JP2016139466A - Image vibration reduction device - Google Patents

Abstract

PURPOSE: To reduce image reduction in an image vibration reduction device for reducing vibration of an image obtained from a sample.CONSTITUTION: An image vibration reduction device includes: a stage on which a sample is moved in an X direction and a Y direction; and vibration reduction means for reducing vibration of an image on the basis of a first signal detected by a first sensor for detecting vibration or position variation, which is mounted on the stage or the sample mounted on the stage or a holder holding the sample, or a second signal of difference between the detected first signal and a signal detected by a second sensor for detecting vibration or position variation, which is mounted on a pole of an objective lens for finely squeezing electron beams to a sample or on the top plate holding the pole.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an image vibration reducing device that reduces vibration of an image obtained from a measurement object.

  Conventionally, scanning electron microscopes are used for observation and measurement of submicron materials. A scanning electron microscope scans a thinly focused electron beam onto a measurement (observation) object to generate secondary electrons, detects and amplifies the secondary electrons, converts the signal amount into contrast, and forms an image. It is a device to do.

  Since shaking of the measurement object leads to blurring of the formed image, that is, resolution degradation, in order to perform high resolution observation of nanometers, it is necessary to make the measurement object shake smaller than nanometers.

  Since the vibration that shakes the measurement object often comes from the installation environment, in order to block the vibration coming from the installation environment, the device is fixed on a vibration isolation table equipped with active suspension or in a room with sound insulation. It has been widely done to arrange.

  When the measurement (observation) target is in the nm order, it is necessary to make the swing of the measurement target smaller than the nm order.

  However, even if the vibration and sound coming from the external environment are thoroughly vibration-isolated, vibration-isolated, and sound-insulated with the above-mentioned vibration isolator and sound-insulating equipment, the amount of attenuation is not sufficient, so there is a problem that vibration appears in the image. there were.

  In addition, vibrations appearing on the image lead to blurring of the image, resulting in problems such as degradation of resolution and SNR, image quality, measurement result, and throughput of the electron microscope.

  In order to reduce conventional image vibrations, it has been considered essential to completely remove vibrations and sound waves flowing from the outside of the apparatus. Therefore, as described above, a vibration isolation table and soundproofing equipment have been installed.

  However, as a result of investigating in detail the correlation between the vibration appearing on the image and the mechanical vibration occurring in each part of the electron microscope, the vibration of the electron beam column used to generate the electron beam and irradiate the measurement object with the electron beam, It has been experimentally confirmed by the inventors that a difference from the vibration of the measurement target held on the stage that moves the measurement target on the XY plane appears as vibration on the image. Specifically, the measurement object was directly fixed to the lower part of the objective lens (not on the XY stage), and the image obtained by photographing the measurement object showed almost no vibration and an extremely high-definition image could be obtained.

  Therefore, in order to reduce the vibration on the image, it is not always necessary to remove the vibration of the entire apparatus. The top plate on which the electron beam column is mounted, the electron beam column itself, or further the vibration of the objective lens, Experiments have shown that the vibration on the image can be eliminated if the vibration of the measurement object can be matched. A mechanical method, an electric control method, an image processing method, or the like can be employed to realize this.

  In order to effectively do this, it is also effective to weaken the vibration transmission from the XY stage vibration including the peripheral vibration to the measurement target (sample).

  The present invention, contrary to the conventional idea of forcibly removing vibrations caused by the environment using a vibration isolation table, soundproofing, and sound insulation equipment, To reduce vibrations by detecting vibrations and correcting them on the computer so that the column and the object to be measured vibrate in the same way. It is aimed.

  Therefore, the present invention provides an image vibration reduction apparatus that reduces vibration of an image obtained from a sample, and is mounted on a stage that moves the sample in the X direction and the Y direction, and a sample mounted on the stage or a holder that holds the sample. The first signal detected by the first sensor for detecting the vibration or position fluctuation, or the pole of the objective lens for narrowing the detected first signal and the electron beam to the sample or the top plate holding the pole is mounted. Vibration reduction means for reducing image vibration based on a second signal that is a difference from a signal detected by a second sensor that detects vibration or position fluctuation is provided.

  At this time, the first sensor and the second sensor are laser interferometers using an acceleration sensor for detecting vibrations or a mirror for detecting positional fluctuations.

  One or a plurality of first sensors and second sensors are provided as necessary.

  Further, a plurality of first sensors and second sensors are provided outside the sample as required, and vibrations or position fluctuations at positions where the electron beam irradiates the sample are calculated.

  Further, the vibration reducing means detects a signal emitted or absorbed from a sample subjected to plane scanning while irradiating the electron beam with the objective lens while narrowing and irradiating it based on the detected first signal or differential second signal. A pixel of an image generated based on a signal detected by a detector, a predetermined continuous pixel, a pixel for one line or a pixel for a plurality of lines is detected as a first signal or a second signal of a difference. The vibration is reduced by shifting by a corresponding distance.

  Further, the vibration reducing means detects an electron beam that is scanned in a plane while irradiating the sample with an objective lens by narrowing the electron beam based on the detected first signal or the second signal of the difference. The vibration is reduced by shifting by a distance corresponding to the first signal or the second signal of the difference.

  Further, the vibration reducing means is a frequency spectrum of the apparatus calculated from the first signal obtained by forced excitation or natural vibration excitation of the apparatus or the second signal of the difference, or the second signal of the first signal or the difference. A frequency mask that removes vibration from the image using the difference in spatial frequency between the image containing the vibration component of the signal and the image not containing the vibration component is created in advance and set in the filter, and the electron beam is narrowed down with the objective lens. An image generated from a signal detected by a detector that detects a signal that is emitted or absorbed from a plane scanned while being irradiated is applied to a set frequency mask to reduce vibration from the image.

  According to the present invention, (1) since image vibration on the order of nm can be reduced, an electron microscope with sub-nm order resolution can be realized if the electron beam is focused to sub-nm.

  (2) The vibration on the image can be removed without employing a conventional large-scale vibration isolation table, soundproofing and sound insulation device.

  (3) Since the entire apparatus for removing image vibration can be simplified and downsized, the cost can be reduced.

  (4) The vibration attenuation rate can be freely set on the computer.

  (5) Not only image shaking due to external vibrations but also image shaking caused by vibrations generated by the electron beam device itself (vibration of a mechanical drive part such as a vacuum exhaust device) can be removed.

  The present invention performs electrical control so that the top plate, column, and object to be measured vibrate in exactly the same manner in response to vibrations and sound waves caused by the surrounding environment. Furthermore, the same image processing is performed on the computer, and the vibration image is removed or reduced, thereby reducing the image vibration with a simple configuration.

  FIG. 1 shows a block diagram of an embodiment of the present invention.

  In FIG. 1, an electron beam column 1 is a lens barrel that generates an electron beam 2, and includes an electron gun that generates an electron beam 2 (not shown), a focusing lens that focuses the electron beam 2 generated by the electron gun, and a focusing lens. Detection to detect secondary electrons generated in the objective lens 4 and the photomask 9 that are XY-scanned by a two-stage deflector (not shown) while irradiating the photomask (sample) 9 with the electron beam 2 focused by the lens narrowly. It is a well-known thing comprised from a container etc. A typical electron beam column 1 has a diameter of about 10 cm and a length of about several tens of centimeters.

  In the vibration mode analysis, various bending motions of the electron beam column 1 are calculated. However, since the rigidity is very large, or the electron beam is projected to be reduced to 1/100 or less due to the characteristics of the electron lens, the electron beam is reduced. It has been found that the influence of the column 1's own bending vibration on the image shake is negligibly small, and the vibration of the electron beam column 1 can be regarded as substantially the same as the vibration of the top plate on which the electron beam column 1 is installed. ing.

  The electron beam 2 is an electron beam generated by an electron gun (not shown) constituting the electron beam column 1.

  The objective lens 3 constitutes the electron beam column 1 and irradiates the photomask (sample) 9 with the electron beam 2 narrowed down. The irradiated narrowed electron beam 2 is XY scanned by a two-stage deflector (not shown).

  The vacuum chamber 4 is a container that is evacuated to a vacuum by an unillustrated evacuation system, and has a bottom plate and a top plate. As shown in the drawing, an XY stage 6, a mask pallet 7, a photomask 9, etc. In a vacuum. An electron beam column 1 for irradiating the electron beam 2 is disposed on the top plate above the vacuum chamber 4. The vacuum chamber 4 is made of a material having a magnetic shielding effect such as pure iron, and has a thickness of about 3 cm and a top plate thickness of about 6 cm.

  The XY stage 6 is fixed on the bottom plate and precisely moves the mask pallet 7 and the photomask 9 mounted on the top in the XY directions. The movement in the XY directions is controlled to move to a desired position (specified position) while measuring the positions in the X and Y directions in real time by a laser interferometer (not shown).

  The mask pallet 7 holds the photomask 9 and is made of a metal such as aluminum or a low thermal expansion ceramic. Since the electron beam 2 passes in the vicinity, a nonmagnetic material is used.

  The top plate is a sturdy top plate above the vacuum chamber (sample chamber) 4 and is rigidly connected (fixed) to the objective lens 3. Usually, pure iron or the like is used for magnetic shielding.

  The bottom plate is a sturdy bottom plate on the lower side of the vacuum chamber (sample chamber) 4.

  Next, the configuration and operation will be described.

  (1) The mask pallet 7 equipped with the photomask 9 is transported and fixed to the position shown by a sample exchange robot (not shown).

  (2) The movement of the XY stage 6 is controlled while being measured in real time by a laser interferometer (not shown), and the XY stage 6 is moved to a position where the electron beam 2 to be XY-scanned is narrowed down by the objective lens 3 and irradiated. Let As a result, a desired location (measurement location) on the photomask 9 is associated with a position where the electron beam 2 performs XY scanning.

  (3) Next, XY scanning is performed with the electron beam 2 narrowed down by the objective lens 3, and secondary electrons emitted at that time are detected and amplified by a detector (not shown) to record an image. At this time, the vibration of the photomask 9 is detected and recorded synchronously by the sensors (2) 8 and (3) 8, and the vibration of the objective lens 3 (or the top plate) is detected and recorded by the sensor (1) 5.

  (4) Here, acceleration sensors are used as the sensors 5 and 8 in order to perform vibration measurement or position measurement. The acceleration sensors are disposed on the photomask 8 side and the electron beam column 1 side (the magnetic pole of the objective lens 3 or the vicinity thereof).

  (5) The vibration signal (first signal) of the photomask 9 recorded in (3) or the difference vibration signal (second signal) between the vibration signal and the vibration signal of the objective lens 3 (or the top plate). For the image of the photomask 9 recorded in (3), a pixel, a predetermined continuous pixel, a single line pixel, or a plurality of line pixels is converted into the first signal or the pixel based on the first signal or the second signal. Shifting by a distance corresponding to the second signal (shifting in both the X direction and the Y direction) is performed to correct image vibration and reduce image shaking. Thereby, even if the photomask 9 vibrates (for example, vibrates at 110 Hz or 100 Hz which is twice the power supply frequency), the vibration of the photomask 9 is directly detected, and the corresponding distance in the image corresponds to the vibration. The vibration of the image can be reduced by shifting pixels, continuous pixels, pixels for one line, or pixels for a plurality of lines.

  (6) While acquiring the image signal, the first signal or the second signal is detected in real time to calculate the first signal or the second signal, and the corresponding pixel portion in the image signal is calculated. The vibration may be removed by shifting (shifting in the X direction and the Y direction). Here, the corresponding pixel portion in the image signal is shifted. However, the present invention is not limited to this, and the vibration may be reduced by shifting the electron beam at the corresponding pixel portion (see FIG. 13 and the like).

  Next, the operation of the configuration of FIG. 1 will be described in detail based on the explanatory diagram of FIG.

  FIG. 2 is a diagram for explaining the operation of the present invention (image vibration reduction). Here, in FIG. 1, the acceleration sensor 8 is mounted on the mask pallet 7 to detect the vibration of the photomask 9 (further, the acceleration sensor 5 is mounted to detect the vibration of the pole of the objective lens 3), and the detected value The removal of image vibration will be described using (or the difference detection value).

  As the acceleration sensor 8, for example, a high-precision and high-sensitivity semiconductor acceleration sensor having a weight smaller than 1 g and smaller than 1 cm is now available due to the development of MEMS technology. A generally available semiconductor sensor can detect an acceleration of 100 μg or less. Some can measure 1 μg at the development level. By using such a high-sensitivity sensor, it is possible to detect the shaking of the photomask 9 due to resonance, and the XY displacement amount of the mask is obtained by double integration of signals separated into components in the X and Y directions. Can know.

  When the acceleration sensor has one axis, by placing a plurality of acceleration sensors on the mask palette 7, X, Y, X-rot, and Y-rot can be independently known from the difference measurement. When using multi-axis sensors (X, Y, X-rot, Y-rot integrated type, etc.), it is necessary to extract the X and Y axis displacements by placing one sensor on the mask pallet 7 Vibration information can be detected.

  In FIG. 2, S1 generates a synchronization signal. This generates a synchronization signal supplied to the sensor (1) 5 on the objective lens 4 side and the sensor (2) 8 on the photomask 9 side in FIG. 1, and a synchronization signal for acquiring an image.

  In S2 to S5, the vibration sensors (X1), (X2), (Y1), and (Y2) detect vibrations on the photomask 9 side and the objective lens 3 side in synchronization with the synchronization signal. Here, the vibration sensors (X1) and (Y1) detect vibrations on the objective lens 4 side, and the vibration sensors (X2) and (Y2) detect vibrations on the photomask 8 side.

  In S6, an XY position difference is calculated. This is because the positional difference between the vibration signals detected by the vibration sensors (X2) and (Y2) on the photomask 9 side and the vibration signals detected by the vibration sensors (X1) and (Y1) on the objective lens 4 side is calculated. calculate.

  In S7, the magnification is adjusted. This is adjusted according to the magnification so that the position difference calculated in S6 becomes a distance corresponding to the magnification on the currently acquired image.

  In S8, an image is acquired in synchronization with the synchronization signal. This is because, while irradiating the photomask 9 with the electron beam 2 narrowed down by the objective lens 4, the X- and Y-directions are scanned in plane, and the secondary electrons emitted at that time are detected and amplified by the detector, and the image is obtained. get.

  In S9, the image pixel is multiplied. In order to shift the image acquired in S8 by an amount corresponding to image vibration, the number of pixels is multiplied (for example, 10 times) on the memory.

  In S10, the acquired images are shifted in position and superimposed. This is because the corresponding pixel, the corresponding continuous pixel, the corresponding pixel for one line, or the corresponding pixels for a plurality of lines are only displayed on the image multiplied in S9 by the positional difference corresponding to the image vibration after the magnification adjustment in S7. Shift (shift in X and Y directions) to remove image vibration. That is, pixels on the image corresponding to the position difference of image vibration calculated by the vibration sensors (X1), (X2), (Y1), (Y2) (one pixel, continuous pixels, one line of pixels, or a plurality of pixels) The line pixels) are shifted in the direction corresponding to the positional difference (X direction and Y direction), and overlapped to reduce image vibration (vibration of the photomask 9). After superposition, the number of pixels is returned to the same as the original image.

  S11 outputs a vibration reduction image.

  As described above, the difference (XY position difference) between the vibration of the photomask 9 detected by the sensors (X2) and (Y2) 8 in FIG. 1 and the vibration of the objective lens 3 detected by the sensors (X1) and (Y1) 5. Based on the above, it is possible to reduce the image vibration by shifting the position of the corresponding pixel (or the corresponding continuous pixel, the corresponding pixel for one line, the corresponding pixel for the plurality of lines) in the image acquired synchronously. . Instead of the difference, the image vibration may be reduced only by the vibration of the photomask 9.

  (1) In this embodiment, there is a synchronization signal (S1) for acquiring a signal at the same time. The signal of the vibration sensor and the image signal are acquired in synchronization with the synchronization signal. As a result, the position or vibration when the image is acquired is in a known state.

  (2) By using this information, a deviation from the relationship between the position of the objective lens 3 and the position of the photomask 9 to be measured when there is no vibration is calculated, and correction for returning to the original position is performed. it can. Since there is a proportional relationship between the output of the sensor and the actual image shake amount, appropriate magnification correction is performed so that the output value of the sensor corresponds to the shake amount on the image (S7).

  (3) Since the minimum unit of a digital image is one pixel, sub-pixel vibration cannot be expressed as it is. Therefore, before superimposing the images, the pixels are multiplied as many times as necessary so that the sub-pixels can be expressed (S9), and the superimposition is performed in consideration of the vibration amount (S10). After the superposition, an image without vibration can be obtained by performing compression processing so as to return to the original number of pixels (S11).

  FIG. 3 shows an explanatory diagram (part 1) of the present invention.

  3A shows a flowchart, and FIG. 3B schematically shows images before and after correction.

  In FIG. 3A, S21 simultaneously measures the vibration of the mask and the objective lens (top plate) and simultaneously acquires the SEM image. 1 and 2, the sensor (2) 8 uses the photomask 9, the sensor (1) 5 uses the sensor (1) 5 to simultaneously measure the vibration of the objective lens (top plate) 3, and the objective lens 3 narrows down the electron beam 2. The secondary electrons emitted when scanning the plane while irradiating the photomask 3 are detected by a detector (not shown) to simultaneously acquire an image.

  In S22, the signal of the acceleration sensor is integrated to calculate the position difference. This is because the vibration signal of the sensor (2) 8 fixed to the photomask 9 acquired in S21 and the vibration signal of the sensor (1) 5 fixed to the objective lens (top plate) 3 are double-integrated to obtain a position difference. Respectively. For example, if the positions after the double integration of the sensor (2) 8 of the photomask 9 and the sensor (1) 5 of the objective lens (top plate) 3 at a certain time are a and b, respectively, the difference is a−b And evaluate. This means that at a certain moment, the position of the photomask 9 and the objective lens (top plate) 3 is shifted by ab due to vibration. At this time, the image information is captured at a high speed so as not to cause a position measurement error.

  In S23, the magnification is adjusted. This adjusts the position difference signal calculated in S22 to a distance corresponding to the magnification of the image acquired in S21.

  In S24, the SEM image position is moved so that the difference becomes zero. This is because, after multiplying the number of pixels in the image acquired in S21, the pixel (1 pixel, multiple pixels, pixels for one line, or multiple lines) of the corresponding part based on the position difference signal after the magnification adjustment in S23. Minute pixels) is moved (shifted in the X and Y directions) to eliminate image vibration.

  S25 is superimposed on the previous SEM image. In S24, after image vibration of the corresponding pixel is removed in S24, this is superimposed on the previous SEM image.

  Here, since the scanning direction of the electron beam for acquiring the image does not necessarily coincide with the vibration direction, the obtained position information is calculated and separated into components parallel to the X and Y axes on the image. Increase the number of pixels by multiplying the acquired image (here 1 line) as necessary (for example, if the original image is 512 pixels, linear approximation to 1024 or 2048 pixels, bilinear, bicubic, nearest neighbor, etc. Conversion using approximate interpolation function). Then, the amount of image shift along the X and Y axes obtained from the sensor is taken into account and added to the previous image. This is repeated as many times as necessary to complete one image frame.

  In S26, it is determined whether the image acquisition is completed. If YES, the process proceeds to S27. In the case of NO, S21 and subsequent steps are repeated.

  S27 outputs an SEM image. This is because it is determined that all images have been acquired in YES in S26, so that the number of pixels is returned to the number of pixels of the original image as necessary, and an SEM image is output.

  By executing S21 to S27 as described above, the SEM image is moved so that the position difference becomes 0, and finally the pixel compression is performed to return to the original image size. It becomes possible to output.

  FIG. 3B is a schematic explanatory diagram of FIG. The left side shows before correction and the right side shows after correction.

  (B-1) in FIG. 3 schematically shows an example in which the image before correction is shifted to the right (shifted by vibration). FIG. 4B-2 schematically shows an example of the corrected image. This (b-2) was found to have been shifted right by b before the correction of (b-1) (the difference in position in S22), so that the difference between the positions was corrected to the left by b and the difference was reduced to zero. This is a correction made to return to the state (candle is in the center position).

  Similarly, (b-3) in FIG. 3 schematically shows an example in which the image before correction is shifted to the left (shifted by vibration). FIG. 3B-4 schematically shows an example of the corrected image. This (b-4) was found to have been shifted left by b before the correction of (b-3) (difference in S22), so that the difference was corrected to the right by b and the difference was reduced to 0 (candle) Is corrected to return to the center position.

  (B-5) in FIG. 3 is an overlay of images in which the position difference due to vibration before correction is shifted by b on the left and b on the right, and becomes “blurred” as shown.

  (B-6) in FIG. 3 is an image obtained by superimposing the images after correcting the difference in position due to the vibration after correction, and becomes “no blur” as shown.

  In addition, the correction of the position of b is performed for each line of the acquired image (or for each pixel when the position fluctuates more than a predetermined range within one line or for each predetermined pixel whose position variation is within the threshold range). To do.

  The correction of the positions 2 and b is performed only in the left-right direction (X direction) for the sake of simplicity in FIG. 3B, but the actual position correction is further performed in the vertical direction (Y direction). Do the same.

  FIG. 4 is a flowchart for explaining the operation of the present invention (part 1).

  In FIG. 4, S31 multiplies the pixels. This is because, for example, the image of the photomask 9 is acquired in S21 of FIG. 3A described above, the number of pixels of the acquired image is multiplied (for example, multiplied by 10), and the number of pixels of the original image is 1 pixel or less. The shift correction is enabled.

  In S32, the position difference is collectively shifted for each unit. This is because pixels having the same position difference (difference) between the position of the photomask 9 calculated in S22 of FIG. 3A and the objective lens (top plate) 3 are shifted together.

  In S33, a smoothing process is performed on the missing part. This is because, in S32, the corresponding pixel in the image is shifted by an amount corresponding to the difference in position between the photomask 9 and the objective lens (top plate) 3, so there is no pixel at the original position and there is a missing part. Then, smoothing processing (for example, Gaussian, median block noise removal, etc.) is performed on the missing portion, and pixels in the missing portion are interpolated.

  As described above, it is possible to perform shift correction and smoothing processing on the corresponding pixel having a difference in position between the photomask 9 and the objective lens (top plate) 3 in the acquired image, thereby reducing image vibration.

  FIG. 5 shows an explanatory diagram (part 2) of the present invention. This is because the accelerations α1 and α2 detected by the sensors (1) 5 and sensor (2) 8 in FIGS. 1 and 2 described above, and the positions (X1, Y1) calculated by double integration of these accelerations α1 and α2. ), (X2, Y2) and the difference (X1, Y1)-(X2, Y2) between the positions.

As above
(1) The accelerations α1 and α2 are detected by the sensor (1) 5 on the objective lens 3 side and the sensor (2) 8 on the photomask 9 side in FIG.

  (2) The accelerations α1 and α2 set in the table in (1) are double-integrated to calculate the positions (X1, Y1) and (X2, Y2), respectively, and set in the table.

  (3) The difference between the positions (X1, Y1) and (X2, Y2) set in the table in (2) is calculated and set in the table, and the vibration of the image can be detected.

  (4) Image vibration is reduced from the acquired image by shift-correcting the image (SEM image) acquired synchronously by the position difference (X1, Y1) − (X2, Y2) set in (3). It becomes possible. Note that image vibration may be reduced from the acquired image using only the sensor (2) 8 on the photomask 9 side.

  6 and 7 show an explanatory diagram for calculating vibration at the irradiation position of the electron beam 2 with two sensors fixed around the photomask 9 in FIG. 1 and a flowchart thereof. This will be described in detail below.

  FIG. 6 shows an explanatory diagram (part 3) of the present invention.

  In FIG. 6, sensors (11) and (12) correspond to, for example, the illustrated sensor (2) 8 and sensor (3) 8 fixed around the photomask 9 of FIG.

  The position (X11, Y11) and the position (X12, Y12) are detected by the sensor (11) (for example, the sensor (2) 8 in FIG. 1) and the sensor (12) (for example, the sensor (3) 8 in FIG. 1). Each position (vibration position) is calculated by double integration of the acceleration of the mask 9.

  The electron beam irradiation position (X1, Y1) is the electron beam irradiation position for each position (X11, Y11) and position (X12, Y12) obtained by double integration of the acceleration detected by the sensors (11), (12). The position of the vibration at is shown.

Here, X1 = (L1 (X11) + L2 (X12)) / (L1 + L2) (Formula 1)
Y1 = (L1 (Y11) + L2 (Y12)) / (L1 + L2) (Expression 2)
L1 represents the distance from the sensor (12) to the electron beam irradiation position, and L2 represents the distance from the sensor (11) to the electron beam irradiation position.

  FIG. 7 shows a flowchart (part 2) for explaining the operation of the present invention.

  In FIG. 7, S41 calculates the positions of the sensors (11) and (12). As described above, when the sensors (11) and (12) are fixed like the sensor (2) 8 and the sensor (3) 8 in FIG. 1, for example, these two sensors (11) and (12) Acceleration is detected and double integration is performed to calculate positions (X11, Y11) and (X12, Y12), respectively.

  In S42, the position of vibration at the electron beam irradiation position is calculated. This is because the respective positions (X11, Y11), (X12, Y12) calculated by double integration of the acceleration detected by the two sensors (11), (12) fixed in S41 are at the electron beam irradiation position. The position (X1, Y1) of the vibration is calculated as described on the right side using (Expression 1) and (Expression 2) described above.

  As described above, the vibration position at the electron beam irradiation position of the photomask 9 can be detected and calculated by the two sensors (11) and (12) fixed to the outside of the photomask 9.

  FIG. 8 shows a configuration diagram (part 2) of one embodiment of the present invention. FIG. 8 shows an example in which an ultra-high precision laser interference type position measuring device is used to detect vibration or position. FIG. 8 shows that the objective lens 3 includes the mirror (1) 51, the mirror (2) 51 and the distance measuring device (1) 10, and the mirror (3) 81 and the mirror (4) 81 and the distance measuring device (2) 10. An example is shown in which the position of vibration and the position of vibration of the mask palette 7 (photomask 9) are detected and correction is performed based on the detection results.

  8, the mirror (1) 51, the mirror (2) 51, and the distance measuring device (1) 10 are arranged such that the mirror (1) 51 and the mirror (2) 51 are arranged on the outer peripheral portion of the hole of the pole piece of the objective lens 3 (or The position of the mirror (1) 51 and the mirror (2) 51, here, the position information that vibrates is detected in real time by a distance measuring device (1) 10, for example, a laser interferometer. It is.

  The mirror (3) 81, the mirror (4) 81, and the distance measuring device (2) 10 fix the mirror (3) 81 and the mirror (4) 81 to the peripheral part of the mask palette 7 (or photomask 9), respectively. The distance measuring device (2) 10, for example, a laser interferometer, detects the positions of the mirror (3) 81 and the mirror (4) 81, in this case, position information that vibrates in real time.

  Next, the configuration and operation will be described.

  (1) In FIG. 8, an extremely precise and real-time detection device such as a laser interferometer is used, and vibration (position change) occurring in the photomask 9 and the objective lens 3 (or top plate) constituting the electron beam column 1 are used. ) Is detected in real time, based on image acquisition and two vibrations (position change) measured simultaneously, only the difference between the two vibrations (position change) (or photo) Image processing for shifting the acquired image (only vibration (position change) of the mask 9) and vibration appearing on the image are removed.

  (2) Therefore, a mirror (3) 81 for interference measurement and a mirror (4) 81 are arranged on the mask pallet 7 for the X axis and the Y axis, respectively, and the pole piece of the objective lens 3 of the electron beam column 1 The mirror (1) 51 and the mirror (2) 51 are arranged for the X-axis and the Y-axis, respectively, on the periphery (or top plate). Each position is continuously measured by each laser interferometer constituting the distance measuring device (1) 10 and the distance measuring device (2) 10, and a difference between the positions is calculated (if necessary, an acquired image and a continuous Recording the position of the measured difference in synchronism). The position difference is calculated based on the vibration position of the two mirrors (3) 81 and the vibration position of the mirror (4) 81 at the electron beam irradiation position on the photomask 9 (described above). Furthermore, the vibration position on the axis of the objective lens 3 based on the vibration position of the two mirrors (1) 51 and the vibration position of the mirror (2) 51. (Refer to (Expression 1) and (Expression 2) described above), and calculate the difference between the vibration positions of the two (if necessary, the acquired image and the position of the continuously measured difference are recorded synchronously. ) Corresponding to the calculated position difference (or each mirror difference), the acquired image is corrected (see FIG. 10 and later described later), and image vibration is reduced (see FIG. 10 and later described later).

  (3) For example, it is known that the frequency range of mechanical vibration that causes image vibration is about 0 to about 2 KHz, and the main component is often near 100/120 Hz (twice the power supply frequency). Moreover, since a specific place such as the XY stage 6 often resonates, it is often configured with a single frequency. Thus, the vibration phenomenon is a relatively slow phenomenon with a response time longer than 1 msec. If the position of vibration is detected in a time sufficiently shorter than this time, the position or phase of vibration can be measured accurately.

  (4) On the other hand, a laser interference position measuring apparatus having a measurement resolution of 0.03 nm or more and capable of measuring a position at a speed of 100 nsec or more is currently available. Therefore, the laser interferometer can measure the position fluctuation caused by the vibration of the objective lens 3 (or the top plate) or the photomask 9 that vibrates in the sub-nanometer in real time.

  (5) On the other hand, image acquisition by electron beam scanning is usually performed using a pixel clock of several tens of MHz or more. For example, when the pixel clock is 10 MHz, one line scan composed of 1000 pixels is 100 μs, and the image acquisition time of one frame composed of 1000 × 1000 pixels is 100 milliseconds. Accordingly, if the position of vibration is measured every time at least one line is scanned, the position information can be obtained with sub-nanometer accuracy necessary for image correction.

  FIG. 9 shows a top view. A laser interference type position measuring device is provided for each of the X and Y axes. If necessary, more than one axis may be measured to measure the rotation. By doing so, it is possible to accurately measure the vibration component of only the X axis and the Y axis.

  In FIG. 9, the distance measuring device (X) 10 is provided with three mirrors (X1) 81, a mirror (X2) 81, and a mirror (X3) 81 in the X direction on the photopallet 7, and these vibrations The position is detected in real time with extremely high accuracy. Any one mirror may be used. In the case of a plurality of mirrors, the vibration position of the position of the photomask 9 irradiated with the electron beam is calculated as described above (Formula 1) and (Formula 2).

  Similarly, the distance measuring device (Y) 10 is provided here with three mirrors (Y1) 81, a mirror (Y2) 81, and a mirror (Y3) 81 in the Y direction on the photopallet 7, and positions of these vibrations. Is detected in real time with ultra-high accuracy. Any one mirror may be used. In the case of a plurality of mirrors, the vibration position of the position of the photomask 9 irradiated with the electron beam is calculated as described above (Formula 1) and (Formula 2).

  FIG. 10 is a diagram for explaining the operation of the present invention (image vibration reduction).

  In FIG. 10, S51 notifies the synchronization signal to the image acquisition of S56, the distance measuring device (X) of S52, and the distance measuring device (Y) of S53. In FIG. 8, an image acquisition start signal is notified to the image acquisition means in S56 as a synchronization signal from a PC (personal computer) not shown in FIG. 8, and the measurement start signal as a synchronization signal is sent to the distance measuring devices (X) 10 and S53 in S52. To the distance measuring device (Y) 10.

  In S56, an image is acquired. XY scanning is performed while irradiating the finely focused electron beam 2 onto the photomask 9, and the secondary electrons emitted at that time are detected and amplified by a secondary electron detector, and an image is obtained based on the detected and amplified signals. (Secondary electron image) is acquired.

  In S57, the image pixel is multiplied. Since the difference in resolution of the target sub-nanometer cannot be expressed with the fineness of the pixel as obtained in S56, the number of pixels in the acquired image can be multiplied to increase, and the target sub-nanometer can be expressed. Like that. In this state, an image to be described later (pixels for one line or a predetermined number of pixels for one line) is superimposed while being shifted so that one frame image can be corrected.

  In S52 and S53, the distance measuring devices (X) and (Y) are connected to the mirrors (3) and (4) and the mirrors (1) and (2) in FIG. 8 based on the synchronization signal (position measurement start signal). Position information is acquired in real time by a known method based on the laser beam emitted toward and the reflected laser beam.

  In S54, the difference is evaluated. This calculates the difference in real-time position information acquired in S52 and S53 (real-time position difference) in the X direction and the Y direction.

  In S55, the magnification is adjusted. Since the change in the position on the image changes corresponding to the magnification when the image is acquired, the adjustment (correction) is performed with the magnification (normalized with the magnification).

  In S58, the position of the acquired image is shifted and superimposed. This is based on the difference in vibration after normalization in S55, and the image obtained by multiplying the image pixels obtained in S56 and S57 is shifted and corrected corresponding to the difference in vibration to cancel the difference in vibration. (It will be described later with reference to FIG. 11).

  In step S59, a vibration removal image is output.

  As described above, an acquired image generated by detecting and amplifying secondary electrons emitted at that time by performing planar scanning while irradiating the finely focused electron beam 2 on the photomask 9 (S51, S56, S57), mirror Based on the difference between signals measured in real time (3), (4), and mirrors (1), (2) (S52 to S57), the acquired image is shifted in position. Thus, it is possible to generate an image from which vibration is removed by performing correction for superimposing (S58).

  Next, image vibration removal will be described in detail with reference to FIG.

  FIG. 11A shows a flowchart, and FIG. 11B shows a schematic explanatory diagram thereof.

  11A, in S61, the positions of the mask and the objective lens (top plate) 3 are simultaneously measured, and SEM images are simultaneously acquired. As described above in S51, S56, S57 and S52, S53 in FIG. 10, the positions of the photomask 9 (mask pallet 7) are mirror (3) 81, mirror (4) 81 and a distance measuring device ( 2) At 10, the position of the objective lens 3 (or the top plate) is simultaneously measured by the mirror (1) 51, the position of the mirror (2) 51 and the distance measuring device (1) 10, and the electron narrowed down to the photomask 9. XY scanning is performed while irradiating the beam 2, and secondary electrons emitted at that time are detected and amplified to simultaneously acquire an SEM image.

  In step S62, a position difference is calculated. This calculates the difference between the positions measured by the distance measuring devices (2) and (1) measured simultaneously in S61. For example, the position of mirror (3), mirror (4) and mirror (1), mirror (2) at a certain time (the position of vibration of the one mirror when there is one mirror, the position when there are two mirrors When (the position of vibration calculated by (Expression 1) and (Expression 2)) is a and b, respectively, the difference is evaluated as a−b. This means that at a certain moment, the objective lens 3 (or the top plate) and the photomask 9 are displaced by ab due to vibration. At this time, the image information is captured at a high speed so as not to cause a position measurement error.

  In S63, the magnification is adjusted. This performs the magnification adjustment in S55 of FIG.

  In S64, the SEM image position is moved so that the difference becomes zero. This moves the position of the acquired image so that the position difference calculated in S62 becomes 0, and corrects the difference in vibration. At this time, the acquired image is subjected to pixel multiplication using an interpolation method or the like in S63 so that a desired position resolution can be expressed. Then, the position coordinates of the acquired image are shifted by a−b on the computer, and an image having a difference of 0 with respect to the reference position is generated. Since the position coordinates depend on the magnification of the acquired image, the amount of pixels to be shifted is calculated considering this as a coefficient.

  S65 superimposes on the previous SEM image. This is repeated as many times as necessary to complete one accumulated image.

  In S66, it is determined whether the image acquisition is completed. This determines whether or not one image has been acquired. If YES, the process proceeds to S67. In the case of NO, S61 and subsequent steps are repeated. In S67, an SEM image is output. This outputs an SEM image after vibration correction.

  FIG. 11B is a schematic explanatory diagram of FIG. The left side shows before correction and the right side shows after correction.

  (B-1) in FIG. 11 schematically shows an example in which the image before correction is shifted to the right (shifted by vibration). FIG. 11B-2 schematically illustrates an example of the corrected image. This (b-2) was found to have been shifted right by b before the correction of (b-1) (difference in S62), so that the difference was corrected to the left by b and the difference was reduced to 0 (candle) Is corrected to return to the center position.

  Similarly, (b-3) in FIG. 11 schematically shows an example in which the image before correction is shifted to the left (shifted by vibration). FIG. 11B-4 schematically shows an example of the corrected image. This (b-4) was found to have been shifted left by b before the correction of (b-3) (difference in S62), so that the difference was corrected to the right by b and the difference was reduced to 0 (candle) Is corrected to return to the center position.

  (B-5) in FIG. 11 is obtained by superimposing images in which the difference in position due to vibration before correction is shifted to the left by b and to the right by b, and is “blurred” as illustrated.

  (B-6) in FIG. 11 is an image obtained by superimposing the images after correcting the difference in position due to the vibration after correction, and becomes “no blur” as shown.

  In addition, the correction of the position of b is performed for each line of the acquired image (or for each pixel when the position fluctuates more than a predetermined range within one line or for each predetermined pixel whose position variation is within the threshold range). To do.

  The correction of the positions 2 and b is performed only in the left and right direction (X direction) for the sake of simplicity in FIG. 11B, but the actual position correction is further performed in the vertical direction (Y direction). Do the same.

  FIG. 12 shows a configuration diagram (part 3) of one embodiment of the present invention. FIG. 12 shows that the position of the electron beam 2 irradiated on the photomask 9 is moved and corrected by the electron beam deflector 11 instead of moving the acquired image and performing vibration correction. Except for the electron beam deflector 11 and the correction coil 12 shown in FIG. 12, they are the same as those shown in FIG. 1 and FIG.

  In FIG. 12, an electron beam deflector 11 is dynamic (real-time) at a high speed in the X and Y directions at the position for XY scanning while irradiating the photomask 9 with the electron beam 2 narrowed down by the objective lens 3. A deflector to be moved. It may be used also as a deflector for XY scanning, or may be provided independently.

  The correction coil 12 is a deflector that moves the XY scanning position at high speed dynamically (in real time) in any direction of X and Y while irradiating the photomask 9 with the electron beam 2 narrowly focused by the objective lens 3. It is an air-core coil or an electrostatic deflection plate.

  The configuration and operation will be sequentially described below.

  (1) The configuration shown in FIG. 12 directly removes (cancels) image vibration when performing image acquisition. The objective lens measured by the distance measuring devices (1) and (2) 10 in real time is used. 3 (or the top plate or the like) and the position of the photomask 9 (mask pallet 7) are measured with sub-nanometer resolution, and the electron beam 2 corresponding to the difference (or the position of the photomask 9) is measured. This extra deflection is used to correct the vibration so that the photomask 9 does not vibrate apparently. That is, the electron beam 2 that is narrowly narrowed following the vibration direction (X, Y direction) of the photomask 9 is deflected, and the vibration is corrected so that the photomask 9 does not vibrate apparently.

  (2) Therefore, the pixel resolution of the electron beam scanning device (scan deflector) used in the configuration of FIG. 12 can be scanned with the electron beam 2 at a resolution (minimum resolution to be subjected to vibration correction) that is finer than that of normal pixel scanning. It is made. For example, in normal image acquisition, a visual field of 3 microns is expressed by about 1000 pixels. In this case, the pixel resolution is 3 nm. The electron beam scanning apparatus shown in FIG. 12 is, for example, one having an ability to scan an electron beam with a resolution of 0.3 nm, which is 1/10 or less.

  (3) For example, when an image vibration that moves 0.3 nm in the right direction (X direction) is applied, the distance measuring devices (1) 10 and (2) 10 are connected to the objective lens 3 (or the top plate) and the photo It is detected that there is an error of 0.3 nm with respect to the mask 9. This detected signal is input in addition to the scanning signal of the electron beam deflector 11 (or correction coil 12) (or input alone). Then, the landing point of the electron beam 2 on the photomask 9 is a place shifted to the right by 0.3 nm (follows), and can be made coincident with the place where the electron beam 2 should be irradiated when there is no vibration. In this way, vibrations appearing on the image can be removed. In this method, since the number of pixels does not increase, it is not necessary to perform image processing with a high-speed computer.

  (4) Since the electron beam deflector 11 (or the correction coil 12) can operate sufficiently faster than MHz (such as an electrostatic deflector or an air core coil), the current position of the photomask 9 is specified (distance measurement). The position is specified with the device (2), the reference position is specified (the position of the top plate 41 (the pole piece of the objective lens) is specified with the distance measuring device (1)), and the photomask from the reference position is specified. The difference up to the position 9 is calculated, and the electron beam 2 is deflected by the electron beam deflector 11 so as to be corrected by the calculated difference. Following this, the position of the electron beam 2 is deflected by the electron beam deflector 11 in real time, and vibrations appearing in the image can be removed. When the vibration frequency to be removed is known, the feedback to the electron beam deflector 11 can be performed so as to intentionally reduce the response amount to other frequencies by increasing the weight to the frequency. In this way, feedback error is reduced, and image shake correction can be performed more accurately.

  (5) Although the position of the electron beam 2 is deflected by the electron beam deflector 11 and is strictly corrected based on the difference in vibration position, the present invention is not limited to this, and one mirror fixed to the mask pallet 7 or the like. The position of the electron beam 2 may be deflected and corrected by the electron beam deflector 11 based on the position of vibration detected only by 81.

  Next, the operation of the configuration of FIG. 12 will be described with reference to FIG.

  FIG. 13 is an explanatory diagram of the present invention (image vibration reduction, part 3).

  In FIG. 13, S71 notifies the synchronization signal to the electron beam scanning device in S75, the distance measuring device (X) in S72, and the distance measuring device (Y) in S73. In FIG. 12, an image acquisition start signal is notified as a synchronization signal from a PC (personal computer) (not shown) to the electron beam scanning device in S75, and a measurement start signal is used as a synchronization signal in the distance measuring device (X) 10 in S72. Each is notified to the distance measuring device (Y) 10 of S73.

  In S75, the electron beam scanning apparatus outputs an electron beam scanning signal. This synchronizes with the synchronization signal notified from S71, and the electron beam scanning device outputs the electron beam scanning signal to the electron beam deflector 11 (or correction coil 12) of FIG.

  In S72 and S73, the distance measuring devices (X) and (Y) are based on the synchronization signal (position measurement start signal), and the mirrors (3) and (4) and mirrors (1) and (2) in FIG. The position information is acquired in real time by a known method based on the laser beam emitted toward the laser beam and the reflected laser beam.

  In S74, the difference is evaluated. This evaluates (calculates) a difference in real-time position information acquired in S72 and S73 (a real-time position difference).

  In S76, addition is performed.

  In S77, the electron beam is deflected by the added signal (XY scanning signal). In S76 and S77, the electron beam scanning signal from S75 is added to the difference calculated in S74 (the position moved by the vibration of the photomask 9 with respect to the position (reference position) of the top plate 41 (the pole piece of the objective lens 3)). ) Is added to deflect the electron beam 2 following the vibration of the photomask 9. As a result, even if the photomask 9 vibrates and moves in position, the electron beam 2 follows and is deflected so as to irradiate the same position.

  In step S78, the vibration removal image is output.

  With the above configuration, the reference position (top plate, position of the pole piece of the objective lens 3) is measured in real time with the distance measuring device (1) 10 and the photomask 9 is measured with the distance measuring device (2). By measuring the position in real time, calculating the difference of the position of the photomask 9 with respect to the reference position, adding a signal corresponding to this difference to the electron beam scanning signal and scanning the electron beam 2 in the XY directions, the photomask 9 However, even if the position is moved by vibration, the electron beam 2 is deflected and the image vibration can be removed.

  FIG. 14 is a flowchart for explaining the operation of the present invention (part 3).

  In FIG. 14, in step S81, the positions of the mask and the objective lens (or the top plate) are simultaneously measured. As described above in S71, S72 and S73 of FIG. 13, the position of the photomask 9 (mask pallet 7) is changed by the mirrors (3) and (4) and the distance measuring device (2) 10 and the objective lens 2 The position of the (or top plate) is simultaneously measured by the mirrors (1) and (2) and the distance measuring device (1) 10.

  In step S82, the difference is calculated. This calculates the difference between the positions measured by the distance measuring devices (2) and (1) measured simultaneously in S81.

  In step S83, the electron beam irradiation position is shifted so that the difference becomes zero. This is done by adding a signal such that the position difference calculated in S82 is 0 to the signal input to the electron beam deflector 11 (or correction coil 12) in FIG. Shifting on the mask 9 follows and follows the movement of the position by the vibration of the photomask 9 to deflect the irradiation position of the electron beam 2.

  In S84, an image is acquired. In this state, in step S83, the position of the electron beam 2 is tracked and deflected following the movement of the position due to the vibration of the photomask 9 in real time, and the electron beam 2 is XY scanned to obtain an image. .

  In S85, it is determined whether the image acquisition is completed. If YES, the process proceeds to S86. In the case of NO, S81 and subsequent steps are repeated.

  In S86, an SEM image is output.

  As described above, the position of the photomask 9 with respect to the reference position (the position of the top plate, the pole piece of the objective lens 3, etc.) is detected in real time, the difference between the positions is calculated, and the electron beam is calculated based on the calculated position difference. By following and deflecting the irradiation position of the photomask 9 in FIG. 2, even if the photomask 9 vibrates, the electron beam 2 is also followed and deflected. it can.

  In FIGS. 1 to 14, the image is shifted or the electron beam is shifted based on the first signal or the difference second signal to reduce the image vibration. However, the present invention is not limited to this, or in addition to this. Furthermore, when the first signal or the second signal of the difference is equal to or greater than a predetermined threshold value, the pixel may be deleted, and a pixel that vibrates greatly at random may be deleted from the image. You may replace the pixel (pixel) of the deleted place with the pixel with the brightness | luminance averaged using the surrounding pixel.

  FIG. 15 shows the configuration of another embodiment of the present invention. In FIG. 15, a vibration mask 13 is vibrated to create a frequency mask in advance, the vibration of the photomask 9 is detected by the sensor (2) 8, and the vibration of the objective lens 3 is detected by the sensor (1) 5. A frequency mask created in advance is applied to an area of the acquired image in which the difference between the two vibrations is equal to or greater than a predetermined threshold, and the vibration image is removed from the acquired image to reduce the vibration component. Details will be sequentially described below.

  In FIG. 15, a sensor (1) 5 detects vibration of the objective lens 3.

  The sensor (2) 8 detects vibration of the photomask 9.

  The vibration means 13 is for creating a frequency mask by vibrating the XY stage 6 or the like (for example, white noise, a predetermined frequency, etc.).

  The frequency mask setting means 14 inputs an excitation vibration to the excitation device 13, measures the frequency characteristics (amplitude change with respect to the frequency) of the device, calculates an appropriate frequency mask, and sets it in the FFT filter device 16. It is.

  The image forming apparatus 15 performs planar scanning in the X and Y directions while irradiating the electron beam 2 finely focused by the objective lens 3 onto the photomask 9, and detects and amplifies the secondary electron signal emitted at that time with a detector. The image is acquired.

  The FFT filter device 16 multiplies the acquired image acquired by the image forming device 15 by a set frequency mask to generate an image from which vibration components are removed (see FIG. 17). Others are the same as those in FIG.

  Next, the operation of the configuration of FIG. 15 will be described.

  (1) The configuration of FIG. 15 is characterized in that it has means for measuring the transfer function of the electron beam apparatus itself and automatically specifying the resonance frequency unique to the apparatus. The device has an inherent mechanical resonance frequency. The vibration amplitude appearing in the image vibration is often caused by the fact that the external vibration does not appear as it is and is amplified hundreds of times by causing the resonance phenomenon inherent to the apparatus using the external signal as an energy source.

  (2) Therefore, in this embodiment, analysis is performed using a plurality of images acquired while intentionally controlling the vibration applied to the device by the vibration device 13, and the spatial frequency component that causes image degradation is specified. To do. Creates a frequency mask that accumulates image signals or images acquired in real time and removes them from the accumulated image using image processing technology for only the identified spatial frequency components, and removes vibration from the acquired image using the frequency mask Is.

  (3) Usually, one image consists of many spatial frequency components. The spatial frequency that constitutes the edge that determines the measurement accuracy is much higher than the spatial frequency of the image vibration, so even if the frequency information corresponding to the image fluctuation is removed, the edge image quality is hardly deteriorated and unnecessary. Only vibration can be removed. As a result, the distance between the edges on the image is stabilized, and higher length measurement reproducibility and high-resolution observation are possible.

  Next, the operation of the configuration of FIG. 15 will be described in detail according to the order of the flowchart of FIG.

  In FIG. 16, S71 acquires an image while applying vibration. For example, the vibration device 13 in FIG. 15 forcibly applies white noise vibration or the like to the XY stage 13 or the like to acquire the image (a) (FIG. 18A). At the same time, the sensor (2) 8 signal or the sensor (1) 5 acquires vibrations present in the apparatus. Therefore, the acquired image includes a vibration component corresponding to the signal of the sensor (2) 8 (or the difference signal between the sensor (2) 8 and the sensor (1) 5).

  In S72, two-dimensional FFT is executed. This is because the image shake due to the resonance of the XY stage 6 is observed in the image acquired in S71 (for example, the image of FIG. 18A). When FFT analysis of this image is performed, FIG. 18B is obtained, and a unique pattern corresponding to XY stage resonance appears in a low frequency region near 100 Hz. Here, the signal (2) 8 of the sensor (or the signal of the difference between the sensor (2) 8 and the sensor (1) 5) also includes 100 Hz.

  S73 acquires an image without vibration. For example, the image is acquired without being vibrated by the vibration device 13 of FIG.

  In S74, two-dimensional FFT is executed.

  In S75, the difference between the two FFT results is used as an FFT filter mask. This calculates the difference between the two-dimensional FFT result when vibration is applied in S71 and S72 and the two-dimensional FFT result when vibration is not applied in S73 and S74, and creates and sets a frequency mask for the FFT filter from this. To do.

  In S76, an image is acquired. This newly acquires a measurement target image.

  S77 measures vibration. The vibration (position) is measured by the sensor (2) 8 fixed to the mask pallet 7 and the sensor (1) 5 fixed to the objective lens (top plate) 3 in FIG. calculate.

  S78 applies an FFT filter to the image signal if it is equal to or greater than the threshold value. If the difference vibration signal measured in S77 is equal to or greater than a preset threshold value, the vibration image is removed by applying an FFT filter to the acquired image. For example, the image vibration is removed by applying a frequency mask of FIG.

  In S79, the image from which vibration is removed is output. The image from which vibration is removed in S78 is output.

  Since the resonance frequency varies slightly depending on the device, frequency analysis is performed for each device and the frequency mask is adjusted. The location to be masked (the pixel to be masked in the acquired image) is obtained as a location where the difference vibration described above is equal to or greater than the threshold, or the difference between the FFT image of the image including vibration and the FFT image not including vibration. Is also possible.

  Here, there are cases where the frequency to be masked is not single, but it is practically important to apply the mask to the main frequencies appearing in the image. The mask range is optimized so as not to impair edge information. Since the frequency of mechanical vibration fluctuates, it may be better to give the mask a little width of frequency.

  2. A new image is acquired, FFT analysis is performed, unnecessary frequency is removed by applying the mask, and then inverse FFT is applied to return to a real space image. In this way, mechanical vibration such as XY stage vibration included in an arbitrary measurement image can be removed from the image. If the frequency is obvious from the beginning, such as the power supply frequency (50 Hz, 60 Hz, etc.), it can be removed in the same way as mechanical vibration by making a mask of that frequency. For those that cause a change in the time axis such as the power supply frequency, a notch filter used in a normal electrical circuit of 50 Hz or 60 Hz can be directly inserted into the detected secondary electron signal (image).

  Since this image processing is a two-dimensional Fourier transform, it is desirable to use a high-speed image processing board mounted with a CPU, DSP, GPU or FPGA in order to perform the calculation in real time. Of course, a two-dimensional spatial filter may be applied to an image stored in a memory or the like.

  In addition, as a method for accurately knowing the mask frequency, in addition to the FFT analysis of the acquired image, the resonance frequency indicated by the output signal of the laser position measuring device or acceleration sensor which is the sensor described above may be used. It can be used as a frequency mask if it is expressed in a two-dimensional FFT space in consideration of the pixel clock at the time of secondary electron image acquisition.

  FIG. 17 is a flowchart for explaining the operation of the present invention (FFT filter procedure).

  In FIG. 17, procedure 1 acquires an image in which vibrations included in the sensor exist (a). For example, an image (for example, the image of FIG. 18A) in a state of being vibrated by the vibration device 13 of FIG. 15 is acquired. If a large vibration can be detected in a natural state, the image at that time may be used. This natural image includes a vibration component included in the signal of the sensor (2) 8 (or the difference between the sensor (2) 8 and the sensor (1) 5).

  In step 2, two-dimensional FFT is performed.

  In step 3, an image vibration component corresponding to the vibration included in the sensor signal that appears in a spot shape or a line shape is extracted (b). For example, when the image of FIG. 18A acquired in step 1 is subjected to two-dimensional FFT analysis, the result of FIG. 18B is obtained, and the image vibration component of the resonance point appearing in a spot shape or a line shape is extracted. To do.

  In step 4, the spot or line is set as a mask (c). This is because, for example, as shown in FIG. 18C, a mask (frequency mask) that surrounds the spot (a line if necessary) is added to the image vibration component of the resonance point that appears in a spot shape or a line shape in step 3. Set.

  In step 5, real time of actual measurement or an accumulated image is acquired. This acquires an image to be measured (an image acquired in real time or an image read from a memory).

  In step 6, two-dimensional FFT is performed.

  In step 7, the frequency component of step 3 is removed from the image. In these procedures 6 and 7, the image acquired in the procedure 5 is applied with an FFT filter in which a frequency mask is set, and the vibration component appearing in the spot shape or the line shape set in the procedure 3 is removed. As described above, the place where the frequency mask is applied (the place in the image where the frequency number mask in the acquired image is applied) is detected by detecting vibrations of the XY stage 6 and the objective lens (top plate) 3 respectively. The vibration component may be removed by applying a frequency mask to a place where the stage signal or the difference signal is equal to or greater than a predetermined threshold.

  In step 8, an inverse two-dimensional FFT is performed to perform image restoration. As a result, it is possible to generate an image restored by inverse FFT shown in FIG.

  FIG. 18 is an explanatory diagram (FFT filter) of the present invention.

  FIG. 18A shows an example of an image to which vibration is applied. It can be seen that resonance vibration of the XY stage 6 exists in the figure.

  FIG. 18B shows an example of the FFT analysis result of the image to which vibration is applied. This is an example of the FFT analysis result of FIG.

  FIG. 18C shows an example of a vibration frequency mask. Here, an example is shown in which a frequency mask is set so as to surround the vibration frequency of the spot-like resonance point in FIG.

  FIG. 18D shows an example of image restoration by inverse FFT. This shows an example in which a newly acquired image is subjected to the frequency mask shown in FIG. 18C, and the result is subjected to inverse FFT to restore the image.

  15 to 18, a frequency mask generated from an image acquired by vibrating the XY stage 6 or the like is used, and the corresponding frequency component is deleted from the newly acquired image using the frequency mask. Since the vibration is reduced, the image vibration can be reduced particularly effectively when the image is greatly vibrated due to resonance of the XY stage 6 or the like.

  FIG. 19 shows another operation explanatory flowchart (No. 2) of the present invention. FIG. 19 shows an example in which forced vibration is applied to the device of FIG. 15 by the vibration device 13, the transfer function of the device is measured, and a frequency mask is created based on the measured transfer function.

  In FIG. 19, S81 acquires a sensor signal while applying vibration to the apparatus. This inputs a vibration signal to the vibration device 13 of FIG. 15 to give forced vibration (for example, white noise) to the device of FIG. If there is a large natural vibration in the surrounding environment, this may be used without applying a forced vibration. Then, in the state of forced vibration (natural vibration) in the apparatus of FIG. 15, a signal is obtained using the acceleration sensor or the laser interference position sensor described above, which is the sensor (2) 8 and the sensor (1) 5.

  In step S82, a transfer function is acquired. This acquires a frequency spectrum (transfer function) based on the signal from the acceleration sensor or laser interference position sensor acquired in S81. For example, a frequency spectrum as shown in FIG.

  In step S83, the mechanical resonance frequency is specified. This specifies the resonance frequency portion of the apparatus from the frequency spectrum (for example, see FIG. 20 described later) of the apparatus of FIG. 15 acquired in S82. For example, in the frequency spectrum of FIG. 20, the peak represents the resonance component of the system including the XY stage 6, and it is specified (estimated) that this is a vibration on the image ((a) of FIG. 18 described above). (B) is identified (estimated) corresponding to 100 Hz (twice the power supply frequency)).

In step S84, the time axis frequency is converted into a spatial frequency and used as a frequency mask for the FFT filter. this is,
(1) As shown in FIG. 20, for example, when the main peak frequency is obtained as 100 Hz, vibration corresponding to this frequency component included in the sensor signal appears on the image.

  (2) An image is a collection of data having an XY two-dimensional spread. In an image acquisition apparatus using the electron beam 2, for example, an electron beam image is acquired by performing horizontal scanning at a high speed in the X-axis direction while sequentially shifting the lines along the Y-axis one pixel at a time. . Therefore, the influence of low-frequency mechanical vibration is unlikely to be placed in the X-axis direction that scans at high speed, and vibration tends to be placed in the Y-axis direction that is slowly scanned. Since the vibration frequency measured on the time axis using the sensor is different from the spatial frequency appearing on the image, it is converted into the spatial frequency appearing on the image in consideration of the frequency used for scanning and the number of pixels.

  (3) For example, when the pixel clock is 10 MHz, the X-axis direction scanning time necessary for acquiring an image having 1000 pixels in the X-axis direction is 0.1 ms. Since the mechanical vibration of 100 Hz has a period of 10 ms, vibration appears on the image with a period of 100 scanning lines in the Y-axis direction. An FFT filter is configured using a frequency corresponding to this spatial frequency as a mask. By inputting a real-time or accumulated electron beam image to the FFT filter, vibration on the image can be removed. Incidentally, since the peak of the mechanical vibration is gentle and broad, it is desirable to give a slight spread to the FFT mask by considering it.

  In S85, a measurement target image is acquired. Then, the frequency mask created in S84 is applied to the acquired image to remove the vibration component.

  FIG. 20 shows an example of a transfer function of the present invention. The horizontal axis represents frequency, and the vertical axis represents displacement / mm. FIG. 20 shows the difference between the signal of the sensor (2) 8 or the difference between the sensor (2) 8 and the sensor (1) 5 in the state where the forced vibration (for example, white noise) is applied by the vibration exciter 13 of FIG. The frequency spectrum of the apparatus (XY stage 6) of FIG. 15 is obtained from the signal. Here, it is found that the main peak frequency is twice the power supply frequency.

1 is a configuration diagram of one embodiment of the present invention. It is operation | movement explanatory drawing (image vibration reduction) of this invention. It is explanatory drawing (the 1) of this invention. It is operation | movement explanatory flowchart (the 1) of this invention. It is explanatory drawing (the 2) of this invention. It is explanatory drawing (the 3) of this invention. It is operation | movement description flowchart (the 2) of this invention. FIG. 3 is a configuration diagram (Part 2) of an embodiment of the present invention. It is explanatory drawing (the 4) of this invention. It is operation | movement explanatory drawing (image vibration reduction, the 2) of this invention. It is explanatory drawing (the 5) of this invention. FIG. 3 is a configuration diagram of an embodiment of the present invention (No. 3). It is operation | movement description (image vibration reduction, the 3) of this invention. It is operation | movement description flowchart (the 3) of this invention. It is another Example block diagram of this invention. It is another operation | movement description flowchart of this invention. It is an operation | movement explanatory flowchart (procedure of FFT filter) of this invention. It is explanatory drawing (FFT filter) of this invention. It is another operation | movement description flowchart (the 2) of this invention. It is an example of a transfer function of the present invention.

1: electron beam column 2: electron beam 3: objective lens 4: vacuum chamber 5: sensor 51: mirror 6: XY stage 7: mask pallet 8: sensor 81: mirror 9: photomask (sample)
10: Distance measuring device 11: Electron beam deflector 12: Correction coil 13: Excitation device 14: Frequency mask setting means 15: Image forming device 16: FFT filter device

Claims (10)

In an image vibration reduction device that reduces vibration of an image obtained from a sample,
A stage for moving the sample in the X and Y directions;
The first signal detected by the first sensor for detecting vibration or position fluctuation mounted on the stage or the sample mounted on the stage or the holder for holding the sample, or the detected first signal and the electron beam. An image based on a second signal that is a difference from a signal detected by a second sensor that detects vibrations or position fluctuations mounted on a pole of an objective lens that is finely focused on the sample or on a top plate that holds the pole. An image vibration reducing apparatus comprising vibration reducing means for reducing the vibration of the image.   The image vibration reduction apparatus according to claim 1, wherein the first sensor and the second sensor are an acceleration sensor that detects vibration or a laser interferometer that uses a mirror that detects position fluctuation.   3. The image vibration reducing device according to claim 1, wherein one or a plurality of the first sensor and the second sensor are provided as necessary.   4. The apparatus according to claim 3, wherein a plurality of first sensors and second sensors are provided outside the sample as needed to calculate vibration or position fluctuation at a position where the electron beam irradiates the sample. The image vibration reducing device according to any one of claims 1 to 3.   The vibration reduction means emits or absorbs a signal emitted or absorbed from a sample scanned in plane while irradiating an electron beam with the objective lens while narrowing and irradiating it based on the detected first signal or the second signal of the difference. A pixel of an image generated based on a signal detected by a detector to be detected, a predetermined continuous pixel, a pixel for one line, or a pixel for a plurality of lines is detected by the detected first signal or a second difference. 5. The image vibration reducing apparatus according to claim 1, wherein the vibration is reduced by shifting a distance corresponding to a signal.   The vibration reducing means, based on the detected first signal or the second signal of the difference, narrows the electron beam with the objective lens and finely irradiates the sample with a plane scan, 5. The image vibration reducing apparatus according to claim 1, wherein the vibration is reduced by shifting by a distance corresponding to the detected first signal or differential second signal.   The vibration reducing means is a frequency spectrum of the apparatus calculated from the first signal or the second signal of the difference obtained by forced vibration or natural vibration excitation of the apparatus, or the first signal or the difference of the difference. A frequency mask that removes vibration from the image using the difference in spatial frequency between the image including the vibration component of the second signal and the image not including the vibration component is created in advance and set in the filter, An image generated from a signal detected by a detector that detects a signal emitted or absorbed from a plane-scanned sample while irradiating with a narrow beam is applied to the set frequency mask to reduce vibration from the image. The image vibration reducing device according to claim 1, wherein the image vibration reducing device is one of the following.   8. The image vibration according to claim 7, wherein the image signal region in which the first signal or the difference second signal exceeds a predetermined threshold is subjected to the set frequency filter to reduce vibration from the image. Reduction device. In the image vibration reduction method for reducing the vibration of the image obtained from the sample,
A stage for moving the sample in the X and Y directions;
A first sensor for detecting vibration or position variation mounted on the stage or a sample mounted on the stage or a holder holding the sample;
If necessary, a pole of an objective lens for narrowing the electron beam to the sample or a second sensor for detecting vibration or position variation mounted on a top plate holding the pole is provided.
Image vibration is reduced based on the first signal detected by the first sensor or the second signal that is the difference between the detected first signal and the signal detected by the second sensor. An image vibration reduction method comprising a vibration reduction step. In the image vibration reduction method for reducing the vibration of the image obtained from the sample,
A stage for moving the sample in the X and Y directions;
A first sensor for detecting vibration or position variation mounted on the stage or a sample mounted on the stage or a holder holding the sample;
If necessary, a pole of an objective lens for narrowing the electron beam to the sample or a second sensor for detecting vibration or position variation mounted on a top plate holding the pole is provided.
The first signal detected by the first sensor, or the difference between the detected first signal and the signal detected by the second sensor, obtained by forced excitation or natural vibration excitation of the apparatus. 2 using the frequency spectrum of the device calculated from the two signals, or the difference in spatial frequency between the image including the vibration component of the first signal or the second signal of the difference and the image not including the vibration component. A frequency mask that removes vibrations from the sample was created in advance and set as a filter, and it was detected by a detector that detects the signal emitted or absorbed from the sample that was scanned in plane while irradiating the objective lens with a finely focused electron beam. An image vibration reduction method characterized by comprising a vibration reduction step of multiplying the image generated from the signal by the set frequency mask and reducing the vibration from the image. .