Charged particle beam device and sample inspection method

Abstract

A charged particle beam device 1 includes: a charged particle gun 2 capable of irradiating a sample 9 with an electron beam EB1; and a detector 11 for detecting emitted electrons EB2 that are emitted when the sample 9 is irradiated with the electron beam EB1. Each of a plurality of specific locations included in the sample 9 are point-irradiated by the electron beam EB1, the emitted electrons EB2 emitted as a result of the point-irradiation are detected, a transient response waveform of the emitted electrons EB2 is measured, and one or more features are extracted from the transient response waveform. The plurality of specific locations are determined to be normal locations or defective locations on the basis of the frequency of a first feature or relationship information relating to a plurality of features, among the one or more features.

Description

Charged particle beam device and sample inspection method

[0001] The present invention relates to a charged particle beam device and an inspection method, and more particularly to a charged particle beam device and a sample inspection method for irradiating a sample with a charged particle beam and inspecting the sample for defects.

[0002] Charged particle beam instruments such as electron microscopes and ion microscopes are used to observe various samples with fine structures. For example, for the purpose of process control in the manufacturing process of semiconductor devices, a scanning electron microscope (SEM), which is one type of charged particle beam instrument, is used to measure the dimensions or inspect defects of semiconductor device patterns formed on semiconductor wafers.

[0003] One known method for inspecting a sample using an electron microscope involves irradiating the sample with a pulsed electron beam, detecting electrons emitted from the sample, forming a voltage contrast image based on the detected emitted electrons, and evaluating the electrical characteristics of semiconductor elements formed on the sample based on an analysis of the voltage contrast image.

[0004] For example, Patent Document 1 discloses a method for determining whether a sample contains both defect-free normal portions and defective portions by using a reference waveform previously obtained from the normal portions and a measured waveform of the time change of emitted electrons obtained by irradiating the sample with an electron beam. The measured waveform and the reference waveform are compared, and if the measured waveform differs from the reference waveform, the measured portion can be determined to be a defective portion.

[0005] Patent Document 2 discloses a method of acquiring a plurality of observation images and deriving the electrical characteristics of a measured location based on the change in brightness of the plurality of observation images.

[0006] U.S. Patent No. 9,805,910

[0007] In Patent Document 1, it is assumed that normal portions can be recognized in advance and that a reference waveform for the normal portions can be prepared. However, if normal portions cannot be recognized in advance, there is a problem in that a reference waveform cannot be generated.

[0008] In Patent Document 2, since it is necessary to acquire a plurality of observation images, there is a problem that the inspection time is long.

[0009] There is a need for technology that can determine whether a measurement point is normal or defective, even when normal points cannot be identified in advance. Furthermore, if this determination can be made without acquiring an observation image, inspection time can be shortened.

[0010] Other objects and novel features will become apparent from the description of this specification and the accompanying drawings.

[0011] A charged particle beam device according to one embodiment includes a charged particle gun capable of irradiating a sample with a charged particle beam and a detector for detecting emitted electrons when the sample is irradiated with the charged particle beam. The device irradiates each of a plurality of specific locations included in the sample with the charged particle beam at a point location, detects the emitted electrons due to the point irradiation, measures a transient response waveform of the emitted electrons, and extracts one or more feature quantities from the transient response waveform. The device distinguishes the plurality of specific locations as normal or defective based on a frequency of a first feature quantity among the one or more feature quantities or relationship information of a plurality of the feature quantities.

[0012] A sample inspection method according to one embodiment includes the steps of: (a) preparing a sample including a plurality of specific locations; (b) point-irradiating one of the specific locations with a charged particle beam; (c) detecting emitted electrons emitted by the point irradiation; (d) measuring a transient response waveform of the emitted electrons; (e) extracting one or more feature quantities from the transient response waveform; and (f) performing steps (b), (c), (d), and (e) on the plurality of specific locations, and discriminating the plurality of specific locations as normal or defective based on the frequency of a first feature quantity among the one or more feature quantities or relationship information of a plurality of the feature quantities.

[0013] A charged particle beam device according to one embodiment includes a charged particle gun capable of irradiating a sample with a charged particle beam and a detector for detecting emitted electrons when the sample is irradiated with the charged particle beam. The device irradiates a plurality of specific locations included in the sample with the charged particle beam at a point, detects the emitted electrons due to the point irradiation, measures a transient response waveform of the emitted electrons, extracts one or more feature quantities from the transient response waveform, and estimates one or more electrical characteristics corresponding to the one or more feature quantities by referring to electrical characteristic information. The device distinguishes the plurality of specific locations as normal or defective based on the frequency of a first electrical characteristic among the one or more electrical characteristics or on relationship information between the plurality of electrical characteristics.

[0014] A sample inspection method according to one embodiment includes the steps of: (a) preparing a sample including a plurality of specific locations; (b) point-irradiating one of the specific locations with a charged particle beam; (c) detecting electrons emitted by the point irradiation; (d) measuring a transient response waveform of the emitted electrons; (e) extracting one or more feature quantities from the transient response waveform; (f) estimating one or more electrical characteristics corresponding to the one or more feature quantities by referring to electrical characteristic information; and (g) performing steps (b), (c), (d), (e), and (f) on the plurality of specific locations, and determining whether the plurality of specific locations are normal or defective based on the frequency of a first electrical characteristic among the one or more electrical characteristics or relationship information between the plurality of electrical characteristics.

[0015] According to one embodiment, even if it is not possible to identify normal portions in advance, it is possible to determine whether a measurement portion is normal or defective. Furthermore, since the above determination can be made without acquiring an observation image, the inspection time can be reduced.

[0016] 1 is a schematic diagram showing a charged particle beam device according to a first embodiment. FIG. 2 is a cross-sectional view showing an example of a cross section of a sample according to the first embodiment. FIG. 3 is a cross-sectional view showing a shape 1 of a transient response waveform when a single-pulse electron beam according to the first embodiment is used. FIG. 4 is a flow chart showing a sample inspection method according to the first embodiment. FIG. 5 is a schematic diagram showing an input screen displayed on a display device according to the first embodiment. FIG. 6 is a function obtained by regression analysis of shape 1 of the transient response waveform according to the first embodiment. FIG. 7 is a function obtained by regression analysis of shape 2 of the transient response waveform according to the first embodiment. FIG. 8 is a schematic diagram showing a GUI screen of statistical analysis results displayed on a display device according to the first embodiment. FIG. 9 is a transient response waveform with different shapes according to the first embodiment. FIG. 10 is a transient response waveform when a multiple-pulse electron beam according to the second embodiment is used. FIG. 11 is a flow chart showing a sample inspection method according to the third embodiment. FIG. 12 is a transient response waveform for each diameter of a plug according to the third embodiment. FIG. 13 is a schematic diagram showing an electrical characteristic estimation system according to the third embodiment. FIG. 14 is a cross-sectional view showing an example of a semiconductor device according to an application example. FIG. 15 is a cross-sectional view showing a defect pattern of a semiconductor device according to an application example. FIG. 16 is a flow chart showing a sample inspection method according to the third embodiment. 10 is a flowchart showing the results of measuring the electrical characteristics of a sample in an application example.

[0017] Hereinafter, embodiments will be described in detail with reference to the drawings. In all drawings for explaining the embodiments, components having the same functions are designated by the same reference numerals, and repeated explanations thereof will be omitted. In the following embodiments, explanations of the same or similar parts will not be repeated unless particularly necessary.

[0018] (First Embodiment) <Charged Particle Beam Apparatus> A charged particle beam apparatus 1 according to a first embodiment will be described below with reference to Fig. 1. The charged particle beam apparatus 1 is, for example, a scanning electron microscope (SEM).

[0019] 1 , the charged particle beam device 1 includes an electron gun (charged particle gun) 2, a condenser lens 3, a pulsed electron generator 4, an aperture 5, a deflector 6, an objective lens 7, a stage 8, a processor 10, a detector 11, a detection signal analyzer 12, a database 16, an input device 17, and a display device 18. The charged particle beam device 1 may also include a light source 13, an optical path blocker 14, an optical path 15, and an energy filter 24.

[0020] The stage 8 can carry a sample 9. When observing the sample 9, the sample 9 is placed on the stage 8. The stage 8 can move in a plane, vertically, rotate, and tilt under the control of the processor 10.

[0021] Note that a wafer on which various semiconductor devices are formed is exemplified as the sample 9 in the first embodiment. For example, the sample 9 includes at least a part of a semiconductor substrate on which a p-type or n-type impurity region is formed, a semiconductor element such as a MOSFET (Metal Oxide Semiconductor Field Effect Transistor) formed on the semiconductor substrate, and plugs and wiring formed on the semiconductor element.

[0022] The electron gun 2 is capable of irradiating an electron beam (charged particle beam) EB1 onto a sample 9. The condenser lens 3 is provided to converge the electron beam irradiated from the electron gun 2.

[0023] The pulsed electron generator 4 is provided to control the electron beam EB1 as a pulsed electron beam (pulsed charged particle beam). The pulsed electron generator 4 generates a pulsed electron beam by controlling a scanning deflector that scans the electron beam EB1 over the sample 9 and a blanking deflector that blocks the electron beam EB1 from irradiating the sample 9. While the scanning deflector is scanning the beam, the blanking electrode intermittently blocks the beam in accordance with set irradiation conditions, thereby generating a pulsed electron beam. Note that, although the pulsed electron generator 4 is provided separately from the electron gun 2 in this embodiment, an electron gun capable of irradiating a pulsed electron beam may also be used.

[0024] The aperture 5 blocks a portion of the electron beam EB1 and passes another portion of the electron beam EB1. The deflector 6 is provided above the sample 9 to two-dimensionally scan the electron beam EB1 over the sample 9. The objective lens 7 focuses the electron beam EB1 onto the sample 9 as a minute spot of several nanometers.

[0025] The detector 11 detects emitted electrons (secondary electrons) EB2 emitted when the electron beam EB1 is irradiated onto the sample 9. The detection signal analysis unit 12 outputs the emitted electrons EB2 detected by the detector 11 as a detection signal to the processor 10. The processor 10 has an arithmetic processing circuit that can generate an observation image based on the detection signal and can measure a transient response waveform.

[0026] The light source 13 can emit a laser beam LB to the sample 9. The emitted laser beam LB is controlled to an optical path 15 by an optical path blocker 14 and irradiated onto the sample 9. The energy filter 24 can discriminate the energy of the emitted electrons EB2 by adjusting the voltage of the energy filter 24. The energy of the emitted electrons EB2 at this time correlates with the surface potential Vs of the sample 9.

[0027] The processor 10 is an arithmetic processing device including a plurality of semiconductor circuits such as a CPU. The processor 10 is electrically connected to the electron gun 2, the condenser lens 3, the pulsed electron generator 4, the aperture 5, the deflector 6, the objective lens 7, the stage 8, the detector 11, the detection signal analyzer 12, the light source 13, the light path interrupter 14, the energy filter 24, the database 16, the input device 17, and the display device 18, and controls these components. That is, each control and operation performed in the charged particle beam device 1 is performed by the processor 10.

[0028] The processor 10 can also control the irradiation time, which is the time for continuous irradiation of the electron beam EB1, and the interruption time, which is the time between irradiations of the electron beam EB1.

[0029] The database 16 is a storage device including a magnetic disk or a semiconductor memory, and stores various types of information used in the charged particle beam device 1. The display device 18 displays various types of information used in the charged particle beam device 1 as a GUI (Graphical User Interface) screen. The input device 17 is a device through which a user inputs instructions such as inputting information about an observation target, changing irradiation conditions for the electron beam EB1, and changing the position of the stage 8. When a user inputs instructions for controlling the charged particle beam device 1 on the display device 18 using the input device 17, the instructions are transmitted to the processor 10.

[0030] 2 shows plugs 33 and 34, which are an example of an inspection target in the first embodiment. As shown in FIG. 2, an insulating film 31 and an interlayer insulating film 32 are formed on a semiconductor substrate 30. Plugs 33 and 34 are formed in the interlayer insulating film 32. Plug 34 is a low-resistance, low-capacitance plug that reaches the semiconductor substrate 30. Plug 33 does not reach the semiconductor substrate 30 and is in contact with the insulating film 31, so plug 33 is a high-resistance, high-capacitance plug.

[0031] 3 and 4 show an example of transient response characteristics of emitted electrons EB2 emitted from an observation target when the observation target is irradiated with an electron beam EB1 (pulse electron beam).

[0032] 5, the plugs 33 and 34 are treated as the observation target as an example. Also, an example is shown in which the observation target is irradiated once with the electron beam EB1 (single-pulse electron beam).

[0033] <Method of Inspecting Sample> A method of inspecting a sample 9 performed using the charged particle beam device 1 will be described below with reference to Fig. 5. The method of inspecting a sample 9 includes steps S1 to S8. In addition, when describing steps S1 to S8, Figs. 3, 4, and 6 to 9 will be used as necessary.

[0034] One of the observation methods using the charged particle beam device 1 is the voltage contrast method. In the voltage contrast method, the object of observation is charged by irradiation with the electron beam EB1, and the difference in the surface potential Vs of the object of observation is reflected as a difference in the signal intensity y of the emitted electrons. The voltage contrast method can be used, for example, to inspect for defects in the electrical characteristics of semiconductor devices. Furthermore, when the amount of charge on the object of observation is small or when the object is negatively charged, the voltage contrast method does not have the sensitivity to detect the difference in the signal intensity y of the emitted electrons, so the emitted electrons EB2 are detected via the energy filter 24.

[0035] In the first embodiment, when an electron beam EB1 is irradiated onto an object to be observed, the transient response waveform of the emitted electrons EB2 emitted from the object to be observed is used to extract the feature values ​​of the transient response waveform, and the object to be observed is discriminated as a normal part or a defective part from the feature values.

[0036] In step S1, first, a sample 9 including a plurality of specific locations is prepared. The plurality of specific locations are, for example, a plurality of plugs 33, 34 (see FIG. 2) formed in the sample 9.

[0037] 6 shows a GUI screen displayed on the display device 18. The user can set the irradiation conditions of the electron beam EB1 and the laser beam LB on the GUI screen using the input device 17. When the user presses the "execute recipe" button, the following steps S2 to S8 are executed.

[0038] In step S2, the irradiation position of the electron beam EB1 is moved so that one specific location becomes the inspection area. The electron beam EB1 is scanned to find the location of the specific location to be inspected. However, during this process, the specific location may be irradiated with the electron beam EB1, which may cause the specific location to become charged. If the specific location is charged, accurate inspection may not be possible depending on the structure of the inspection target. Therefore, before or simultaneously with point irradiation of the specific location with the electron beam EB1, a laser beam LB having a wavelength appropriate for the structure of the inspection target is irradiated onto the specific location. This removes charge from the specific location.

[0039] In step S3, the electron beam EB1 is point-irradiated onto one specific location. Point irradiation is an irradiation method in which the electron beam EB1 (pulse electron beam) is continuously irradiated for a fixed time period without scanning the observation target. Point irradiation also includes scanning the electron beam EB1 on the observation target, for example, within the plug surface, in order to prevent irradiation marks caused by the irradiation of the electron beam EB1.

[0040] In step S4, the transient response waveform of one specific location is measured. First, the emitted electrons EB2 emitted from the specific location by point irradiation are detected by the detector 11. At this time, depending on the charged state of the sample 9, the detection sensitivity by the voltage contrast method may be low and it may be impossible to acquire the transient response waveform. In this case, the emitted electrons EB2 are detected via the energy filter 24 shown in FIG. 1. The detected emitted electrons EB2 are output to the processor 10 as a detection signal. Next, the processor 10 measures the transient response waveform of the emitted electrons EB2 based on the detection signal. FIGS. 3 and 4 show the measured transient response waveform of the emitted electrons EB2.

[0041] In step S5, one or more feature quantities are extracted from the transient response waveform shown in Figures 3 and 4. Specifically, as shown in Figures 3 and 4, a steady-state value C is extracted from the end of the transient response waveform. In Figure 3, the amplitude A is calculated from the difference between the steady-state value C and the maximum value of the transient response waveform. In Figure 4, the amplitude A is calculated from the difference between the steady-state value C and the minimum value of the transient response waveform. The time constant B is the time required for the transient response to converge. Furthermore, if there are multiple gradients in the transient response, it is also possible to calculate multiple time constants B with different values.

[0042] Another method for extracting one or more feature quantities from a transient response waveform is to perform regression analysis on the transient response waveform in Fig. 7 using Equation 1. Equation 1 calculates the amplitude A, time constant B, and steady-state value C as one or more feature quantities, where y is signal strength and t is time.

[0043] Furthermore, as a method for extracting one or more feature quantities from the transient response waveform, the transient response waveform in Fig. 8 is subjected to regression analysis using Equation 2. Equation 2 calculates the amplitude A, time constant B, and steady-state value C as one or more feature quantities, where y is signal intensity and t is time.

[0044] In step S6, it is determined whether inspection has been completed for multiple specific locations within the sample 9. If there are specific locations that require inspection, the irradiation position of the electron beam EB1 is moved to the next specific location. In other words, steps S2 to S5 are performed for multiple specific locations until inspection of all specific locations is completed.

[0045] In step S7, statistical analysis is performed on one or more feature quantities (amplitude A, time constant B, steady-state value C) extracted for the multiple specific locations. For example, the multiple amplitudes A, multiple time constants B, and multiple steady-state values ​​C are plotted as histograms, and the frequencies of the amplitudes A, time constants B, and steady-state values ​​C are analyzed. Also, scatter plots are created showing the relationships between the multiple amplitudes A and the multiple time constants B, the relationships between the multiple time constants B and the multiple steady-state values ​​C, and the relationships between the multiple amplitudes A and the multiple steady-state values ​​C. A wafer map is also created showing the trends of the multiple amplitudes A, multiple time constants B, and multiple steady-state values ​​C within the wafer surface, and the distribution within the wafer surface is analyzed. This provides relationship information for the multiple amplitudes A, multiple time constants B, and multiple steady-state values ​​C.

[0046] As shown in FIG. 9, the user can check a histogram, a scatter plot, and a wafer map as examples of the statistical analysis results on the GUI screen.

[0047] In step S8, multiple specific locations are classified as normal or defective based on the frequency of a specific feature among the one or more feature values. For example, as shown in FIG. 10, the transient response waveforms of plugs 33 and 34 in FIG. 2 have different waveform shapes due to the voltage contrast method. By calculating one or more feature values ​​from the transient response of this waveform shape, it is possible to classify the locations as normal or defective. Here, the frequencies of the amplitude A, the time constant B, and the steady-state value C are checked, and the one that is easiest to distinguish can be used.

[0048] Furthermore, instead of the frequency of a specific feature, multiple specific locations can be determined as normal or defective based on relationship information of multiple feature values. Here, the relationship between multiple amplitudes A and multiple time constants B, the relationship between multiple time constants B and multiple steady-state values ​​C, and the relationship between multiple amplitudes A and multiple steady-state values ​​C can be checked and the one that is easiest to determine can be used.

[0049] Furthermore, one or more boundary values ​​between normal and defective portions are determined based on the frequency of a specific feature or on the relationship information between a plurality of feature amounts.

[0050] After determining one or more boundary values, when inspecting a sample 9 having a similar structure, if one or more feature quantities are extracted from the transient response waveform, there is no need to again obtain the frequency of a specific feature quantity and information on the relationship between multiple feature quantities. In other words, when one or more boundary values ​​have been determined, steps S2, S3, S4, and S5 are performed to extract one or more feature quantities from the transient response waveform, and then the one or more feature quantities are compared with one or more boundary values ​​to determine whether multiple specific locations are normal or defective. This allows inspection processes to be omitted, thereby shortening the inspection time.

[0051] As described above, according to the first embodiment, even if it is not possible to identify normal portions in advance, it is possible to determine whether a specific portion is normal or defective. Furthermore, since the above determination can be made without acquiring an observation image, the inspection time can be reduced.

[0052] (Embodiment 2) A charged particle beam device 1 and an inspection method for a sample 9 according to embodiment 2 will be described below with reference to Fig. 11. In the following description, differences from embodiment 1 will be mainly described, and descriptions of points that overlap with embodiment 1 will be omitted.

[0053] In the first embodiment, the electron beam EB1 is a pulsed electron beam, and one specific location is irradiated with the electron beam EB1 once. In the second embodiment, as shown in Fig. 11, one specific location is irradiated with the electron beam EB1 (pulsed electron beam) multiple times.

[0054] Furthermore, the interruption time of the second and subsequent point irradiations of the electron beam EB1 is longer than the immediately preceding interruption time. For example, the interruption time t2 is longer than the interruption time t1. In other words, the interruption time from the (n+1)th point irradiation of the electron beam EB1 to the (n+2)th point irradiation of the electron beam EB1 is longer than the interruption time from the nth point irradiation of the electron beam EB1 to the (n+1)th point irradiation of the electron beam EB1. Here, n is a natural number greater than or equal to 2.

[0055] The irradiation of the electron beam EB1 to a specific location can be considered as charging the specific location, and the cut-off time can be considered as the charge relaxation in which the charge is released from the specific location. The transient response waveform reflects the material characteristics of the specific location and the characteristics (electrical characteristics) of the circuit in which the specific location is involved. Since the discharge only involves pure electrical characteristics, excluding material characteristics allows for more sensitive measurements.

[0056] By gradually lengthening the interruption time from the second time onwards, the maximum value of the transient response waveform from the third time onwards increases. From the respective maximum values ​​of the multiple transient response waveforms measured by point irradiation with the electron beam EB1 from the second time onwards, the amplitude A, the time constant B and the steady-state value C are extracted or the maximum values ​​are subjected to regression analysis using the above-mentioned Equation 2, thereby calculating the amplitude A, the time constant B and the steady-state value C as one or more feature quantities. y is the signal intensity, and t is the time.

[0057] In order to calculate the maximum value of each of the plurality of transient response waveforms, for example, the amplitude A extracted by regression analysis using Equation 1 can also be used.

[0058] As described above, according to the second embodiment, one or more feature amounts can be extracted with higher accuracy than in the first embodiment.

[0059] 12 to 14, a charged particle beam device 1 and a sample inspection method according to a third embodiment will be described below. Note that the following description will mainly focus on differences from the first embodiment, and descriptions of points that overlap with the first embodiment will be omitted.

[0060] 12 , in the third embodiment, one or more feature quantities are extracted from the transient response waveform in step S5, and then step S9 is performed. In step S9, one or more electrical characteristics corresponding to the one or more feature quantities are estimated by referring to electrical characteristic information, such as a charge model of the sample 9, that has been input in advance. Thereafter, in step S8, multiple specific locations are classified as normal locations or defective locations based on the frequency of a specific electrical characteristic among the one or more electrical characteristics or on the relationship information between multiple electrical characteristics.

[0061] In defect inspection of semiconductor devices using the charged particle beam device 1, not only is the location of the defect identified by the difference in brightness, but the defect can also be classified or evaluated based on electrical properties such as electrical resistance or capacitance, leading to highly accurate process management.

[0062] For example, if defects can be classified based on electrical resistance and capacitance, it is possible to distinguish between short defects, which have a lower electrical resistance than normal semiconductor devices, and depletion capacitance defects, which have a higher capacitance than normal semiconductor devices. In order to properly evaluate electrical defects, it is desirable to analyze electrical parameters such as the electrical resistance and capacitance of normal and defective parts.

[0063] When the electron beam EB1 positively charges the sample and the charge accumulation reaches a steady state, the signal intensity y of the emitted electrons EB2 is inversely proportional to the electrical resistance R and follows Ohm's law as shown in the following equation 3.

[0064] Here, Vs is the surface potential when the charge is saturated, which is the charged state obtained by irradiating the electron beam EB1 for a long time. Since the surface potential Vs when the charge is saturated is almost constant, when the electrical resistance R is small, the signal intensity y becomes high, and when the electrical resistance R is large, the signal intensity y becomes dark.

[0065] On the other hand, when the charging is in a transient state, if the electron beam EB1 is irradiated onto, for example, an RC parallel circuit of electrical resistance R and capacitance C, the signal intensity y of the emitted electrons EB2 is proportional to the capacitance C and is expressed by Equation 4.

[0066] Here, Q is the charge absorbed by the sample 9 by the electron beam EB1, t is the irradiation time, and R is the electrical resistance. Therefore, a transient response waveform is formed by the time constant of the electrical resistance R and the electrostatic capacitance C.

[0067] 13 is a schematic diagram showing an electrical characteristic estimation system 19 included in the processor 10. The electrical characteristic estimation system 19 can estimate one or more electrical characteristics corresponding to one or more feature quantities. As shown in FIG. 13 , the electrical characteristic estimation system 19 includes an input unit 20, a learning unit 21, a learning model storage unit 22, and an electrical characteristic estimation unit 23.

[0068] The input unit 20 receives information on the irradiation conditions of the electron beam EB1 and information on one or more feature quantities from the database 16, and receives electrical characteristic information acquired in advance by the user from the input device 17. The learning unit 21 receives this information as training data.

[0069] The electrical characteristic information is information obtained in advance by the user, such as actual measurements using simulation or an EB tester, values ​​obtained from images obtained by cross-sectional processing and observation of an actual device, or values ​​according to the type of defect based on the operator's experience, etc. The learning unit 21 accepts at least one of these pieces of information as a label.

[0070] FIG. 14 shows data on transient response waveforms for different diameters of the plug 33 in FIG. 2. Diameter B is larger than diameter A, and diameter C is larger than diameter B. As shown in FIG. 14, the transient response waveform differs depending on the diameter. Changing the diameter means that the capacitance of the plug changes. If the capacitance differs, the charging time and the transient response waveform also differ.

[0071] For example, data such as that shown in FIG. 14 is output to the input unit 20 as electrical characteristic information, and this electrical characteristic information is used as training data. It should be noted that not only capacitance but also electrical resistance can be used as training data. Furthermore, comparison information between electrical resistance and capacitance (e.g., information on whether the resistance is greater than or smaller than a reference value) may also be included in the training data. In this way, a data set of electrical characteristic information and irradiation conditions of the electron beam EB1 is trained in advance by the learning unit 21.

[0072] The learning unit 21 is, for example, a neural network, a regressor, or a Bayesian classifier. The learning unit 21 performs machine learning using the received training data. The learning model storage unit 22 stores the learning model constructed by the learning unit 21. The learning model constructed by the learning unit 21 is transmitted to the electrical characteristic estimation unit 23 and used for electrical characteristic estimation. Note that such estimation processing may be performed using an AI accelerator.

[0073] The electrical characteristic estimation unit 23 estimates one or more electrical characteristics corresponding to one or more feature quantities based on the learning model. The estimation results are stored in the database 16. Thereafter, the inspection method for the sample 9 according to the third embodiment is performed by replacing the feature quantities used in the first embodiment with the electrical characteristics. The estimation results can also be displayed on the display device 18. That is, a histogram, a scatter diagram, and a wafer map in which the feature quantities are replaced with the electrical characteristics can be displayed on the GUI screen of FIG. 9.

[0074] Furthermore, the technique described in the third embodiment may be applied in combination with the techniques described in the first and second embodiments. That is, the inspection method in which the feature quantities are replaced with electrical characteristics can also be applied to the first and second embodiments.

[0075] 15 to 18, an application example of the third embodiment will be described below. In this application example, the technical concept of the third embodiment is used to determine whether a semiconductor device to be inspected is a normal semiconductor device or a semiconductor device containing a defect. Here, a MOSFET (Metal Oxide Semiconductor Field Effect Transistor) is used as an example of the semiconductor device.

[0076] 15 is a cross-sectional view showing a MOSFET included in sample 9. The MOSFET has a well region 40, a gate insulating film 41, a gate electrode 42, a sidewall spacer 43, a diffusion region 44, and a diffusion region 45. One of the diffusion region 44 and the diffusion region 45 constitutes the source region of the MOSFET, and the other constitutes the drain region of the MOSFET.

[0077] 15 , a well region 40 is formed in a semiconductor substrate. A gate insulating film 41 is formed on the well region 40. A gate electrode 42 is formed on the gate insulating film 41. Sidewall spacers 43 are formed on both side surfaces of the gate electrode 42. A diffusion region 44 is formed in the well region 40 on one side surface of the gate electrode 42. A diffusion region 45 is formed in the well region 40 on the other side surface of the gate electrode 42. A portion of the well region 40 that is located below the gate electrode 42 and between the diffusion regions 44 and 45 constitutes a channel region of the MOSFET.

[0078] In the case of an n-type MOSFET, the well region 40 exhibits p-type conductivity, and the gate electrode 42, diffusion region 44, and diffusion region 45 exhibit n-type conductivity. In the case of a p-type MOSFET, the well region 40 exhibits n-type conductivity, and the gate electrode 42, diffusion region 44, and diffusion region 45 exhibit p-type conductivity.

[0079] An interlayer insulating film 46 is formed on the semiconductor substrate so as to cover the MOSFET. A plurality of holes are formed in the interlayer insulating film 46. Plugs 47, 48, and 49 are formed by filling the holes with a conductive film mainly made of, for example, a tungsten film.

[0080] The sample 9 includes a plurality of MOSFETs and a plurality of plugs 47, 48, and 49. Essentially, the plurality of plugs 47 should be electrically connected to a plurality of gate electrodes 42, the plurality of plugs 48 should be electrically connected to a plurality of diffusion regions 44, and the plurality of plugs 49 should be electrically connected to a plurality of diffusion regions 45.

[0081] If the MOSFET is normal, a channel region is formed by applying a potential to the gate electrode 42. In this state, a current flows from the drain region to the source region (ON state) by generating a potential difference between the diffusion region 44 and the diffusion region 45. If no potential is applied to the gate electrode 42 and no channel region is formed, no current flows (OFF state).

[0082] FIG. 16 is a cross-sectional view illustrating the types of defects that may occur in the MOSFET of FIG.

[0083] In defect pattern 1, plug 47 does not reach gate electrode 42 (open defect). Therefore, no channel region is formed even if electron beam EB1 is irradiated onto plug 47. In this state, even if electron beam EB1 is irradiated onto plug 48 or plug 49, charges are accumulated in diffusion region 44 or diffusion region 45, and no current flows.

[0084] In defect pattern 2, a defect exists in gate insulating film 41, causing a leak path from gate electrode 42 to well region 40 (gate leak defect). Therefore, even if electron beam EB1 is irradiated onto plug 47, no channel region is formed. In this state, even if electron beam EB1 is irradiated onto plug 48 or plug 49, charges are accumulated in diffusion region 44 or diffusion region 45, and no current flows.

[0085] In defect pattern 3, a leak path (junction leak defect) occurs from diffusion region 44 or diffusion region 45 to well region 40. When electron beam EB1 is irradiated onto plug 47, a channel region is formed. However, even if electron beam EB1 is irradiated onto plug 48 or plug 49 in this state, charges leak into well region 40, and no current flows.

[0086] 17 is a flowchart showing a method for inspecting a sample 9 in an application example. In the application example, a plurality of MOSFETs are set as a plurality of specific locations. As shown in FIG. 17 , after estimating one or more electrical characteristics of a plug 47 (gate electrode 42), one or more electrical characteristics of a plug 48 or a plug 49 (diffusion region 44 or diffusion region 45) are estimated.

[0087] First, steps S3, S4, S5 and S9 are performed on the plug 47 (gate electrode 42).

[0088] The plug 47 is irradiated with an electron beam EB1, a transient response waveform is measured, and one or more feature quantities are extracted from the transient response waveform. The one or more extracted feature quantities have a correlation with at least the capacitance of the gate electrode 42. Therefore, one or more electrical characteristics are estimated from the one or more feature quantities. The one or more electrical characteristics can be used to determine whether the structure near the gate electrode 42 is normal or defective.

[0089] Next, steps S3, S4, S5 and S9 are performed on the plug 48 or the plug 49 (diffusion region 44 or diffusion region 45).

[0090] Irradiating the electron beam EB1 onto the plug 47 would normally form a channel region under the gate electrode 42, allowing the MOSFET to operate normally. It is necessary to inspect whether or not there are defects other than in the structure near the gate electrode 42. To this end, the electron beam EB1 is irradiated onto the plug 48 or 49, the transient response waveform is measured, one or more feature quantities are extracted from the transient response waveform, and one or more electrical characteristics are estimated from the one or more feature quantities.

[0091] For a plurality of MOSFETs, the plug 47 is irradiated with the electron beam EB1, and the plug 48 or the plug 49 is irradiated with the electron beam EB1.

[0092] 9 are created for plug 47, and the frequency of a specific electrical characteristic or relationship information for multiple electrical characteristics is obtained. Also, a histogram, scatter plot, and wafer map such as those in FIG. 9 are created for plug 48 or plug 49, and the frequency of a specific electrical characteristic or relationship information for multiple electrical characteristics is obtained.

[0093] It is also possible to create a scatter diagram combining multiple electrical characteristics related to the plug 47 and multiple electrical characteristics related to the plug 48 or plug 49, and to obtain relationship information combining these. In addition, the relationship between the charge Q absorbed in the sample 9 when the plug 47 is irradiated with the electron beam EB1 and the electrical resistance R estimated when the plug 48 or plug 49 is irradiated with the electron beam EB1 can be shown as in Figure 18. In this way, by charging the gate electrode 42 in stages, it is possible to show the change in the electrical resistance R of the channel region.

[0094] In this way, it is possible to distinguish between normal MOSFETs and defective MOSFETs based on the frequency of a particular electrical characteristic or on the relationship information between a plurality of electrical characteristics.

[0095] The present invention has been specifically described above based on the above embodiment, but the present invention is not limited to the above embodiment and can be modified in various ways without departing from the spirit of the present invention.

[0096] REFERENCE SIGNS LIST 1 Charged particle beam device 2 Electron gun 3 Condenser lens 4 Pulse electron generator 5 Aperture 6 Deflector 7 Objective lens 8 Stage 9 Sample 10 Processor 11 Detector 12 Detection signal analysis unit 13 Light source 14 Light path interrupter 15 Light path 16 Database 17 Input device 18 Display device 19 Electrical characteristic estimation system 20 Input unit 21 Learning unit 22 Learning model storage unit 23 Electrical characteristic estimation unit 24 Energy filter 30 Semiconductor substrate 31 Insulating film 32 Interlayer insulating film 33, 34 Plug 40 Well region 41 Gate insulating film 42 Gate electrode 43 Sidewall spacer 44, 45 Diffusion region (source region, drain region) 46 Interlayer insulating film 47, 48, 49 Plug (conductive film) EB1 Electron beam EB2 Emitted electrons LB laser beam

Claims

1. A charged particle beam device comprising: a charged particle gun capable of irradiating a sample with a charged particle beam; and a detector that detects emitted electrons when the sample is irradiated with the charged particle beam; the device irradiates the charged particle beam at a plurality of specific locations included in the sample, detects the emitted electrons emitted by the point irradiation, measures a transient response waveform of the emitted electrons, extracts one or more feature amounts from the transient response waveform, and classifies the plurality of specific locations as normal or defective based on the frequency of a first feature amount among the one or more feature amounts, or relationship information between a plurality of the feature amounts.

2. A charged particle beam device according to claim 1, wherein one or more boundary values ​​between the normal portion and the defective portion are determined based on the frequency of the first feature or on relationship information between the plurality of feature amounts.

3. A charged particle beam device according to claim 2, wherein, when the one or more boundary values ​​have been determined, the one or more feature values ​​are extracted from the transient response waveform, and then the one or more feature values ​​are compared with the one or more boundary values, thereby discriminating the plurality of specific locations as normal locations or defective locations.

4. A charged particle beam device according to claim 1, wherein an amplitude A, a time constant B, and a steady-state value C are extracted as the one or more feature quantities of the transient response waveform.

5. The charged particle beam device according to claim 1, wherein the one or more feature quantities are calculated as amplitude A, time constant B, and steady-state value C by performing regression analysis on the transient response waveform using formula 1 or formula 2. (y is signal intensity, and t is time.) 6. A charged particle beam device according to claim 1, further comprising a pulsed electron generator for controlling the charged particle beam as a pulsed charged particle beam, and for point-irradiating the pulsed charged particle beam onto one of the specific locations.

7. A charged particle beam device according to claim 1, further comprising a pulsed electron generator for controlling the charged particle beam as a pulsed charged particle beam, wherein the pulsed charged particle beam is irradiated multiple times onto one specific location, and the interruption time from the n+1th point irradiation of the pulsed charged particle beam to the n+2th point irradiation of the pulsed charged particle beam is longer than the interruption time from the nth point irradiation of the pulsed charged particle beam to the n+1th point irradiation of the pulsed charged particle beam (n is a natural number of 2 or more).

8. A charged particle beam device according to claim 7, wherein amplitude A, time constant B and steady-state value C are extracted as the one or more feature quantities from the maximum values ​​of each of the plurality of transient response waveforms measured by point irradiation of the pulsed charged particle beam from the second time onwards, or the maximum values ​​are subjected to regression analysis using Equation 2, thereby calculating amplitude A, time constant B and steady-state value C as the one or more feature quantities. (y is signal intensity, and t is time.) 9. A charged particle beam device according to claim 1, further comprising an energy filter for energy filtering the emitted electrons, wherein the emitted electrons are detected via the energy filter.

10. A charged particle beam device according to claim 1, further comprising a light source capable of irradiating a laser beam, wherein the laser beam is irradiated onto one of the specific locations before or simultaneously with point irradiation of the charged particle beam onto one of the specific locations.

11. A sample inspection method comprising: (a) preparing a sample including a plurality of specific locations; (b) point-irradiating one of the specific locations with a charged particle beam; (c) detecting emitted electrons emitted by the point irradiation; (d) measuring a transient response waveform of the emitted electrons; (e) extracting one or more feature amounts from the transient response waveform; and (f) performing steps (b), (c), (d), and (e) on the plurality of specific locations, and discriminating the plurality of specific locations as normal or defective based on the frequency of a first feature amount among the one or more feature amounts or relationship information between a plurality of the feature amounts.

12. A method for inspecting a sample according to claim 11, wherein one or more boundary values ​​between the normal portion and the defective portion are determined based on the frequency of the first feature or on relationship information between the plurality of feature amounts.

13. A method for inspecting a sample according to claim 12, further comprising: (g) when the one or more boundary values ​​have been determined, performing steps (b), (c), (d) and (e) on the plurality of specified locations, and comparing the one or more feature amounts with the one or more boundary values ​​to determine whether the plurality of specified locations are normal or defective.

14. A charged particle beam device according to claim 11, wherein an amplitude A, a time constant B, and a steady-state value C are extracted as the one or more feature quantities of the transient response waveform.

15. A sample inspection method according to claim 11, wherein the one or more feature quantities are calculated as amplitude A, time constant B, and steady-state value C by performing regression analysis on the transient response waveform using Equation 1 or Equation 2. (y is signal intensity, and t is time.) 16. A method for inspecting a sample according to claim 11, wherein the charged particle beam is a pulsed charged particle beam, and in step (b), the pulsed charged particle beam is irradiated at a point on the one specific location.

17. A method for inspecting a sample according to claim 11, wherein the charged particle beam is a pulsed charged particle beam, and in step (b), the one specific location is irradiated with the pulsed charged particle beam a plurality of times, and the interruption time from the n+1th point irradiation of the pulsed charged particle beam to the n+2th point irradiation of the pulsed charged particle beam is longer than the interruption time from the nth point irradiation of the pulsed charged particle beam to the n+1th point irradiation of the pulsed charged particle beam (n is a natural number equal to or greater than 2).

18. A method for inspecting a sample according to claim 17, further comprising the step of: (i) extracting amplitude A, time constant B, and steady-state value C as the one or more feature quantities from the maximum values ​​of each of the multiple transient response waveforms measured by point irradiation with the pulsed charged particle beam from the second time onwards, or calculating amplitude A, time constant B, and steady-state value C as the one or more feature quantities by performing regression analysis of the maximum values ​​using Equation 2. (y is signal intensity, and t is time.) 19. A method for inspecting a specimen according to claim 11, wherein the emitted electrons are detected via an energy filter.

20. A method for inspecting a sample according to claim 11, further comprising: (j) irradiating said one specific location with a laser beam before point-irradiating said one specific location with the charged particle beam, or simultaneously with point-irradiating said one specific location with the charged particle beam.

21. A charged particle beam device comprising: a charged particle gun capable of irradiating a sample with a charged particle beam; and a detector that detects emitted electrons when the sample is irradiated with the charged particle beam, wherein the device irradiates each of a plurality of specific locations included in the sample with the charged particle beam at a point, detects the emitted electrons emitted by the point irradiation, measures a transient response waveform of the emitted electrons, extracts one or more feature amounts from the transient response waveform, and estimates one or more electrical characteristics corresponding to the one or more feature amounts by referring to electrical characteristic information, and discriminates the plurality of specific locations as normal or defective based on the frequency of a first electrical characteristic among the one or more electrical characteristics or relationship information between a plurality of the electrical characteristics.

22. A charged particle beam device as defined in claim 21, wherein the sample includes a plurality of MOSFETs, each having a gate insulating film, a gate electrode, a first diffusion region, and a second diffusion region; a plurality of first conductive films to be electrically connected to the plurality of gate electrodes; a plurality of second conductive films to be electrically connected to the plurality of first diffusion regions; and a plurality of third conductive films to be electrically connected to the plurality of second diffusion regions; the plurality of MOSFETs are defined as the plurality of specific locations; one or more electrical characteristics of the first conductive film are estimated, and then one or more electrical characteristics of the second conductive film or the third conductive film are estimated; and the plurality of MOSFETs are distinguished as normal MOSFETs or MOSFETs including defects based on the frequency of the first electrical characteristic or relationship information of the plurality of electrical characteristics.

23. A sample inspection method comprising: (a) preparing a sample including a plurality of specific locations; (b) point-irradiating one of the specific locations with a charged particle beam; (c) detecting emitted electrons emitted by the point irradiation; (d) measuring a transient response waveform of the emitted electrons; (e) extracting one or more feature quantities from the transient response waveform; (f) estimating one or more electrical characteristics corresponding to the one or more feature quantities by referring to electrical characteristic information; and (g) performing steps (b), (c), (d), (e), and (f) on the plurality of specific locations, and discriminating the plurality of specific locations as normal or defective based on the frequency of a first electrical characteristic among the one or more electrical characteristics or relationship information between the plurality of electrical characteristics.

24. A method for inspecting a sample as defined in claim 23, wherein the sample includes: a plurality of MOSFETs, each having a gate insulating film, a gate electrode, a first diffusion region, and a second diffusion region; a plurality of first conductive films to be electrically connected to the plurality of gate electrodes; a plurality of second conductive films to be electrically connected to the plurality of first diffusion regions; and a plurality of third conductive films to be electrically connected to the plurality of second diffusion regions; the method defines the plurality of MOSFETs as the plurality of specific locations; estimates one or more electrical characteristics of the first conductive film, and then estimates one or more electrical characteristics of the second conductive film or the third conductive film; and distinguishes the plurality of MOSFETs as normal MOSFETs or MOSFETs containing defects based on the frequency of the first electrical characteristic or relationship information between the plurality of electrical characteristics.

Images