Scanning electron microscope with inspection and measurement model, and method for correcting inspection and measurement model

Abstract

The present invention provides a scanning electron microscope with an inspection and measurement model that can be adjusted efficiently while making a user aware of reduced reliability of inspection or measurement results, and a method for correcting an inspection and measurement model. The scanning electron microscope with an inspection and measurement model is characterized by comprising: a feature analysis unit that accepts input of an electron microscope image or waveform of a sample to be inspected or measured to obtain a first feature, and accepts input of an electron microscope image or waveform of a calibration sample to obtain a second feature; a learning unit that generates an inspection and measurement model by learning on the basis of the second feature; and a calibration unit that compares the first feature to a prescribed calibration range on the basis of the first feature and the inspection and measurement model, and activates an alarm if the first feature deviates from the calibration range.

Description

Scanning electron microscope having inspection and measurement model and method for correcting inspection and measurement model

[0001] The present invention relates to scanning electron microscopes, and more particularly to a scanning electron microscope having an inspection metrology model and a method for correcting the inspection metrology model.

[0002] Known background art in this technical field is Patent Document 1. Patent Document 1 describes an image processing system that stores in advance in a storage device the detectable range of a detector included in a charged particle beam device, generates a simulated image of a three-dimensional shape pattern using the detectable range, and learns in advance the relationship between the simulated image and the three-dimensional shape pattern.

[0003] Japanese Patent Application Laid-Open No. 2022-29505

[0004] Patent Document 1 describes image processing for estimating a three-dimensional shape, and describes using an image of a calibration pattern whose three-dimensional shape is known and reflecting this in the output of the three-dimensional shape. However, Patent Document 1 cannot detect whether the characteristics of the actual sample to be measured fall outside the calibration range, which can prevent correct calibration and result in errors in the output of the three-dimensional shape.

[0005] Therefore, the present invention provides a scanning electron microscope having an inspection and measurement model that can be efficiently adjusted while making the user aware of a decrease in the reliability of the inspection or measurement results, and a method for correcting the inspection and measurement model.

[0006] In order to solve the above problems, a scanning electron microscope having an inspection measurement model according to the present invention is characterized by comprising: a feature analysis unit that inputs an electron microscope image or waveform of a sample to be inspected or measured to determine a first feature, and inputs an electron microscope image or waveform of a calibration sample to determine a second feature; a learning unit that generates an inspection measurement model by learning based on the second feature; and a calibration unit that compares the first feature with a predetermined calibration range based on the first feature and the inspection measurement model, and issues an alarm if the first feature deviates from the calibration range.

[0007] In addition, the method for correcting an inspection measurement model according to the present invention is characterized by comprising the steps of: a feature analysis unit inputting an electron microscope image or waveform of a sample to be inspected or measured to determine a first feature; and inputting an electron microscope image or waveform of a calibration sample to determine a second feature; a learning unit generating an inspection measurement model by learning based on the second feature; and a calibration unit calibrating the output of the inspection measurement model so that it matches the second feature, and correcting the inspection measurement model.

[0008] According to the present invention, it is possible to provide a scanning electron microscope having an inspection and measurement model that can be efficiently adjusted while making a user aware of a decrease in the reliability of the inspection or measurement results, and a method for correcting the inspection and measurement model. Problems, configurations, and effects other than those described above will become apparent from the description of the following embodiments.

[0009] 1 is a schematic overall configuration diagram of a scanning electron microscope having an inspection and measurement model according to a first embodiment of the present invention. FIG. 1 is a functional block diagram of a computer shown in FIG. 1. FIG. 2 is a diagram showing a program stored in a storage medium shown in FIG. 2. FIG. 3 is a flowchart showing an operation of creating an inspection and measurement model (learning and correction) using a scanning electron microscope having the inspection and measurement model according to the first embodiment. FIG. 4 is a flowchart showing an operation of measurement using a scanning electron microscope having the inspection and measurement model according to the first embodiment. FIG. 5 is a diagram showing a pattern shape formed on a sample and an electron microscope image thereof. FIG. 6 is a diagram showing dimensional parameters for creating learning data. FIG. 7 is a diagram showing an example of a result of analyzing feature quantities. FIG. 8 is a diagram showing an example of calibration. FIG. 9 is a diagram showing an example of a calibration range expressed in feature quantity space. FIG. 10 is a diagram showing an example of a calibration range defined by dimension values. FIG. 11 is an explanatory diagram of deviation from the calibration range. FIG. 12 is a diagram showing the relationship between the calibration range and the learning range. FIG. 13 is an explanatory diagram of deviation from the learning range. FIG. 14 is a flowchart of creating an inspection and measurement model taking deviation from the learning range into consideration using a scanning electron microscope having the inspection and measurement model according to the first embodiment. FIG. 15 is a measurement flowchart taking deviation from the learning range into consideration using a scanning electron microscope having the inspection and measurement model according to the first embodiment. FIG. 16 is a diagram showing an example of measurement data deviating from the calibration range. FIG. 17 is a diagram showing an example of a screen display (GUI) of the display device shown in FIG. 2. FIG. 10 is a diagram showing an example in which measurement data deviates from a learning range in a scanning electron microscope having an inspection and measurement model according to Example 2 of the present invention.

[0010] The following description of the embodiments will be made with reference to the drawings. Note that the embodiments described below do not limit the scope of the invention as claimed, and not all of the elements and combinations thereof described in the embodiments are necessarily essential to the solution of the invention. The following description of the embodiments of the present invention will be made with reference to the drawings.

[0011] 1 is a schematic diagram of the overall configuration of a scanning electron microscope having an inspection and measurement model according to Example 1 of the present invention. As shown in Fig. 1, the scanning electron microscope 1 having the inspection and measurement model includes a scanning electron microscope 100, a control device 120, a power supply device 121, a power supply device 122, and a computer 200.

[0012] A scanning electron microscope 100 irradiates a sample 108 with an electron beam (electron beam) 103. The scanning electron microscope 100 outputs a detection signal obtained based on the irradiation of the electron beam 103. A scanning electron microscope 1 having an inspection and measurement model includes components necessary for forming a signal waveform and an image based on the detection signal from the scanning electron microscope 100. First, an example of the scanning electron microscope 100 will be specifically described with reference to FIG. 1 .

[0013] An electron beam 103 extracted from an electron source 101 by an extraction electrode 102 is accelerated by an acceleration electrode (not shown). The accelerated electron beam 103 is focused by a condenser lens 104, which is a type of focusing lens. The focused electron beam 103 is scanned one-dimensionally or two-dimensionally on a sample 108 by a scanning electrode 105. The electron beam 103 is decelerated by a negative voltage (retarding voltage) applied to an electrode built into a sample stage 109, and is focused by the lens action of an objective lens 106, and is irradiated onto the sample 108.

[0014] When the electron beam 103 is irradiated onto the sample 108, it scatters from the irradiated point (irradiation point) into the interior of the sample 108 and is emitted as electrons 110 such as secondary electrons (SE) and backscattered electrons (BSE). The emitted electrons 110 are accelerated toward the electron source 101 by an acceleration action based on a negative voltage (retarding voltage) applied to the sample 108, and collide with a conversion electrode 112, generating secondary electrons 111. The secondary electrons 111 emitted from the conversion electrode 112 are captured by a detector 113, and a detection signal, which is an output of the detector 113, changes depending on the amount of the captured secondary electrons 111.

[0015] The detection signal output from the detector 113 is supplied to the computer 200 by the control device 120. The computer 200 is equipped with a display device, which will be described later. The brightness of the image displayed on this display device changes according to the detection signal. That is, the amount of electrons captured by the detector 113 (electron quantity) is displayed as brightness on the display device. For example, when a two-dimensional image is displayed on the display device, the deflection signal supplied to the scanning electrode 105 and the detection signal output from the detector 113 are synchronized, so that the brightness of the image in the scanning area scanned by the deflection signal is displayed on the display device.

[0016] The scanning electron microscope 100 shown in FIG. 1 also includes, but is not limited to, a deflector (not shown) that moves the scanning area of ​​the electron beam 103. This deflector is used to display images of patterns of the same shape that exist at different positions on a display device. This deflector is also called an image-shift deflector, and can move the field of view (FOV) position of the scanning electron microscope 100 without moving the sample 108 using a sample stage (e.g., a sample table) 109 that moves the sample 108. The image-shift deflector and the scanning electrode 105 may be a common deflector, and an image-shift signal and a deflection signal may be superimposed and supplied to the deflector.

[0017] Detection signals (image, brightness profile, brightness, etc.) from the scanning electron microscope 100 are supplied to the computer 200 via the control device 120. The computer 200 calculates values ​​related to changes in the shape of the target shape of the inspection or measurement object based on the supplied detection signals, and outputs one or more of these calculated values. Note that the computer 200 may be integrated with the scanning electron microscope 100.

[0018] The control device 120 controls the power supply devices 121 and 122 in accordance with instructions from the computer 200. By controlling the power supply device 122, the voltage applied to the extraction electrode 102 and the acceleration electrode (not shown) changes. Similarly, by controlling the power supply device 121, the voltage (retarding voltage) applied to the sample 108 changes. Furthermore, the control device 120 controls the deflection signal supplied to the scanning electrode 105 and the signal supplied to the objective lens 106 in accordance with instructions from the computer 200. Furthermore, the control device 120 supplies the detection signal output from the detector 113 to the computer 200, as described above.

[0019] <Computer Configuration> Next, the computer 200 constituting the scanning electron microscope 1 having the inspection and measurement model according to this embodiment will be described with reference to the drawings. Fig. 2 is a functional block diagram of the computer shown in Fig. 1. As shown in Fig. 2, the computer 200 constituting the scanning electron microscope 1 having the inspection and measurement model according to this embodiment includes a pseudo-SEM waveform calculation unit 201, a feature amount analysis unit 202, a learning unit 203, a calibration unit 204, a measurement unit 205, a memory unit 206, a communication I / F 207, an input / output I / F 208, a storage medium 223, and an inspection and measurement model 230, which are interconnected via an internal bus 209. In addition, an input / output device 221 and a display device 222 are connected via the input / output I / F 208. Here, the pseudo-SEM waveform calculation unit 201, the feature analysis unit 202, the learning unit 203, the calibration unit 204, and the measurement unit 205 constitute a processing unit, which is realized by, for example, a processor such as a CPU (not shown), a ROM for storing various programs, a RAM for temporarily storing data in the calculation process, and a storage device such as an external storage device, and the processor such as a CPU reads and executes the various programs stored in the ROM, and stores the calculation results, which are the execution results, in the RAM, the external storage device, or cloud storage via a network connection, etc.

[0020] The input / output device 221 is, for example, a mouse, keyboard, etc., and is used by the user to input data, instructions, etc. to the processing unit. The display device 222 is used to display data, etc. obtained by the processing unit. The memory unit 206 is used to store data, etc. when the processing unit executes a program. The processing unit reads and executes a program stored in the storage medium 223. Figure 3 is a diagram showing the programs stored in the storage medium shown in Figure 2. The storage medium 223 stores a pseudo-SEM waveform calculation program 211, a feature analysis program 212, a learning program 213, a calibration program 214, and a measurement program 215 in predetermined areas.

[0021] A pseudo SEM (Scanning Electron Microscope) waveform calculation unit 201 accesses a storage medium 223 via an internal bus 209, and reads and executes a pseudo SEM waveform calculation program 211. The pseudo SEM waveform calculation unit 201 receives as input a simulation model including, for example, information on the pattern shape and material, information on incident electrons such as the acceleration voltage, and information on the detector arrangement and detection energy filter, and outputs a pseudo SEM image or waveform by, for example, Monte Carlo calculation. The shape of the simulation model, the pseudo SEM image, and the waveform will be described in detail when explaining FIG. 6.

[0022] The feature amount analysis unit 202 accesses the storage medium 223 via the internal bus 209, and reads and executes the feature amount analysis program 212. The feature amount analysis unit 202 receives, via the internal bus 209, an SEM image or waveform captured by the scanning electron microscope 100, or an image waveform generated by the pseudo-SEM waveform calculation unit 201, and outputs feature amounts fi (i = 1 to n) of the image or waveform. The feature amounts fi are values ​​that capture the characteristics of the waveform, such as the maximum brightness of the image, the peak brightness of the waveform, or the width of the peak.

[0023] The learning unit 203 accesses the storage medium 223 via the internal bus 209 and reads and executes the learning program 213. The learning unit 203 receives, via the internal bus 209, for example, the pattern shape used in the calculation by the pseudo SEM waveform calculation unit 201 and the feature values ​​fi obtained by analyzing the output of the pseudo SEM waveform calculation unit 201 with the feature value analysis unit 202. Then, using the feature values ​​fi as input, the learning unit 203 generates an inspection and measurement model 230 that outputs the dimensions of the pattern shape through learning, and stores the inspection and measurement model 230 in, for example, a recording medium (not shown). Note that the inspection and measurement model 230 may be configured to be stored in the storage medium 223.

[0024] The calibration unit 204 accesses the storage medium 223 via the internal bus 209 and reads and executes the calibration program 214. The calibration unit 204 receives, via the internal bus 209, the feature values ​​fi obtained by analyzing, in the feature value analysis unit 202, SEM images and waveforms of a calibration sample, the dimensions of which pattern shape are known, captured by the scanning electron microscope 100, the dimensions of the pattern shape of the calibration sample, and the inspection and measurement model 230. Then, the calibration unit 204 performs calibration so that the output of the inspection and measurement model 230 matches the dimensions of the pattern shape of the calibration sample, and corrects the inspection and measurement model 230. Details of the calibration will be described in detail when explaining FIG. 9.

[0025] The measurement unit 205 accesses the storage medium 223 via the internal bus 209, and reads and executes the measurement program 215. The measurement unit 205 receives as input an inspection and measurement model 230 and feature values ​​fi obtained by analyzing, in the feature value analysis unit 202, SEM images and waveforms of the sample 108 ( FIG. 1 ) to be inspected or measured using the scanning electron microscope 100, and outputs the dimensions of the sample 108 ( FIG. 1 ).

[0026] The pseudo SEM waveform calculation unit 201, the feature analysis unit 202, the learning unit 203, the calibration unit 204, and the measurement unit 205 may all or partly be implemented by a computer other than the computer 200, and the results may be output to the computer 200.

[0027] <Inspection and Measurement Model Creation (Learning and Calibration) Flow> Figure 4 is a flowchart showing the operation of inspection and measurement model creation (learning and correction) using a scanning electron microscope having an inspection and measurement model according to this embodiment. As shown in Figure 4, in step S301, a simulation model and dimensional parameters are input via the input / output device 221 constituting the computer 200. The input / output device 221 transfers the input simulation model and dimensional parameters to the pseudo SEM waveform calculation unit 201 via the input / output I / F 208 and the internal bus 209. In step S302, the pseudo SEM waveform calculation unit 201 constituting the computer 200 performs a simulation in accordance with the simulation model and dimensional parameters transferred from the input / output device 221 to generate a pseudo SEM image. The pseudo SEM waveform calculation unit 201 transfers the generated pseudo SEM image to the feature amount analysis unit 202 via the internal bus 209. In step S303, the feature amount analysis unit 202 constituting the computer 200 analyzes the feature amounts of the pseudo SEM image transferred from the pseudo SEM waveform calculation unit 201. The feature amount analysis unit 202 transfers the analyzed feature amounts fi and dimensional parameters to the learning unit 203 via the internal bus 209. In step S304, the learning unit 203 constituting the computer 200 creates learning data that is a set of the feature amounts fi and dimensional parameters transferred via the internal bus 209. In step S305, the learning unit 203 creates an inspection and measurement model 230 by learning based on the learning data. Steps S302 to S305 are collectively referred to as a learning step (S311).

[0028] Once the real sample is obtained in step S306, it is imaged with the scanning electron microscope 100 in step S307. If the dimensions of the real sample are unknown, the correct values ​​for the dimensions are obtained using another device, such as a cross-sectional transmission electron microscopy (TEM) (step S308). If the dimensions of the real sample are known or if design information is used as the correct values, step S308 can be omitted. The image results from the scanning electron microscope 100 in step S307 are input from the control device 120 via the communication I / F 207. The communication I / F 207 transfers the image results from the scanning electron microscope 100 to the feature amount analysis unit 202 via the internal bus 209. The feature amount analysis unit 202 transfers the results of the feature amount analysis to the calibration unit 204 via the internal bus 209. In step S309, the calibration unit 204 compares the dimensions obtained by inputting the feature analysis results transferred from the feature analysis unit 202 into the inspection and measurement model 230 with the correct values ​​of the dimensions, and corrects the inspection and measurement model. The calibration unit 204 then outputs the corrected measurement model 230 and the calibration range 330. Steps S306 to S309 are collectively referred to as a calibration step (step S310).

[0029] 5 is a flowchart showing the operation of measurement by a scanning electron microscope having an inspection and measurement model according to this embodiment. The scanning electron microscope 100 performs measurement using the inspection and measurement model 230 and calibration range 330 created in the inspection and measurement model creation flow.

[0030] As shown in FIG. 5 , in step S401, the scanning electron microscope 100 captures an image of the sample 108 ( FIG. 1 ) to be inspected or measured. In step S402, the feature amount analysis unit 202 inputs the captured SEM image and waveform via the control device 120, the communication I / F 207, and the internal bus 209, and performs feature amount analysis. Thereafter, in step S403, the feature amount analysis results by the feature amount analysis unit 202 are transferred via the internal bus 209 to the measurement unit 205, which outputs measurement values. Thereafter, in step S404, the calibration unit 204 inputs the feature amount analysis results into the inspection measurement model 230, compares the resulting measurement values ​​or feature amount analysis results with the calibration range 330, and determines whether the calibration range 330 is exceeded. Details of the calibration range 330 will be described later. If the determination results in an excess of the calibration range 330, an alarm is issued (step S405) to prompt adjustment (step S406). The adjustment (step S406) refers to re-executing the calibration step (step S310) using the sample 108 ( FIG. 1 ) to be inspected or measured when the alarm was issued. On the other hand, if there is no deviation, a message indicating that the calibration is within the range is displayed (step S407). <Examples of Samples, SEM Images / Profiles, and Feature Amounts> Next, examples of pattern shapes formed on the sample 108 and examples of electron microscope images obtained when the pattern shapes are imaged (observed) with the scanning electron microscope 100 will be described with reference to the drawings. FIG. 5 shows the pattern shapes formed on the sample and their electron microscope images. Here, the upper left diagram of FIG. 5 shows a perspective view of the pattern shapes, and the upper right diagram of FIG. 5 shows the electron microscope image. The lower left diagram of FIG. 5 shows a cross-sectional view of the pattern shapes shown in the upper left diagram of FIG. 5, and the lower right diagram of FIG. 5 shows the brightness profile (change in brightness due to change in coordinates) in the electron microscope image shown in the upper right diagram of FIG. 5.

[0031] Here, a semiconductor wafer used in the manufacture of semiconductor devices will be described as an example of the sample 108. Because it is a semiconductor wafer, the material of the sample 108 is silicon (Si). By etching the sample 108, a pattern shape 500 as shown in the upper left diagram of FIG. 5 is formed. Here, an L / S (Line and Space) having a bowing shape will be described as an example of the pattern of the pattern shape 500.

[0032] In the upper left diagram of Fig. 5, UP indicates the main surface (first surface) of a line L of the pattern shape 500, and SD indicates the sidewall (second surface) of the line L. Furthermore, DW indicates the back surface of the sample 108 facing the main surface UP. The electron microscope image 510 in the upper right diagram of Fig. 5 is an image captured by the scanning electron microscope 100 from vertically above the sample 108. That is, the electron microscope image 510 is an image captured by irradiating the sample 108 with an electron beam 103 from vertically above the main surface UP.

[0033] The lower left diagram of Fig. 5 shows a cross section 502 of the pattern shape 500 at a cross section position 501 in the pattern shape 500 shown in the upper left diagram of Fig. 5. The lower right diagram of Fig. 5 shows a brightness profile 512 at a cross section position 511 in the electron microscope image 510 shown in the upper right diagram of Fig. 5. The feature amount fi is, for example, a pattern top brightness 513 at the pattern top 503, a peak brightness 514 which is the brightness in the vicinity of the pattern edge 504, a profile width 515, etc.

[0034] <Details of Learning and Calibration> Next, details of learning and calibration will be described. FIG. 7 is a diagram showing dimension parameters for creating learning data. Dimension parameter sets 602 are prepared for a range larger than the expected range 601 of dimension fluctuation ranges during mass production for dimensions A and B. Here, an example is shown in which 6 × 5 = 30 dimension parameter sets 602 are prepared. For the sake of illustration, only dimensions A and B are shown, but other dimensions may also be varied to prepare more dimension parameter sets 602. Also, dimension parameter sets 602 may be prepared at non-equidistant intervals in parameter space. Here, the range of dimension parameters of the learning data is referred to as a learning range 603. Note that dimensions A and B in FIG. 7 are, for example, the height of the pattern, and dimension B is the width of the top of the pattern. However, this is not limited to this.

[0035] FIG. 8 shows an example of the results of generating a pseudo-SEM image using the dimension parameter set 602 in FIG. 7 and performing feature analysis. As shown in FIG. 8, a feature data set 702 is obtained corresponding to the dimension parameter set 602. At this time, the internal region connecting the outermost points in the feature space is referred to as a learning range 703. Learning is performed, for example, by linear or nonlinear regression of the feature values ​​fi. At this time, the inspection and measurement model 230 is expressed as a nonlinear combination of weighted feature values, for example, as shown in the following equation (1):

[0036] In this case, n is the number of features, and a ０ , a ｉｊ is a coefficient (constant value), f ｉ ｊ is a feature quantity, and N is the order of the nonlinear combination to be considered. An example when n = 3 and N = 1 is shown in the following formula (2), and an example when n = 2 and N = 2 is shown in the following formula (3). Note that instead of formula (1), it may be expressed by a more complex polynomial, or a method that does not use a formula, such as a deep learning model, may be used.

[0037]

[0038] An example of calibration will be described using FIG. 9 . For simplicity, an example of a linear combination of equations (2) is shown. Assume that the learning unit 203 ( FIG. 2 ) constituting the computer 200 has learned learning data 801 and obtained an inspection measurement model 801. Plot 803 is a plot of the estimated value p and the correct value of dimension A when measuring a calibration sample. In the left diagram of FIG. 9 , only the offset is corrected while maintaining the slope of the line, resulting in a corrected inspection measurement model 804. In the right diagram of FIG. 8 , the slope and offset of the line are corrected, resulting in a corrected inspection measurement model 805. While a simple case of linear combination is shown here, this is referred to as correcting the inspection measurement model (sometimes referred to as calibration) from the actual measurement data and correct value of the calibration sample. However, in addition to correcting the inspection measurement model, corrections can also be made to convert the feature value fi to a special value fi' using an arbitrary function. Generally, higher-order corrections require measurement data from a larger number of calibration samples.

[0039] The calibration range, which is created simultaneously with calibration, will now be described. FIG. 10 is a diagram showing an example of a calibration range expressed in feature space. The left diagram of FIG. 10 shows an example in which the range 901 of each feature value fi when a calibration sample is measured is defined as the calibration range. The right diagram of FIG. 10 shows an example in which a calibration range 902 is defined based on the relationship between multiple feature values ​​from the ranges of existence between multiple feature values ​​when a calibration sample is measured. Note that the right diagram of FIG. 10 is shown as a two-dimensional graph for illustrative purposes, but the calibration range 902 may be defined in a higher-order multidimensional space. Furthermore, the calibration range may be defined not only in feature space but also by dimension values. The left diagram of FIG. 11 shows an example in which a calibration range 1001 of a measurement value p is defined by the maximum and minimum values ​​of the correct dimension of the calibration sample. Furthermore, as shown in the right diagram of FIG. 11, the calibration range 1002 may be defined by multiple dimensions (pi and pj). Note that while FIG. 11 is shown as a two-dimensional graph for illustrative purposes, the calibration range may be defined in a higher-order multidimensional space.

[0040] <Regarding deviation from calibration range in measurement> Fig. 12 is an explanatory diagram of deviation from the calibration range. Here, the description will be given using the calibration range 902 and the calibration range 1002 in the right diagram of Fig. 10 and the right diagram of Fig. 11. Deviation from the calibration range means that the feature value f or the estimated value p obtained from an image of the sample 108 to be inspected or measured is outside the calibration range 902 or the calibration range 1002. In Fig. 12, the measurement data 1101 and the measurement data 1111 deviate from the calibration range (902, 1111), while the measurement data 1102 and the measurement data 1112 do not deviate from the calibration range (902, 1111).

[0041] Up to this point, for the sake of simplicity, deviations from the calibration range have been described, but there are cases in which the data obtained by measuring the sample 108 deviates not only from the calibration range but also from the learning range, in which case there is a high possibility that a measurement error greater than that caused by deviations from the calibration range will occur. Below, deviations from the learning range will be described, and at the end, a flowchart including deviations from the learning range will be shown.

[0042] 13 is a diagram showing the relationship between the calibration range 902 and the learning range 703. If the data acquired during calibration exceeds the learning range 703, it is necessary to review the dimension parameters in step S301 (FIG. 4) and execute the learning step S311 (FIG. 4) again. Therefore, during measurement, the calibration range 902 is included in the learning range 703.

[0043] 14 is an explanatory diagram of deviation from the learning range 703. Deviation from the learning range 703 means that the feature value f and the estimated value p obtained from an image of the sample 108 to be inspected or measured are outside the learning range 703 and / or the learning range 603 ( FIG. 7 ). In FIG. 14 , measurement data 1301 deviates from the learning range 703, and measurement data 1302 deviates from the calibration range 902, but does not deviate from the learning range 703. Measurement data 1303 does not deviate from the calibration range 902 or the learning range 703.

[0044] <Flow of Inspection and Measurement Model Creation (Learning and Calibration) Taking Deviation from the Learning Range Into Account> Figure 15 is a flowchart of the creation of an inspection and measurement model taking deviation from the learning range into account using a scanning electron microscope having an inspection and measurement model according to this embodiment. Here, only the differences from Figure 4 will be explained. After step S304 when creating learning data, the learning unit 203 (Figure 2) constituting the computer 200 creates a learning range (step S1401) in parallel with step S305, and creates a learning range 1401. Steps S302 to S305 and step S1401 are collectively referred to as learning step S1402.

[0045] <Measurement Flow Taking Deviation from the Learning Range into Account> FIG. 16 is a measurement flowchart taking deviation from the learning range into account using a scanning electron microscope having an inspection and measurement model according to this embodiment. Here, only the differences from FIG. 5 will be described. The calibration unit 204 determines whether the calibration range 330 has been deviated (step S404). If it has been determined that the calibration range 330 has been deviated, it further determines whether the calibration range 1401 has been deviated (step S1501). If not, an alarm indicating deviation from the calibration range 330 is issued (step S405) and adjustment (step S406) is prompted. On the other hand, if deviation from the learning range 1401 has been detected, an alarm indicating deviation from the learning range 1401 is issued (S1503) and re-learning (step S1503) is prompted. Here, re-learning refers to reviewing the simulation model and dimensional parameters of step S301 ( FIG. 15 ) and re-executing the learning step of step S1402 ( FIG. 15 ).

[0046] Fig. 17 is a diagram showing an example in which measurement data deviates from the calibration range. As shown in the left diagram of Fig. 17, if a measurement data group 1601 deviates from the calibration range 902, an alarm is issued. At this time, as shown in the right diagram of Fig. 17, the distribution of the calibration range 902 and the measurement data group 1601 is analyzed, and additional calibration points are selected so that the entire measurement data point group 1601 is included. Here, measurement data 1602, measurement data 1603, and measurement data 1604 are used as additional calibration data to redefine a new calibration range 1605.

[0047] When performing calibration, it is necessary to measure the location where measurement data was acquired using a cross-sectional TEM, an AFM (Atomic Force Microscope), etc. As in this embodiment, the calibration range 902 and the measurement data group 1601 are analyzed and adjustment is made using a small number of calibration points, thereby reducing the effort required to acquire correct data.

[0048] Fig. 18 is a diagram showing an example of a screen display (GUI: Graphical User Interface) of the display device shown in Fig. 2. As shown in Fig. 18, the GUI is composed of a group of execution buttons 1710 for each program, a setting area 1720 for setting the simulation model and parameters of the pseudo SEM waveform calculation program 211, a setting area 1730 for the feature amount analysis program 212, a setting area 1740 for the learning program 213, a setting area 1750 for the calibration program 214, a setting area 1760 for the measurement program 215, and a display area 1770.

[0049] In the simulation model and parameter setting area 1720, the shape of the model and the setting values ​​of the model parameters are input. In the feature analysis program setting area 1730, which part of the waveform to extract feature values ​​from is specified on the display, or a feature calculation formula (not shown on the GUI) is defined. In the learning program setting area 1740, a learning method such as regression or deep learning is selected, and the model to be used, for example, linear or nonlinear, is selected from a pull-down menu or defined using a formula. In the calibration program setting area 1750, a calibration method, such as offset, for the model selected in the learning program setting area 1740 is selected from a pull-down menu (see the explanation of FIG. 9 for other methods). In the calculation program setting area 1760, a model to be used is selected from a pull-down menu from multiple models defined for each pattern shape (such as line and space or hole) and each representative dimension of the pattern.

[0050] Display area 1770 displays the execution results of various programs. FIG. 18 shows an example of the execution of measurement program 215, displaying measurement values ​​1771 and a warning 1772 indicating deviation from the calibration range. Additional calibration data used for adjustment, for example, indicates which chip on wafer map 1773 should be remeasured, and measurement coordinate information 1774 at that time is displayed. Note that chips 1775 to 1777 on wafer map 1773 correspond to additional calibration data 1602 to 1604 (the right diagram in FIG. 17). The user performs adjustments based on these displays. Note that, although described in terms of feature space, the distribution of the calibration range and the measurement data group may also be compared in the space of estimated values ​​(measurement values) shown in the right diagram in FIG. 11.

[0051] When performing calibration, it is necessary to measure the location where measurement data was acquired using a cross-sectional TEM, AFM (Atomic Force Microscope), etc. As in this embodiment, the calibration range 902 and the measurement data group 1601 (left diagram in FIG. 17 ) are analyzed and adjustment is made using a small number of calibration points, thereby reducing the effort required to acquire correct data.

[0052] In this embodiment, learning and calibration are described separately, but it is also possible to prepare a large number of calibration samples as learning data, acquire SEM images of the calibration samples to create a feature data set 702 ( FIG. 8 ), measure the calibration samples using, for example, a cross-sectional TEM or AFM to acquire a dimension parameter set 602 ( FIG. 7 ), and use the dimension parameter set 602 ( FIG. 7 ) and the feature data set 702 ( FIG. 8 ) to learn equations (1) to (3) and a deep learning model. In this case, the learning range and the calibration range are equal.

[0053] Furthermore, in this embodiment, dimensional measurement has been described as an example, but the monitoring (alarm) of deviations from the learning range or calibration range and the fixed timing of observation points for efficient adjustment can be widely used in processes using mathematical formulas or models, and are not limited to dimensional measurement. They can also be applied to defect inspection, defect classification, pattern recognition, and image conversion (image quality improvement, high magnification estimation).

[0054] As described above, according to this embodiment, it is possible to provide a scanning electron microscope having an inspection measurement model that can be efficiently adjusted while making the user aware of a decrease in the reliability of the inspection or measurement results, and a method for correcting the inspection measurement model.

[0055] In this embodiment, an example of a computer 200 will be described that not only issues an alarm when the learning range is deviated, but also proposes a learning range suitable for relearning.

[0056] The above-described flowchart of creating an inspection and measurement model taking into account deviations from the learning range in Fig. 15 and the measurement flowchart taking into account deviations from the learning range in Fig. 16 are the same as those in Example 1. The difference is the operation when an alarm is issued in step S1502 (Fig. 16).

[0057] Figure 19 is a diagram showing an example in which measurement data deviates from the learning range in a scanning electron microscope having an inspection and measurement model according to this embodiment. As shown in the upper left diagram of Figure 19, when an alarm regarding deviation from the learning range is issued in step S1502 (Figure 16), i.e., when measurement data group 1801 is located outside learning range 703, a dimension group 1811 estimated from measurement data group 1801 shown in the upper right diagram of Figure 19 is analyzed to determine where in learning range 603 in the space of dimension parameters a dimension group 1811 is located, and a new learning range 1812 is defined so as to include dimension group 1811. Then, as shown in the lower diagram of Figure 19, a dimension parameter set 1813 is proposed for learning the new learning range 1812.

[0058] This is possible because the learning model can sometimes estimate (i.e., extrapolate) dimensions even outside the learning range, and in such cases, the results of estimating dimensions outside the learning range can be used to identify a new learning range.

[0059] According to this embodiment, in addition to the effect of the first embodiment, it is easy to know how much simulation data should be added when re-simulating.

[0060] The present invention is not limited to the above-described embodiments, but includes various modifications. For example, the above-described embodiments have been described in detail to clearly explain the present invention, and the present invention is not necessarily limited to those including all of the described configurations. Furthermore, it is possible to replace part of the configuration of one embodiment with the configuration of another embodiment, and it is also possible to add the configuration of another embodiment to the configuration of one embodiment.

[0061] Furthermore, each of the above configurations may be realized in hardware, for example, by designing a part or all of them as an integrated circuit. Furthermore, each of the above configurations may be realized in software, with a processor interpreting and executing a program that realizes each function. Information such as the program, table, and file that realizes each function can be stored in a memory, a recording device such as a hard disk or SSD (Solid State Drive), or a recording medium such as an IC card, SD card, or DVD.

[0062] In addition, the control lines and information lines shown are those that are considered necessary for the explanation, and do not necessarily show all the control lines and information lines in the product. In reality, it can be assumed that almost all components are interconnected.

[0063] 1...Scanning electron microscope with inspection and measurement model, 100...Scanning electron microscope, 101...Electron source, 102...Extraction electrode, 103...Electron beam, 104...Condenser lens, 105...Scanning electrode, 106...Objective lens, 108...Sample, 109...Sample stage, 110...Electrons, 111...Secondary electrons, 112...Conversion electrode, 113...Detector, 120...Control device, 121, 122...Power supply device, 200...Computer, 201...Pseudo SEM waveform calculation unit, 202...Feature amount analysis unit, 203...Learning unit, 204...Calibration unit, 205...Measurement unit, 206...Memory unit, 207...Communication I / F, 208...Input / output I / F, 209...Internal bus, 211...Pseudo SEM waveform calculation program, 212...Feature amount analysis program, 213...Learning program, 214...Calibration program, 215...Measurement program, 2 21...input / output device, 222...display device, 223...storage medium, 230...inspection measurement model, 330, 902, 1002...calibration range, 500...pattern shape, 501...cross-sectional position, 502...cross-section, 510...electron beam microscope image, 511...cross-sectional position, 512...brightness profile, 601...assumed area, 602...dimension parameter set, 603, 703...learning range, 702...feature amount data set, 901...range of each feature amount fi (calibration range), 902...calibration range defined by the relationship between multiple feature amounts, 1001...calibration range of measurement value p, 1002...calibration range defined by multiple dimensions (pi and pj), 1101, 1102, 1111, 1112...measurement data, 1801...measurement data group, 1811...estimated dimension group, 1812...new learning range, 1813...dimension parameter set

Claims

1. A scanning electron microscope with an inspection and measurement model, comprising: a feature analysis unit that inputs an electron microscope image or waveform of a sample to be inspected or measured to determine a first feature, and inputs an electron microscope image or waveform of a calibration sample to determine a second feature; a learning unit that generates an inspection and measurement model by learning based on the second feature; and a calibration unit that compares the first feature with a predetermined calibration range based on the first feature and the inspection and measurement model, and issues an alarm if the first feature deviates from the calibration range.

2. A scanning electron microscope having an inspection and measurement model as described in claim 1, characterized in that the calibration unit calibrates the output of the inspection and measurement model so that it matches the second feature amount, and corrects the inspection and measurement model.

3. A scanning electron microscope having an inspection and measurement model as described in claim 1, characterized in that the calibration unit issues an alarm when the first feature deviates from the calibration range and deviates from a learning range that includes the specified calibration range or is wider than the specified calibration range.

4. A scanning electron microscope having an inspection and measurement model as described in claim 1, characterized in that, when the first feature deviates from the calibration range, the calibration unit corrects the inspection and measurement model by correcting the calibration range so as to include the first feature.

5. A scanning electron microscope having an inspection and measurement model as described in claim 3, characterized in that, when the first feature deviates from the calibration range and deviates from a learning range that includes the specified calibration range or is wider than the specified calibration range, the calibration unit corrects the learning range to include the first feature.

6. A scanning electron microscope having an inspection and measurement model as described in claim 4, characterized in that the calibration unit corrects the learning range to include the first feature when the first feature deviates from the calibration range and deviates from a learning range that includes the specified calibration range or is wider than the specified calibration range.

7. A scanning electron microscope having an inspection and measurement model according to claim 5, characterized in that the comparison between the first feature value and the calibration range is performed by analyzing a distance in a feature value space.

8. A scanning electron microscope having an inspection and measurement model as described in claim 7, comprising a display device, the display device having on its screen at least an area for displaying an alarm, an area for visually suggesting positions within a sample to be recalibrated for inspection or measurement, and an area for selecting a learning method and model.

9. A method for correcting an inspection measurement model of a scanning electron microscope having an inspection measurement model, comprising the steps of: a feature analysis unit inputting an electron microscope image or waveform of a sample to be inspected or measured to obtain a first feature, and inputting an electron microscope image or waveform of a calibration sample to obtain a second feature; a learning unit generating an inspection measurement model by learning based on the second feature; and a calibration unit calibrating the output of the inspection measurement model so that it matches the second feature, thereby correcting the inspection measurement model.

10. A method for correcting an inspection measurement model as described in claim 9, characterized in that the calibration unit has a step of issuing an alarm when the first feature deviates from a predetermined calibration range and deviates from a learning range that includes the predetermined calibration range or is wider than the predetermined calibration range.

11. A method for correcting an inspection measurement model as described in claim 9, characterized in that the calibration unit has a step of correcting the inspection measurement model by correcting the calibration range to include the first feature when the first feature deviates from the calibration range.

12. A method for correcting an inspection measurement model as described in claim 10, characterized in that the calibration unit has a step of correcting the learning range to include the first feature when the first feature deviates from the calibration range and deviates from a learning range that includes the specified calibration range or is wider than the specified calibration range.

13. A method for correcting an inspection measurement model as described in claim 11, characterized in that the calibration unit has a step of correcting the learning range to include the first feature when the first feature deviates from the calibration range and deviates from a learning range that includes the specified calibration range or is wider than the specified calibration range.

Images