weighted average of these distances is obtained, in the above formula, instead of dXij, dYij, dXij ', dYij', where the normalized weight can be determined by the distance between each matching pair and/or the minimum/average of the matching similarity of each pair of templates. And averaging results in a relatively higher accuracy. There may then be a basic formula for the partial measurement QM method:

wherein the content of the first and second substances,

dXij and dYij are respectively the template distances in the X and Y directions of a pair of templates formed by the ith and jth templates in the template image, 1 ≦ i ≦ M, 1 ≦ j ≦ M, M is the total number of templates meeting the requirements of the templates, the target distances in the X and Y directions of corresponding successful matches in the template image are dXij 'and dYij', respectively, the sum traverses all (at least 1 pair of templates) successful matches meeting the distance threshold/requirement between the templates, and is not repeated, and ω ij is a set weight, for example, the weight occupied by the pair of matches in all successful matches, as described above, can be determined by the distance between each matched pair and/or the minimum value/average value of the matching similarity of each pair of templates, or can be a constant value, i.e., there is no weight difference. Px, Py are still correct, e.g. the actual pixel size after the last full measurement FM has been completed, Px ', Py' are the actual pixel sizes currently estimated with the partial measurement QM method.

The description is supplemented otherwise. When Px ', Py' is changed relative to the original Px, Py, the template image itself will be changed to the same extent as the whole image, but the whole result will not be affected, because for the normal device system, Px ', Py' is still slightly changed relative to the original Px, Py, and there is usually no need to perform a scaling search during template matching (the scaling search slightly increases the computation time, but the algorithm usually runs under the current normal CPU and increases the time by no more than 10 ms), so as not to affect the template matching itself. Further the template matching includes not only the above mentioned image class template matching, e.g. with NCC algorithm, but also or feature based template matching, e.g. using feature points, many of which are substantially unaffected by the image scaling, which is true both for low magnification LM and very high magnification HM SEM images.

When the theoretical magnification of the actual pixel to be measured is high, the method of partially measuring QM may be similar to that of low magnification LM, and also a method of relatively changing the template distance. As shown in the template image 1010 in fig. 10B, where

templates

1012, 1013, 1014 can also be selected, the distance between them can also be used to estimate the actual pixel size at this magnification, but the measurement accuracy is slightly worse at lower magnifications LM. Because on the one hand the FOV becomes smaller at high magnification HM, the distance between the templates becomes smaller and the local irregularity of the image increases, and also because of the variation of the image (the template image and the target image may be from the same kind but different wafers, or different locations on the wafers), and the SEM imaging mechanism decides that the SEM image at high magnification HM is always worse at lower magnification LM. But the corresponding error should be within 1 pixel and the distance between the templates is 10 ² At the pixel level, so the relative error is still small, still at 10 ^-2 Magnitude.

In rare cases, when the magnification is high, such as shown in fig. 9 above, after the image is edge extracted, there are still usually very local curve segments with matching conditions that satisfy the template, and more than 2 desired templates can be given, wherein the distance between 2 templates 1-1 that meets the template requirement and the template distance threshold/requirement can still be used in the above method.

Still further, referring to fig. 11, in the case where the theoretical magnification of the pixel to be measured is extremely high (rare), there are sometimes only one template 1102 satisfying the template matching condition in the template image 1101, and the above-described actual pixel size measurement method based on the distance between templates cannot be used. Also as in the case of a wafer aligned WA template image 1103, only one usable template 1104 can be generated.

In the embodiment, when only a single template meeting a preset template threshold exists in the template image, the device is used for biasing the charged particle beam to obtain a bias image, extracting the template from the bias image, forming at least one pair of templates with the single template in the template image and acquiring a template distance; acquiring target images at preset positions on the wafer, acquiring offset target images by using the offset which is the same as the offset, performing template matching on the templates in the unbiased template images and the offset template images to the corresponding unbiased target images and the offset target images to obtain at least one pair of matching positions, and acquiring a target distance between two matching positions in each pair of matching positions.

In the method provided in the embodiment of the present invention, the aforementioned biasing function of the electronic optical system 140 is adopted, so that the electronic optical system 140 can scan the area in the neighborhood of the predetermined distance from the system center, and finally obtain two templates located in 2 images, and the distance between the two templates can be used for the calculation in the partial measurement QM. At this point, electron optics 140 is given an appropriate bias voltage without moving the wafer. The offset here is the same as the offset described above in terms of working principle but for a different purpose, the former offset is for acquiring an image near the WA template image acquisition position as a template image, and the offset here is for stitching measurement images. At this time, an offset is set to the electron optical system 140, and an image of the adjacent area at the same magnification is obtained. It should be noted that, there is no need to actually stitch the two images (based on a stitching method of the overlapped part features in 2 images), the distance between the two images can be obtained completely by calibration of the electronic optical system 140, for example, the calibration can generate a Lookup Table (LUT) of offset and distance, and the offset distance can be converted into an error corresponding to the offset in units of pixels, which should not exceed 1 pixel in the case of accurate calibration of the system, and is 2-3 orders smaller than the distance between a pair of templates constituting part of the measured QM template, and can be ignored. The calibration itself is outside the embodiments of the present invention and is prior art, and is not described herein.

Another template image acquisition and matching method, also used for partial measurement QM at high magnification HM, is described below, which is applicable to situations where features in small view field images are rare at high magnification HM. Referring to fig. 12, there is a template image 1200 at an original high magnification HM, in which there is a template 1202, in which there is a feature area 1206. At this time, the electron optical system 140 is set with a bias to obtain an image 1230 of the adjacent area at the same magnification. In addition, the template feature area 1202 may generate the image-like template 1203, but in such a case of high magnification, the interference effect due to feature content, local deformation, noise, etc. is not necessarily good, but other types of templates may be generated, for example, a certain number of points 1204 on two adjacent sides of a corner are used to fit a 2-line, an intersection 1205 or a certain distance along an angular bisector thereof determines a reference point, which is defined as the position of a feature template, and the template may still use a part of the feature-like template (a Descriptor) extracted by the FAST algorithm, for example. Similarly, the template position is determined in the offset image 1230 in the same way, for example, by using the image class template 1232 and obtaining the intersection 1235 of the sampled and fitted lines on the two sides of the corner as the feature point. The template matching of the feature class usually uses vector dot product/descriptor matching correlation algorithm, the result is also between [0, 1], and is similar to the result of NCC. Thus, the distance between template 1202 (in the original image) and template 1233 (in the offset image) can be obtained using the above-described pixel size acquisition method based on the distance between templates, and thus:

wherein, the first and the second end of the pipe are connected with each other,

DXod＝Xd-Xo+XDef

DYod＝Yd-Yo+YDef

DXod‘＝Xd’-Xo‘+XDef

DYod’＝Yd‘-Yo’+YDef

where (Xo, Yo) is the position of the template 1202 in the (original/unbiased) template image 1200, (Xd, Yd) is the position of the template, e.g., 1232, in its neighboring offset image 1230, XDef, YDef is the distance (in pixels) in the X, Y direction over which the two frames of images are offset, and both (Xo, Yo) and (Xd, Yd) are previously retained correct/within system specifications, and (Xo ', Yo') and (Xd ', Yd') are on the target image to which the template is matched in QM, one is the template matching position on the original/unbiased image and the other is the template matching position on the offset image. Incidentally, the error due to the offset is included in the fixed XDef and YDef, and the maximum value thereof is δ D. However, as described above, the δ D < 1 pixel of the precisely calibrated system is negligible because it is 2-3 orders of magnitude different from the template distance.

It should be noted that, in the embodiment of the present invention, there is no limitation on the direction and position of the offset image with respect to the original image and the number of the offset images, as long as the electronic optical system 140 can support (within a certain error range), for example, in fig. 13, the original image 1301 when the wafer is aligned to the WA may have the offset images 1302 and/or 1303, and there are many other possibilities, which are not described herein again.

In this embodiment, there is also provided a method for monitoring an actual pixel size of a charged particle beam scanning imaging device, including:

1417, when the first counter Nq does not reach the first parameter Nq, returning to the step 1415, and performing homework by the equipment as usual; otherwise go to step 1418;

1418. performing QM based on said partial measurement while clearing the first counter nq;

1419. when the result of the actual pixel size is greater than or equal to the second threshold value Thfi, returning to the step 1411 to perform the calibration maintenance of the device again; otherwise, when the monitored value is smaller than a first threshold value Thqi, returning to the step 1415, and continuing to execute the job work of the equipment; otherwise, when the monitored value is smaller than the second threshold value Thfi but greater than or equal to the first threshold value Thqi, the first parameter Nq and the second parameter Nf are changed, and then the process returns to step 1415.

A method of monitoring an actual pixel size in the embodiment of the present invention is explained below.

Firstly, reasonably planning and monitoring based on respective characteristics of the comprehensive measurement FM and the partial measurement QM. The FM is measured completely and needs to be done off-line, all pixels to be measured need to be covered, the result precision is high, and the FM is used for the work of the equipment. And partial measurement QM can be performed on line, a few theoretical amplification factors can be selected, the speed is high, and the method is used for determining the time for comprehensively measuring FM. The partial measurement QM is performed in the equipment performing the job, all after the equipment performs wafer alignment. Specifically, this occurs at a time after the apparatus completes wafer alignment WA at the low magnification LM and/or the high magnification HM, or after the apparatus completes the entire wafer alignment WA and before the job is started. For example, usually, the wafer alignment WA includes a first-order low magnification LM and a first-order high magnification HM, and if the magnification of the pixel to be measured in the full-scale measurement FM is the same as the magnification of the first-order low magnification LM and the first-order high magnification HM in the wafer alignment WA, part of the measurement QM is also performed with these two magnifications.

The embodiment of the invention specifically plans the frequencies for making the overall measurement FM and the partial measurement QM in the monitoring. The cumulative number of times N the wafer is processed by the tool, or the cumulative time T for which the tool is operating, may be used. Both meanings are substantially the same. For convenience, the accumulated number of times N the wafer is processed by the apparatus, and there are many options for counting, for example, the apparatus (computer 150) may record once after each wafer. In the embodiment of the present invention, the number of wafer processes Nq required to be accumulated for partial measurement QM is referred to as a first parameter Nq, and the number of wafer processes Nf required to be accumulated for full measurement FM is referred to as a second parameter Nf, Nf > Nq. Nf and Nq are both operating parameters under monitoring, and they may be kept in a fixed proportional relationship, for example, Nf may be uNq, and the scale factor u may be a positive integer, for example, the scale factor u may be 10, Nq may be 1000, and Nf may then be 10000. These values are generally determined by the device characteristics and user experience, but may be subsequently further adjusted in use, and are provided in connection with the methods described herein.

In addition, Px, Py have their initial values when monitoring has not yet begun, which are typically ideal values determined in advance based on equipment performance and experience, and are subsequently replaced by new values after a full FM measurement has been successful. Usually the actual pixel change will be larger than the initial value, so the following references to thresholds are all unidirectional, only considering the case where the actual pixel size is larger than the ideal value.

Alternatively, it is also possible to consider Pxy max (Px/Py, Py/Px), since the system often requires the ratio of the two, and the difference between the two is too large, which is a problem if not designed intentionally, and causes image distortion. The values to be measured can now be written uniformly in the form Pi, i ═ x, y, xy, and accordingly the second threshold value Thfi and the first threshold value Thqi are respectively set to include three threshold values to be compared with the actual pixel size Px in the x direction, the actual pixel size Py in the y direction, and the proportional relation Pxy, respectively. They have different threshold values for the overall measurement FM and the partial measurement QM, including the second threshold value Thfi, i ═ x, y, xy and the first threshold value Thqi, i ═ x, y, xy, and Thqi < Thfi, i ═ x, y, xy, which, of course, exceed the initial value of Pi. The significance of this is that when a certain Pi (i ═ x, y, x) exceeds the corresponding threshold value Thfi, meaning that the actual pixel size of the device deviates more severely from the index at that time, the system must be maintained and calibrated as soon as possible; when Pi is lower than Thqi, the actual pixel size of the equipment is normal, and the next partial measurement QM and the full measurement FM can be carried out according to the set time; whereas when all Thqi ≦ Pi < Thfi (i ≦ x, y, xy), i.e., between the two thresholds, it means that Pi, although within the normal range, deviates from its initial value to some extent, and monitoring needs to be more frequent/intimate. The above threshold values Thfi, Thqi, i ═ x, y, xy are mainly derived from system metrics/limitations, typically based on user experience. In addition, different charged particle beam scanning imaging systems and different applications are different. For example, the Thqi used by EBR equipment is much higher than the Thqi used by CD-SEM because EBR equipment is also primarily concerned with the presence or absence of defects and classification of defects, and the requirements for measuring dimensions are relatively relaxed.

The specific steps of the above monitoring are set forth below. Please refer to fig. 14. First, although fig. 14 includes a step of determining whether the result is "yes" or "no", fig. 14 is labeled only with "yes" and is not labeled with "no" for convenience.

It should be further noted that at the beginning of the monitoring process, the values involved in the parameters Nq, Nf, Thfi, Thqi, i ═ x, y, xy, and the partial measurement QM for the device are determined. The parameters required for the full measurement of FM and the partial measurement of QM have also been determined.

In this embodiment, fig. 14 shows a closed loop flow of the monitoring method according to the embodiment of the present invention.

The flow begins at step 1411, normal maintenance calibration of the system. Since usually when a new plant or its system/components are updated, appropriate maintenance calibrations need to be performed to ensure that the plant is working properly. After the system is calibrated for the first time in operation/maintenance, the first parameter Nq, the second parameter Nf, the first threshold Thqi, and the second threshold Thfi, i ═ x, y, xy are required, and of course, the initial value of Pi is also required. As previously mentioned, they are determined when creating the Recipe.

Then, in step 1412, a full measurement FM is performed, covering all theoretical pixels/magnifications to be measured, as previously described, while the second counter nf for the full measurement FM and the first counter nq for the partial measurement QM are cleared together. The method for measuring FM globally at this time is as described above. The second counter nf and the first counter nq are both counting wafers, e.g. counting each wafer, and are recorded once.

Then, in step 1413, the actual pixel size at each level/theoretical magnification is checked and the FM is measured globally as described above. If any one of the obtained actual pixel sizes exceeds the second threshold value Thfi, i.e. Pi ≧ Thfi, i ═ x, y, xy, a warning is given to let the user timely return to step 1411 for system calibration, and then the device resumes operation.

Otherwise, in a next step 1414, all Pi, i ═ x, y currently remain for further study of the monitoring process, if necessary. Then, step 1415 is performed, the equipment performs the job normally (for example, the job of EBR is to re-inspect the defect and classify the defect), and the result of the latest Pi, i ═ x, y after the FM is fully measured may be used when the equipment fails, and the second counter nf and the first counter nq are counted again from zero, for example, as described above, and are recorded once per wafer.

Then, in step 1416, the second counter Nf is checked, and when it reaches the predetermined threshold Nf, the routine returns to step 1412 to perform the full FM measurement, otherwise, in step 1417, the first counter Nq is checked, and when it reaches the predetermined threshold Nq, the next step 1418 is performed to perform the partial QM measurement, and the method described above is used while the first counter Nq is also cleared. Note that the device is still doing its job at this point, rather than stopping it as it would when measuring FM comprehensively.

Then, in the next step 1419, the result is checked by using a second threshold value Thfi, if any one of the actual pixel size measurement results included in the partial measurement QM exceeds the second threshold value Thfi, that is, Pi is greater than or equal to Thfi, i is equal to x, y, xy, which indicates that the problem is serious, a warning is provided, the user returns to the step 1411 to perform system maintenance and calibration at a proper time, and then the monitoring process is restarted.

Otherwise, the result is checked by using the first threshold Thqi, and if any one of the actual pixel size measurement results included in the partial measurement QM exceeds the first threshold Thqi, i.e. Pi ≧ Thqi, i ═ x, y, xy, this indicates that there is a slightly large variation or a tendency of variation (usually, the actual pixel size becomes larger) in the actual pixel size of the system, so that the detection needs to be performed slightly frequently, which can be implemented by adjusting the first parameter Nq timely. Then, in step 1422, optionally, the first parameter Nq is adjusted (in the embodiment of the present invention, the proportional relationship between the second parameter Nf and the first parameter Nq, which tends to be fixed, changes the first parameter Nq, and thus changes the second parameter Nf), and after that, the process returns to step 1415. Otherwise, Pi < Thqi, i ═ x, y, xy, which indicates that there is no abnormality, the process returns directly to step 1415, the loop continues, the system continues to perform the job, the monitoring is also in progress, nothing is changed, the second counter nf, and the first counter nq continues to count. In a timely tuning method of the first parameter Nq (or even the second parameter Nf because of a fixed relationship therebetween), the tuning amplitude may be proportional to the deviation from the previous/previous actual pixel size, for example, setting the new first parameter Nq to be S% times the previous first parameter Nq, that is:

Nq‘＝Nq×S％,

Nf‘＝uNq‘

therefore, there are:

Nf’＝Nf×S％

where Nf 'and Nf are the adjusted and current second parameters, Nq' and Nq are the adjusted and current first parameters, λ i is the set weight, S is a bounded adjustment factor, which may be determined by the degree of deviation from the last actual pixel dimension value, and the adjustment factor S is calculated by the average of the degree of deviation of the actual pixel dimension Px in the X direction, the degree of deviation of the actual pixel dimension Py in the Y direction, and the degree of deviation of the proportional relation Pxy, for example:

where S has a limit, S ≦ St, which is a predetermined threshold, e.g., S ≦ 5. Pi is the value of the qualified actual pixel sizes Px and Py stored after the current full-scale measurement FM and the value of the proportional relation Pxy, Pi' is the value of the actual pixel sizes Px and Py of the current partial measurement QM and the value of the proportional relation Pxy, and i is x, y and xy. As mentioned above, the new results are always larger than the former ones, so that the absolute values of the above-mentioned molecules are not true. Then, as above, since the second parameter Nf is related to the first parameter Nq and the scaling factor u is still fixed, the second parameter Nf is updated when the first parameter Nq is updated. In addition, the above formula may also include a weighting factor, and thus there may be

Where λ i (i ═ x, y, xy) represents weights, all of which sum to 1, each weight being determined by the experience of the user, e.g., Px, Py with a weight λ x ═ λ y ═ 0.4, and Pxy with a weight λ xy ═ 0.2, and some of the foregoing are associated with the former two, the next to the others. S still has a limit such as S ≦ 5. Similar methods for updating the second parameter Nf and the first parameter Nq are many, and are not described herein again.

As can be seen from the above description of the monitoring process, the monitoring method is a closed loop process that can be automatically tuned. The advantage is that the use of the full measurement FM and the partial measurement QM is a second parameter Nf and a first parameter Nq with different frequencies. Thus, the precision of the FM is comprehensively measured and the speed of the QM is partially measured, so that the time of measuring the actual pixel size is accurately grasped in the working process of the equipment, the time of measuring the actual pixel size is not delayed, and precious machine time and resources are not wasted, the problems in the prior art described in the background technology are solved, and the performance, the efficiency and the use cost of the equipment are greatly improved.

The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.

Claims

1. A method of measuring an actual pixel size of a charged particle beam scanning imaging device, comprising:

carrying out offline comprehensive measurement FM on the patterned wafer by using charged particle beam scanning imaging equipment to obtain the actual pixel size of each pixel to be measured;

and performing partial measurement QM based on the comprehensive measurement FM method, wherein the partial measurement QM is used for the equipment to measure the actual pixel size of partial pixels to be measured on line.

2. The measurement method of claim 1, wherein the comprehensive measurement FM method comprises:

and acquiring the actual pixel size according to the accumulated wafer movement distance and the weighted target displacement.

3. The measurement method of claim 1, wherein the partial measurement QM method comprises:

4. The measurement method of claim 3, wherein the template requirements include a template distance threshold with respect to the template distance and the target requirements include a target distance threshold with respect to the target distance.

5. The measurement method according to claim 3, wherein when there is only a single template satisfying a preset template threshold in the template image, the apparatus is used to bias the charged particle beam to obtain a bias image in which templates are extracted and at least one pair of templates is formed with the single template in the template image and a template distance is acquired; acquiring target images at preset positions on the wafer, acquiring offset target images by using the offset which is the same as the offset, performing template matching on the templates in the unbiased template images and the offset template images to the corresponding unbiased target images and the offset target images to obtain at least one pair of matching positions, and acquiring a target distance between two matching positions in each pair of matching positions.

6. The measurement method according to claim 3, wherein the template image is selected according to any one of the following methods:

when the magnification of an SEM image used for wafer alignment is the same as the theoretical magnification of a pixel to be detected, a template image used for wafer alignment is used as the template image;

acquiring a template image acquisition position on a wafer and acquiring the template image when the wafer is aligned;

based on a template image acquisition location on the wafer at wafer alignment, biasing a charged particle beam using the apparatus and acquiring the template image at the biased location;

and acquiring the template image in an image acquisition area in a crystal grain of which the number is greater than or equal to the preset number of template pairs on the wafer.

7. The measurement method according to claim 6, wherein the successfully matched target image is selected according to any one of the following methods:

obtaining a target image corresponding to the template image acquisition mode at a successfully matched position closest to the center of the wafer in wafer alignment;

obtaining a target image corresponding to the template image acquisition mode at a successfully matched position with the highest matching similarity in wafer alignment;

and acquiring a target image in an image acquisition area of the same crystal grain or an image acquisition area at the same position in other crystal grains on the wafer.

8. The method according to claim 3, wherein the combination of template pairs is selected and the target distance is obtained according to any one of the following methods:

selecting a pair of templates which meet the preset template requirement and are farthest away from each other in the template images for template matching, and acquiring the target distance after matching;

selecting a plurality of pairs of templates which meet a preset template distance threshold value in the template image for template matching, and using an average value of a plurality of pairs of corresponding target distances which meet preset target requirements as the target distance after matching.

9. The measurement method of claim 8, wherein the partial measurement QM method has the following basic formula:

wherein the content of the first and second substances,

dXij and dYij are respectively the template distance in the X direction and the Y direction of a pair of templates formed by the ith template and the jth template in the template image, i is more than or equal to 1 and less than or equal to M, j is more than or equal to 1 and less than or equal to M, M is the total number of templates meeting the requirements of the templates, the corresponding target distance in the X direction and the Y direction successfully matched in the template image is dXij 'and dYij', omega ij is the set weight, Px and Py are the actual pixel size after the FM is completed in the last overall measurement, and Px 'and Py' are the actual pixel size currently estimated by using the partial measurement QM method.

10. A method of monitoring an actual pixel size of a charged particle beam scanning imaging device, comprising:

1411. after an electronic optical system in the device undergoes maintenance and calibration, determining monitoring working parameters, wherein the monitoring working parameters comprise a second parameter Nf, a second threshold value Thfi, a first parameter Nq and a first threshold value Thqi, the second parameter Nf is greater than the first parameter Nq, and the second threshold value Thfi is greater than the first threshold value Thqi;

1412. performing a comprehensive measurement FM according to any of the claims 1-9 while clearing the second counter nf and the first counter nq;

1414. storing all actual pixel sizes measured in the current comprehensive measurement FM;

1418. performing a partial measurement QM based on any of the claims 1-9 while clearing the first counter nq;

11. The monitoring method according to claim 10, wherein the second parameter Nf and the first parameter Nq have a fixed proportional relationship Nf uNq, and the scaling factor u is a positive integer; counting by the first counter nq and the second counter nf after the wafer is mounted on the equipment wafer; the partial measurement QM is performed in the equipment performing the job, all after the equipment performs wafer alignment.

12. The monitoring method according to claim 10, wherein the second threshold value Thfi and the first threshold value Thqi are respectively set to include three threshold values to be compared with an actual pixel size Px in an x direction, an actual pixel size Py in a y direction, and a proportional relation Pxy, where Pxy is max (Px/Py, Py/Px).

13. The monitoring method according to claim 12, wherein the first parameter Nq and the second parameter Nf are changed by the following formulas:

Nq‘＝Nq×S％

Nf‘＝uNq‘

Nf’＝Nf×S％

wherein Nf 'and Nf are respectively the adjusted and current second parameters, Nq' and Nq are respectively the adjusted and current first parameters, the proportionality factor u is a positive integer, λ i is the set weight, S is the bounded adjustment factor,

or

Pi is the value of the qualified actual pixel size Px and Py stored after the current full measurement FM and the value of the proportional relation Pxy, Pi' is the value of the actual pixel size Px and Py of the current partial measurement QM and the value of the proportional relation Pxy, S is less than or equal to St, and St is a preset threshold.