JP2008151797A - Length measuring method by scanning electron microscope - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a length measuring method by a scanning electron microscope, capable of preventing a length measuring error from being generated by the constitution of a sample, a kind of film or the like. <P>SOLUTION: In this length measuring method by the scanning electron microscope, the scanning electron microscope is auto-focused with a measuring point, using an output from a sensor for detecting a distance from an objective lens to the sample, and a change of a sample signal emitted from the sample is monitored while changing an excitation current of the objective lens, to detect the excitation current ΔIobj of the objective lens corresponding to a focus shift. The excitation current ΔIobj of the objective lens is converted into an acceleration voltage ΔV of an electron beam getting incident upon the sample, and the acceleration voltage of the electron beam getting incident upon the sample is changed by the acceleration voltage ΔV. The length is measured thereafter in the measuring point. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a length measuring method using a scanning electron microscope, and more particularly to a length measuring method using a scanning electron microscope suitable for length measurement and inspection of a semiconductor device.

  Recent scanning electron microscopes (SEMs), especially length measurement SEMs, have measured values that are true values, and the measured values have variations and offsets depending on the sample configuration, film type, etc. There is no need for it. Conventionally, as a dimension calibrating means of a length measurement SEM, a pitch pattern is created using a material (for example, Al, W, etc.) with less charge on Si, and a length measurement value is adjusted to the pitch pattern. The length measurement calibration value of the length measurement SEM was obtained.

Further, in recent measurement SEMs, the acceleration voltage of incident electrons on the primary side is increased to improve the resolution of secondary electrons, and a negative voltage is applied to the sample side to decelerate the acceleration voltage of incident electrons. An electric field method (hereinafter referred to as a retarding method) is employed (see, for example, Patent Document 1). Specifically, in the retarding method, the acceleration voltage of electrons emitted from the electron gun portion is increased by + Vr, and a voltage of −Vr is applied to the sample, compared with the conventional method (sample-side voltage 0). The electrons emitted from the gun unit with an acceleration voltage of V + Vr are decelerated by the voltage of −Vr applied to the sample side after passing through the objective lens, and finally enter the sample with the acceleration voltage V. Has been. The characteristics of the retarding method are that the incident electrons on the primary side pass under the objective lens with the acceleration voltage V + Vr, so that the chromatic aberration of the lens can be reduced, and the secondary electrons generated from the sample are detected by the electric field of −Vr. The secondary electron collection efficiency is improved because the electron beam is accelerated in the direction of the vessel.
JP-A-10-294074

  However, in the scanning electron microscope that employs the retarding method as described above, the acceleration voltage of the electron beam incident on the sample (hereinafter referred to as the “Landing Voltage”) varies depending on the voltage applied to the sample side. When the sample to be performed is covered with an insulating film, or when the insulating film is attached to the surface, the landing voltage changes and the magnification (Field of View) shifts. The dimension measurement of the length measurement SEM is performed under the condition that the Landing Voltage is constant and the magnification is constant. Therefore, when one of the values changes, there is a problem that the value of the measured dimension changes. That is, there is a problem that a dimensional error does not occur due to a change in the Landing Voltage, and the measured dimension is not a true value.

  This problem also occurred in the conventional length measuring SEM that does not employ the retarding method. That is, the landing voltage is changed due to the change in the surface potential of the sample, resulting in a dimensional error and a problem that the measured dimension is not a true value.

Table 1, the conventional length measuring SEM, Si wafer Si pattern formed, on the B-PSG film having a film thickness of 400nm pattern, on the SiO ₂ film with a thickness of 10nm patterns on SiO ₂ film having a thickness of 100nm pattern 3 shows the results of measuring the pitch in the pattern on the SiN film, the pattern on the HTO + SiN film (35 nm), the pattern on the SiON film with a film thickness of 25 nm, and the pattern on the real device A and the real device B. The measured pitch has a design value of 350 nm, 400 nm, 500 nm, 600 nm, 700 nm, and 800 nm. The unit of numerical values in the table is nm.

  FIG. 18 shows the pitch design value (horizontal axis) and SEM length measurement value (vertical) obtained from the data of the Si pattern, the B-PSG film pattern and the actual device A pattern among the length measurement data shown in Table 1. (Axis) is illustrated. Although the pitch measurement value of the Si pattern substantially matches the design value, the pitch measurement value of the pattern on the B-PSG insulating film and the pitch measurement value of the pattern on the actual device have a large deviation from the design value. I understand that. This is presumably because an error occurred in the measured value because the sample surface was charged up by electron beam irradiation during pattern length measurement on the B-PSG insulating film or the actual device.

  The present invention has been made in view of such problems in length measurement with a conventional scanning electron microscope, and is capable of measuring with a scanning electron microscope that does not cause a length measurement error due to the configuration of the sample or the film type. It aims to provide a method.

  In order to solve the above problems, the length measuring method using the scanning electron microscope of the present invention is a step of autofocusing the scanning electron microscope to the measurement point using the output of the sensor that detects the distance from the objective lens to the sample. Monitoring the change in the sample signal emitted from the sample while changing the excitation current of the objective lens to detect the excitation current ΔIobj of the objective lens corresponding to the focus shift, and the excitation current ΔIobj of the objective lens to the sample. The method includes the steps of converting the acceleration voltage ΔV of the incident electron beam into ΔV, changing the acceleration voltage of the electron beam incident on the sample by ΔV, and measuring the measurement point.

  The autofocus may be performed by controlling the excitation current of the objective lens, or may be performed by mechanically moving a sample stage or the like on which the sample is placed. According to this method, the change in the electron beam acceleration voltage caused by the charge-up of the sample surface can be canceled and the acceleration voltage of the electron beam incident on the sample can be set as a set value. The error due to the charge-up is not included.

  The length measuring method using the scanning electron microscope of the present invention also includes the step of autofocusing the scanning electron microscope to the measurement point using the output of the sensor that detects the distance from the objective lens to the sample, and the voltage applied to the sample. Monitoring the change in the sample signal emitted from the sample while changing the sample, detecting the sample applied voltage ΔVb corresponding to the focus shift, changing the electron beam acceleration voltage incident on the sample by ΔVb, and measurement points And a step of measuring the length.

  The autofocus may be performed by controlling the excitation current of the objective lens, or may be performed by mechanically moving a sample stage or the like on which the sample is placed. According to this method, similarly to the above method, the change in the electron beam acceleration voltage caused by the charge-up of the sample surface can be canceled out, and the acceleration voltage of the electron beam incident on the sample can be set to the set value. The long value does not include an error due to charge-up of the sample surface.

  The length measuring method using the scanning electron microscope of the present invention also includes a step of obtaining a correspondence relationship between the acceleration voltage of the electron beam incident on the sample and the magnification correction coefficient in advance, and electrons incident on the length measuring point on the sample. A step of obtaining an actual acceleration voltage of the beam, a step of obtaining a magnification correction coefficient corresponding to the actual acceleration voltage of the electron beam incident on the length measurement point with reference to the correspondence relationship obtained in advance, and a measurement point And a step of obtaining a true dimension by multiplying the length measurement value by the magnification correction coefficient.

  In this method, the step of obtaining the actual acceleration voltage of the electron beam incident on the measurement point on the sample may be realized in any way, for example, the acceleration of the electron beam set in the scanning electron microscope A step of obtaining an objective lens correction current ΔIobj that eliminates a focus shift under an objective lens excitation condition corresponding to the voltage, and a step of converting the objective lens correction current ΔIobj into an acceleration voltage ΔV of an electron beam incident on the sample. be able to. Alternatively, it may be realized by obtaining the sample application voltage ΔVb that eliminates the focus shift under the objective lens excitation condition corresponding to the acceleration voltage of the electron beam set in the scanning electron microscope.

  According to the length measurement method of the present invention, in the length measurement using the scanning electron microscope, it becomes possible to obtain a true dimension value by compensating for the length measurement error caused by the potential change of the sample surface. In a general length measurement SEM that does not use the retarding method, the length measurement error due to charge-up of the sample surface due to incident electron irradiation is sufficiently corrected, and the true value is obtained regardless of the sample configuration and type. Can be obtained. That is, when the mayor is charged up, the surface potential of the sample changes, and accompanying this, the magnification deviation due to the error in the landing voltage of the incident beam (the occurrence of magnification deviation means that the offset value is included in the measurement value) However, since this problem is solved, it is possible to always measure the true value even if the surface potential of the sample changes due to the electron beam irradiation.

  Further, in the length measuring SEM employing the retarding method, the problem of magnification shift due to the setting error of the Landing Voltage, which has been a problem until now, is solved. In other words, when an insulating film is attached to the back surface of the sample, or when the interlayer structure of the portion irradiated with the electron beam changes, it is possible to solve the problem that a dimensional error may occur depending on the sample to be measured, It is always possible to measure the true value. Furthermore, according to one aspect of the present invention, the acceleration voltage of the electron beam incident on the sample can always be set to the design value regardless of the configuration and type of the sample.

  According to the present invention, it is possible to provide a length measurement method using a scanning electron microscope that does not cause a length measurement error due to the configuration of the sample, the film type, or the like.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the following description, the case where the present invention is mainly applied to pattern measurement on a semiconductor wafer by a length measurement SEM will be described, but the scope of application of the present invention is not limited to that.

  FIG. 1 is a flowchart for explaining an example of a length measuring method of a length measuring SEM according to the present invention. First, a wafer whose length is to be measured with a length measurement SEM is placed in a chamber (S101), and global alignment is performed (S102). Global alignment, for example, detects two known points on the wafer with an optical microscope and performs alignment in the translational direction and rotational direction of the wafer. The alignment of the wafer coordinate system and the stage coordinate system of the length measurement SEM is performed. Done as a purpose. After that, before measuring the first length measurement point, it moves to the correction data acquisition pattern (S103).

  FIG. 2 shows an example of a correction data acquisition pattern. The correction data acquisition pattern is a test pattern formed in an empty area in the chip or in a boundary part (scribe) between the chip and the chip, and this test pattern is a line and space pattern as shown in FIG. A space and line pattern as shown in FIG. 2B, a contact hole pattern as shown in FIG.

  After moving to the correction data acquisition pattern, the pattern is measured to obtain length measurement data (S104). Next, the length measurement data is compared with the design value (true value) (S105). At this time, in general, a pattern exposed and formed with light changes in line width and space width due to proximity effects, light interference, and the like, so it is better to use measurement data obtained by measuring the pitch. However, if there is no such restriction, measurement data of line width or space width may be used. As a result of comparing the measurement data with the design value (true value), if there is no dimensional error, use the magnification correction coefficient obtained during standard wafer measurement (S106) and measure all measurement points (S107). ).

  If a dimension error has occurred as a result of comparing the length measurement data with the design value (true value), a magnification correction coefficient for removing the dimension error component is calculated (S108). For example, it is assumed that the scanning width (Field of View) El of the electron beam is 1500 nm when the magnification Mag of the SEM is 100,000. When measuring a pitch width of 400 nm with a standard wafer, assuming that the output value is 402 nm, 2 nm of the error must be removed. In this case, the correction is performed by changing the El × magnification correction coefficient, but is normally El × 1.0 (magnification correction coefficient Mag (a) = 1.0), and in this case, the magnification correction coefficient is handled as Mag (a) = 0.95. If the same result is not obtained even if the other wafer patterns are measured in this state, the magnification correction coefficient is calculated again from the correction pattern. The magnification correction coefficient Mag (a) is calculated by dividing the design value (true value) by the length measurement value. For example, when the value calculated from the correction pattern of the wafer with SiN film is 410 nm and the design data (true value) is 400 nm, the magnification correction coefficient Mag (a) is 0.9756.

  Thereafter, the true value can be obtained by performing length measurement using the magnification correction coefficient Mag (a) (= 0.9756). Accordingly, first, the first length measurement point is measured (S109), and the true length measurement value is calculated by multiplying the length measurement value by the magnification correction coefficient (S110). Subsequently, other length measurement points are measured, and a true length measurement value is calculated based on the magnification correction coefficient obtained in the same manner (S111). When the length measurement for all length measurement points is completed, the wafer is unloaded (S112), and the process is terminated.

  FIG. 3 is a flowchart for explaining another example of the length measuring method of the length measuring SEM according to the present invention. Here, the magnification correction coefficient is calculated by measuring a pattern having a known dimension formed in the actual device (chip) without using a special correction data acquisition pattern. Therefore, in this case, it is not necessary to previously form a test pattern in the scribe or chip empty area.

  First, a wafer whose length is to be measured by the length measuring SEM is placed in the chamber (S201), and global alignment is performed (S202). After that, before measuring the first length measurement point, it moves to the position of a known line and space pattern or contact hole pattern formed in the actual device (S203), and measures the pitch pattern ( S204). FIG. 4 is a diagram showing an example of a pattern that is formed in an actual device and can be used for calibration of the measurement value of the length measurement SEM. This pattern includes a line and line pattern as shown in FIG. There are a space pattern, a space and line pattern as shown in FIG. 4B, a contact hole pattern as shown in FIG.

  Next, the length measurement data obtained by pitch pattern length measurement is compared with the design value (true value) (S205). Here, in general, a pattern exposed and formed with light changes in line width and space width due to proximity effects, light interference, and the like, so it is better to use measurement data obtained by measuring the pitch. However, if there is no such restriction, measurement data of line width or space width may be used. As a result of comparing the measurement data with the design value (true value), if there is no dimensional error, use the magnification correction coefficient obtained during standard wafer measurement (S206) and measure all measurement points (S207). ).

  If there is a dimension error as a result of comparing the length measurement data with the design value (true value), a magnification correction coefficient Mag (a) for removing the dimension error component is calculated (S208). This magnification correction coefficient Mag (a) is calculated by dividing the design value (true value) by the length measurement value. At subsequent measurement points, if the length measurement is performed using the magnification correction coefficient Mag (a) thus calculated, a true value can be obtained. The first length measurement point is measured (S209), and the length measurement value is multiplied by a magnification correction coefficient Mag (a) to calculate a true length measurement value (S210). Subsequently, other length measurement points are measured, and a true length measurement value is calculated based on the magnification correction coefficient Mag (a) obtained in the same manner (S211). When the length measurement for all length measurement points is completed, the wafer is unloaded (S212), and the process is terminated.

  FIG. 5 is a flowchart for explaining another example of the length measuring method of the length measuring SEM according to the present invention. Here, the line width, the space width, and the pitch are collectively measured at the first length measurement point, and the magnification correction coefficient Mag (a) for length measurement calibration is calculated using the pitch length measurement data. Therefore, it is not necessary to previously form a test pattern in the scribe or chip empty area. However, in this case, since a pitch pattern is required, the effect cannot be obtained if the pattern of the first length measurement point is a single pattern or a pattern having no periodicity. Therefore, it is necessary to select a length measurement point including a periodic pattern as the first length measurement point.

  First, a wafer whose length is to be measured with a length measurement SEM is placed in a chamber (S301), and global alignment is performed (S302). After that, it moves to the actual device pattern of the first length measurement point (S303), and the line width, space width and pitch are collectively measured (S304). Next, the pitch measurement data is compared with the pitch design value (true value) (S305). In general, a pattern exposed and formed with light changes in line width and space width due to proximity effects, light interference, and the like, so measurement data obtained by measuring the pitch is used. If there is no dimensional error in the pitch, the length correction points obtained during standard wafer length measurement are used (S306) to measure all length measurement points (S307).

  If a pitch dimension error occurs as a result of comparing the pitch measurement data with the pitch design value (true value), the magnification correction coefficient Mag (a) is used to remove the dimension error component from the pitch error. The calculated value is fed back to the length measurement values of the line width and space width of the first length measurement point (S308). That is, the true value is obtained by multiplying the length measurement values of the line width and space width of the first length measurement point by the magnification correction coefficient Mag (a) (S309). Thereafter, other length measurement points are measured, and a true length measurement value is calculated based on the magnification correction coefficient Mag (a) obtained in the same manner (S310). When the length measurement for all length measurement points is completed, the wafer is unloaded (S311), and the process is terminated.

  In measuring the length of a length measuring SEM according to the present invention, a magnification correction coefficient may be calculated for each wafer in a lot as schematically shown in FIG. For example, the magnification correction coefficient obtained by the method of FIG. 1, FIG. 3 or FIG. 5 for the wafer 11 is a, the magnification correction coefficient obtained by the method of FIG. 1, where the magnification correction coefficient obtained by the method of FIG. 3 or FIG. 5 is c, the true dimension of the wafer 11 is obtained by multiplying the length measurement value of the length measurement SEM by the magnification correction coefficient a. Similarly, for the wafer 12, the true dimension is obtained by multiplying the length measurement value of the length measurement SEM by the magnification correction coefficient b, and for the wafer 13, the true dimension is obtained by multiplying the length measurement value of the length measurement SEM by the magnification correction coefficient c. Determine the dimensions.

  In the embodiments of the present invention described so far, the degree of charge and the influence of the bias voltage change depending on the insulating film and the process that has passed for each process. It is obtained for each wafer or a single wafer), and the true dimension is obtained by correcting the length measurement value. However, even with a single wafer, the bias voltage application state may change between the wafer center and the wafer periphery. Therefore, a magnification correction coefficient may be obtained for each chip of each sample, and if the length is measured in the same chip, the same magnification correction coefficient may be used to obtain the true value of the length measurement value. That is, instead of using one magnification correction coefficient for one wafer, magnification correction coefficients a to e are calculated for each chip in the wafer 10 as schematically shown in FIG. The true dimension may be obtained by multiplying the length measurement value of the long SEM by the magnification correction coefficients a to e of the chip.

  Furthermore, even if the length is measured within the same chip, the degree of charge changes depending on the irradiation area of the incident beam and the underlying structure (film type, film thickness, and combination) of the sample. It may be obtained for each of the measurement points, and the true value of the length measurement value may be obtained. That is, as schematically shown in FIG. 8, a magnification correction coefficient may be calculated in the wafer, for each chip, and for each length measurement point, and the true dimension may be obtained using the magnification correction coefficient.

  The method described so far has been a method for obtaining a true dimension by multiplying a length measurement value measured by a length measurement SEM by a magnification correction coefficient. On the other hand, the method described below with reference to FIGS. 9 to 14 is a method of controlling the incident voltage of the electron beam to the sample so that the Landing Voltage becomes a design value. According to this method, the length measurement value measured by the length measurement SEM becomes a true dimension as it is without correction using a magnification correction coefficient.

  First, the case where there is no Z-axis drive system for the wafer stage 21 will be described with reference to FIGS. 9, 10, and 11. FIG. 9 is a schematic diagram of the vicinity of the objective lens of the length measuring SEM apparatus adopting the retarding method. A retarding voltage −Vr is applied from the power source 22 to the wafer 10 held on the wafer stage 21. The incident electron beam 24 accelerated from the electron gun 23 with the acceleration voltage V 0 is converged on the wafer 10 by the objective lens 25 excited by the exciting coil 26. The incident electron beam 24 is scanned on the wafer 10 by the deflection coil 27, and secondary electrons emitted from the wafer are detected by a secondary electron detector (not shown). The extraction electrode 30 is an electrode for extracting secondary electrons emitted from the wafer upward. A distance Z from the objective lens 25 to the surface of the wafer 10 is detected by a Z sensor including a light source 28a and a photodetector 28b. The Z sensor provides the focal length of the objective lens 25 by causing the light beam 29 from the light source 28a to enter the surface of the wafer 10 obliquely and detecting the position where the light beam reflected by the wafer enters the photodetector 28b. The error ΔZ between the measured distance and the distance from the objective lens 25 to the surface of the wafer 10 can be detected.

  In this method, as shown in the flowchart of FIG. 10, the wafer 10 is first loaded on the wafer stage 21 (S401), global alignment (S402) is performed, and then the stage is moved to the first length measurement point (S403). ). FIG. 11A shows the positional relationship between the focal length of the objective lens 25 and the wafer 10 at this time. The focal length of the objective lens is set to the initial value of the apparatus, and the surface of the wafer 10 is not focused. Next, the focus shift ΔZ is detected by the Z sensors 28a and 28b (S404), and the correction current ΔIobj of the objective lens excitation current corresponding to ΔZ is applied to the excitation coil 26 (S405). Thus, as shown in FIG. 11B, the focal length of the objective lens 25 is set to the focus Fz given by the Z sensors 28a and 28b (S406).

  However, it is an ideal case that the incident beam 24 is focused on the wafer surface as shown in FIG. 11B by applying the correction current ΔIobj to the exciting coil 26. In practice, the surface of the wafer 10 is charged or Due to the setting error of the retarding voltage applied to the wafer 10, a focus shift ΔZ ′ as shown in FIG. Next, while observing the secondary electron detection signal of the length measurement SEM, the current value ΔIobj ′ of the exciting coil 26 is changed to obtain the best focus point Fb that compensates for the focus shift ΔZ ′ that could not be detected by the Z sensor. In this way, the current value ΔIobj ′ of the exciting coil 26 corresponding to the focus shift amount ΔZ ′ is obtained (S407). Next, the current value ΔIobj ′ into which ΔZ ′ is added is converted into an acceleration voltage ΔV ′ (S408), and the incident voltage of the electron beam to the sample is controlled (S409). When controlling the incident voltage, it is sufficient that the LandingVoltage is constant. Therefore, ΔV ′ can be controlled on the incident voltage (electron gun) side, on the bias voltage (retarding voltage) 22 side, or both. You may control using a means together.

  In this way, length measurement is performed using the Landing Voltage as a set value (S410). The excitation current of the objective lens 25 at this time is the excitation current set corresponding to the detection value of the Z sensor in step 405, but the focus may be adjusted again using an image signal. After the length measurement, it is determined whether the measurement is completed by measuring the length measurement point (S411). If the measurement is completed, the wafer is unloaded (S412), and the process is terminated. If length measurement points still remain, the stage is moved to the next length measurement point (S403), and the same processing is repeated until the length measurement for all length measurement points is completed.

By the above method, previously shown in Table 1, Si Si pattern formed on the wafer, B-PSG film pattern having a thickness of 400 nm, on the SiO ₂ film with a thickness of 10nm patterns, thickness 100nm of SiO ₂ film The pitch among the pattern, the pattern on the SiN film, the pattern on the HTO + SiN film (35 nm), the pattern on the SiON film having a film thickness of 25 nm, and the pattern on the actual device A and the actual device B was measured. The measured pitch has a design value of 350 nm, 400 nm, 500 nm, 600 nm, 700 nm, and 800 nm. The results are shown in Table 2. The unit of numerical values in the table is nm.

  FIG. 17 shows the pitch design value (horizontal axis) and SEM length measurement value (vertical) obtained from the data of the Si pattern, the B-PSG film pattern and the actual device A pattern among the length measurement data shown in Table 1. (Axis) is illustrated. The three measured values almost overlap, and not only the Si pattern, but also the pitch measured values of the pattern of the B-PSG film and the actual device A are in good agreement with the design values. This is because the influence of the charge-up of the sample surface due to the electron beam irradiation is taken into account in the pattern length measurement on the B-PSG insulating film or the actual device.

  In the conventional technology, if the conditions such as the material, material, film thickness, film type, film combination, etc. of the sample are different between the pitch calibration pattern and the sample to be measured, charge up of incident electrons and insulation resistance of the sample itself Therefore, as shown in Table 1 and FIG. 18, the length measurement value was rarely always a true value (absolute value). However, according to the present invention, a true value can always be obtained as shown in Table 2 and FIG. 17 even if the pitch calibration pattern and the conditions of the sample to be measured are different.

  Next, with reference to FIGS. 9 and 12, a case where there is a Z-axis drive system of the wafer stage will be described. As shown in the flowchart of FIG. 12, the wafer 10 is first loaded on the wafer stage 21 (S501), and global alignment is performed (S502). Thereafter, the stage is moved to the first length measurement point (S503), and the defocus amount ΔZ is detected by the Z sensors 28a and 28b (S504). Next, the Z axis of the wafer stage 21 is driven corresponding to the detected ΔZ (S505), and the focus Fz given by the Z sensors 28a and 28b is set (S506).

  In this case as well, even if the wafer 10 is positioned at the focus position given by the Z sensor, it is ideal that the incident beam 24 is focused on the wafer surface. Due to the setting error of the retarding voltage to be applied to 10, a focus shift ΔZ ′ occurs. Therefore, while observing the secondary electron detection signal of the length measurement SEM, the current value ΔIobj ′ of the exciting coil 26 is changed by ΔZ ′ that cannot be moved by the Z sensor, and the best focus point Fb is obtained. In this way, the current value ΔIobj ′ of the exciting coil 26 corresponding to the focus shift amount ΔZ ′ is obtained (S507). Next, the current value ΔIobj ′ into which ΔZ ′ is added is converted into an acceleration voltage ΔV ′ (S508), and the incident voltage of the electron beam to the sample is controlled (S509). When the incident voltage is controlled, it is sufficient that the landing voltage is constant. Therefore, ΔV ′ can be controlled on the incident voltage (electron gun) side, on the bias voltage (retarding voltage) 22 side, or You may control using both means together.

  Thus, length measurement is performed with the Landing Voltage as a set value (S510). The excitation current of the objective lens 25 at this time is an excitation current that can be focused at a distance given by the Z sensor, but the focus may be adjusted again using an image signal. After the length measurement, it is determined whether the measurement is completed by measuring the length measurement point (S511). If the measurement is completed, the wafer is unloaded (S512), and the process is terminated. If length measurement points still remain, the stage is moved to the next length measurement point (S503), and the same processing is repeated until the length measurement for all length measurement points is completed.

  Another embodiment of the present invention will be described with reference to FIGS. 9 and 13. Here, a case of a length measuring SEM apparatus adopting a retarding method without a Z-axis drive system of a wafer stage will be described. As shown in the flowchart of FIG. 13, first, the wafer 10 is loaded onto the wafer stage 21 (S601), global alignment is performed (S602), and then the stage is moved to the first length measurement point (S603). Defocus amount ΔZ is detected by Z sensors 28a and 28b (S604), and current ΔIobj corresponding to ΔZ is applied to exciting coil 26 of objective lens 25 (S605), and set to focus Fz provided by Z sensor (S605). S606). Next, the best focus point Fb is obtained by changing the wafer bias voltage (retarding voltage) ΔVb for the focus shift ΔZ ′ that could not be detected by the Z sensor (S607). Thus, the wafer bias voltage ΔVb giving the best focus is obtained (S608). Since the wafer bias voltage ΔVb corresponding to the focus deviation ΔZ ′ becomes an incident voltage error, the landing voltage is made constant by applying this ΔVb to the incident voltage side (electron gun side) or the wafer bias side, or both of them. S609).

  In this way, length measurement is performed using the Landing Voltage as a set value (S610). The excitation current of the objective lens 25 at this time is an excitation current that can focus on the distance given by the Z sensor in step 605, but the focus may be adjusted again using an image signal. After the length measurement, it is determined whether the measurement is completed by measuring the length measurement point (S611). If the measurement is completed, the wafer is unloaded (S612), and the process is terminated. If length measurement points still remain, the stage is moved to the next length measurement point (S603), and the same processing is repeated until the length measurement for all length measurement points is completed.

  Next, a case where there is a wafer stage Z-axis drive system will be described using the flowchart shown in FIG. First, the wafer 10 is loaded onto the wafer stage 21 of the length measurement SEM (S701), global alignment is performed (S702), and then the first length measurement point is moved (S703). Next, the amount of focus deviation ΔZ is detected by the Z sensor (S704), the Z axis of the wafer stage 21 is operated corresponding to ΔZ (S705), and the focus Fz determined by the Z sensor is set (S706). ). Next, the best focus point Fb is obtained by changing the wafer bias voltage (retarding voltage) ΔVb for ΔZ ′ that could not be detected by the Z sensor (S707). Thus, the wafer bias voltage ΔVb giving the best focus is obtained (S708). Since the wafer bias voltage ΔVb corresponding to the focus deviation ΔZ ′ becomes an incident voltage error, the landing voltage is made constant by applying this ΔVb to the incident voltage side (electron gun side) or the wafer bias side, or both of them. S709).

  Thus, length measurement is performed with the Landing Voltage as a set value (S710). After the length measurement, it is determined whether or not the measurement is completed by measuring the length measurement point (S711). If the measurement is completed, the wafer is unloaded (S712), and the process is terminated. If length measurement points still remain, the stage moves to the next length measurement point (S703), and the same processing is repeated until the length measurement for all the length measurement points is completed.

  Next, another embodiment of the present invention will be described. In this embodiment, the actual landing voltage of the sample incident electron beam affected by ΔV due to the charge-up of the sample surface (the surface potential change at the place irradiated with the incident electrons) is obtained, and the actual landing voltage is obtained. The length measurement value of the SEM is corrected using the magnification correction coefficient.

As shown in the flowchart of FIG. 15, first, a standard wafer of each process, for example, a semiconductor wafer that has undergone a polysilicon process, a semiconductor wafer that has undergone a SiO ₂ process, a semiconductor wafer that has undergone a SiON process, a semiconductor wafer that has undergone a SiN process, B -A semiconductor wafer that has undergone a PSG process, a semiconductor wafer that has undergone an Al process, a semiconductor wafer that has undergone a W process, a semiconductor wafer that has undergone a Cu process, or a semiconductor wafer that has undergone multiple processes, or a standard wafer of an actual device. The pitch is measured for each (Landing Voltage) (S801). An example of the measured value is shown in Table 3. Table 3 shows the results of measuring the pitch of 600nm design value by changing Vacc (Landing Voltage) from 500V to 800V every 50V for Si wafer, actual device C and actual device D adopted as standard wafers. It is a table.

  From this data table, the design value is divided by the measurement value for the standard wafer of each process to obtain a magnification correction coefficient for each Vacc, and a Vacc data table is created (S802). An example of the Vacc data table is shown in Table 4.

  Subsequently, an approximate curve expression representing the relationship between Vacc (Landing Voltage) and a magnification correction coefficient is calculated for each standard wafer from the Vacc table of each standard wafer (S803). An example of the approximate curve equation obtained from Table 4 is shown in FIG. This is the preparation stage, and data is obtained before the actual device length measurement.

  Next, the actual device is measured. For example, a wafer that has undergone the SiON film process will be described. First, the SiON wafer is loaded into a chamber and global alignment is performed. Since the SiON wafer has an insulating film, the landing voltage differs by ΔV. This ΔV is calculated by, for example, the method described with reference to the flowcharts of FIGS. 10 and 12 to 14 (S804). If it is confirmed that the Landing Voltage is shifted by ΔV, the Landing Voltage actually applied to the wafer can be found. A magnification correction coefficient at the length measurement point is obtained from the curve equation (S805).

  Finally, the true value is calculated from the measured value using the magnification correction coefficient thus obtained (S806). For example, assume that the SEM is set so that the landing voltage is 600 V on the wafer with the SiON film. Suppose that the magnification correction coefficient at this time is 0.954 for the standard wafer. Next, when ΔV was measured, when the actual Landing Voltage was 550V, the length correction value was multiplied by 0.956 using the magnification correction coefficient (assuming 0.956) when the Landing Voltage was 550V from 0.954. Find the true value.

  According to the present invention, in a length measurement using a scanning electron microscope, it is possible to obtain a true dimension value by compensating for a length measurement error caused by a potential change of the sample surface. For example, in a general length measurement SEM that does not use a retarding method, the length measurement dimension error caused by charge-up of the sample surface due to incident electron irradiation is sufficiently corrected, regardless of the sample configuration and type. A value can be obtained. In a length measuring SEM employing the retarding method, the difference between the voltage set in the electron microscope to be applied to the sample and the voltage actually applied to the sample can be accurately corrected. The acceleration voltage applied is always constant regardless of the sample configuration and type.

The flowchart explaining an example of the length measuring method of length measuring SEM by this invention. The figure which shows an example of the pattern for correction data acquisition. The flowchart explaining the other example of the length measuring method of length measuring SEM by this invention. The figure which shows an example of the pattern which is formed in the real device and can be utilized for length measurement value calibration of length measurement SEM. The flowchart explaining the other example of the length measuring method of length measuring SEM by this invention. The schematic diagram explaining the method of calculating a magnification correction coefficient for every wafer in a lot. The schematic diagram explaining the method to calculate a magnification correction coefficient for every chip | tip in a wafer. The schematic diagram explaining the method of calculating a magnification correction coefficient for every length measurement point. The schematic diagram of the objective lens vicinity of the length measurement SEM apparatus which employ | adopted the retarding system. The flowchart which shows an example of the method of controlling the incident voltage to the sample of an electron beam so that Landing Voltage may become a design value. The schematic diagram explaining the method of height adjustment of length measurement SEM. The flowchart which shows the other example of the method of controlling the incident voltage to the sample of an electron beam so that Landing Voltage may become a design value. The flowchart which shows the other example of the method of controlling the incident voltage to the sample of an electron beam so that Landing Voltage may become a design value. The flowchart which shows the other example of the method of controlling the incident voltage to the sample of an electron beam so that Landing Voltage may become a design value. The flowchart explaining the other example of the length measuring method by this invention. The figure which shows an example of the approximate curve type | formula showing the relationship between Vacc (Landing Voltage) and a magnification correction coefficient. The figure which shows the relationship between the SEM length measurement value (vertical axis) by the method of this invention, and a pitch design value (horizontal axis). The figure which shows the relationship between the SEM measured value (vertical axis) and the pitch design value (horizontal axis) by the conventional method.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Wafer, 21 ... Wafer stage, 22 ... Power supply, 23 ... Electron gun, 24 ... Incident electron beam, 25 ... Objective lens, 26 ... Excitation coil, 27 ... Deflection coil, 28a ... Light source, 28b ... Photo detector, 29 ... light beam, 30 ... extraction electrode

Claims (5)

Auto-focusing the scanning electron microscope to the measurement point using the output of the sensor that detects the distance from the objective lens to the sample;
Monitoring the change of the sample signal emitted from the sample while changing the excitation current of the objective lens to detect the excitation current ΔIobj of the objective lens corresponding to the focus shift;
Converting the excitation current ΔIobj of the objective lens into an acceleration voltage ΔV of an electron beam incident on the sample;
Changing the acceleration voltage of the electron beam incident on the sample by ΔV;
A step of measuring the measurement point;
A length measuring method using a scanning electron microscope. Auto-focusing the scanning electron microscope to the measurement point using the output of the sensor that detects the distance from the objective lens to the sample;
Monitoring the change in the sample signal emitted from the sample while changing the voltage applied to the sample to detect the sample application voltage ΔVb corresponding to the focus shift; and
Changing the electron beam acceleration voltage incident on the sample by ΔVb;
A step of measuring the measurement point;
A length measuring method using a scanning electron microscope. Obtaining a correspondence relationship between the acceleration voltage of the electron beam incident on the sample and the magnification correction coefficient in advance;
Determining the actual acceleration voltage of the electron beam incident on the measurement point on the sample;
Obtaining a magnification correction coefficient corresponding to an actual acceleration voltage of an electron beam incident on the length measurement point with reference to the correspondence relationship obtained in advance;
A step of measuring the measurement point;
Obtaining a true dimension by multiplying the length measurement value by the magnification correction coefficient;
A length measuring method using a scanning electron microscope. In the length measuring method by a scanning electron microscope according to claim 3,
The step of obtaining the actual acceleration voltage of the electron beam incident on the length measurement point on the sample includes an objective for eliminating the focus shift under the objective lens excitation condition corresponding to the acceleration voltage of the electron beam set in the scanning electron microscope. A length measuring method using a scanning electron microscope, comprising: obtaining a lens correction current ΔIobj; and converting the objective lens correction current ΔIobj into an acceleration voltage ΔV of an electron beam incident on a sample. In the length measuring method by a scanning electron microscope according to claim 3,
The step of obtaining the actual acceleration voltage of the electron beam incident on the measurement point on the sample is a sample that eliminates the focus shift under the objective lens excitation condition corresponding to the acceleration voltage of the electron beam set in the scanning electron microscope. A step of obtaining an applied voltage ΔVb, and a length measuring method using a scanning electron microscope.

Images