Charged particle beam apparatus

It is an object of the present invention to obtain an image which is focused on all portions of a sample and to provide a charged particle beam apparatus capable of obtaining a two-dimensional image which has no blurred part over an entire sample. In order to achieve the above object, the present invention comprises means for changing a focus condition of a charged particle beam emitted from a charged particle source, a charged particle detector for detecting charged particles irradiated from a surface portion of said sample in response to the emitted charged particle beam, and means for composing a two-dimensional image of the surface portion of the sample based on signals on which said charged particle beam is focused, said signals being among signals output from the charged particle detector.

 

1. A method for forming an image by a scanning charged particle apparatus, comprising steps of:

forming a plurality of two dimensional images each at different focus height by detecting particles emitted from a sample using a scanning charged particle beam;

calculating a focus evaluation value of each pixel on each of the two dimensional images;

for each of a plurality of different image points, selecting a pixel having a larger focus evaluation value than at least one other pixel from among pixels at same coordinates of the two dimensional images; and

synthesizing the selected pixels for each of the plurality of different image points, two dimensionally into a composite image.

2. A charged particle beam apparatus comprising:

a charged particle source;

a scanning deflector for scanning a charged particle beam emitted from the charged particle source on a sample;

an objective lens for adjusting a focus of the charged particle beam;

a detector for detecting particles emitted from the sample; and

an image processor for forming an image based on the particles detected by the detector, wherein said image processor:

memorizes a plurality of two dimensional images formed at different focus heights,

for each of a plurality of different image points, selects a pixel having a larger focus evaluation value than at least one other pixel from among pixels at same coordinates of the two dimensional images, and

forms the image by arranging selected pixels for each of the plurality of different image points two dimensionally.

3. A charged particle beam apparatus comprising:

a charged particle source;

a scanning deflector for scanning a charged particle beam emitted from the charged particle source on a sample;

an objective lens for focusing the charged particle beam;

a detector for detecting particles emitted from the sample; and

a controller for adjusting the objective lens, for:

adjusting a charged particle beam to a focus and computing a focal depth for an image if taken at that focus; and

shifting focus of the charged particle beam by an amount equal to or less than the computed focal depth.

4. A charged particle apparatus as claimed in claim 3, wherein said controller calculates said focal depth based on image forming conditions.

5. A charged particle apparatus as claimed in claim 4, wherein said image forming conditions include magnification of the image, an acceleration voltage of the charged particle beam, beam resolution, and/or a number of pixels of the image.

6. A charged particle apparatus as claimed in claim 3, wherein said controller has an input device for inputting a number of images, and determines a focus shift amount based on said calculated focal depth and the number of images inputted.

BACKGROUND OF THE INVENTION

The present invention relates to a charged particle beam apparatus, and more particularly to a charged particle beam apparatus having a function of properly adjusting the focus of an image obtained by charged particle beam irradiation.

A charged particle beam apparatus such as a scanning electron microscope is suitable for measuring or observing patterns formed on a semiconductor wafer, which has been becoming finer. Incidentally, samples used for such a purpose have taken a shape extending more three-dimensionally as semiconductor wafers have been multilayered. For example, currently, deeper contact holes have been formed in a sample.

SUMMARY OF THE INVENTION

A charged particle beam apparatus such as a scanning electron microscope thinly converges a beam and irradiates it onto a sample, which requires proper focusing of the beam on the sample. As semiconductor wafers have been multilayered, however, the distance between the surface of a sample and the bottom surface of a contact hole therein, for example, has become longer, causing a problem that the surface of the sample and the bottom surface of the contact hole have different focal distances. That is, focusing a beam onto the surface of the sample causes the bottom surface of the contact hole to be out of focus, producing a blurred image of the bottom surface of the contact hole.

Incidentally, Japanese Laid-Open Patent Publication No. 5-128989 (1993) discloses a technique which irradiates an electron beam onto a three-dimensional object while changing the focus of the beam, and extracts the contours of in-focus portions of the object to construct a three-dimensional image. In such an apparatus, however, when a two-dimensional image of a sample including the bottom of a contact hole is observed, for example, it is not possible to observe the details of the sample surface and the bottom portions of the contact hole since only the contour of the contact hole is indicated.

Japanese Laid-Open Patent Publication No. 5-299048 (1993) discloses another example which basically extracts contours of an object having concave/convex portions and produces a pseudo three-dimensional image in a representation similar to a contour chart.

The technique disclosed in Japanese Laid-Open Patent Publication No. 5-299048 (1993) performs differential processing on an image obtained by changing a focus, and extracts portions whose differential values exceed a preset extraction level.

This process is repeated on a plurality of images obtained by changing a focus, and finally the extracted portions are combined to extract contours of concave/convex portions of the imaged object. At that time, no consideration is given to portions whose differential values are less than the extraction level. Furthermore, since the extraction level, which is an evaluation level for determining a contour, depends on the S/N ratio of an image and the shape of the object, it is not possible to set a constant value for all portions. When there are two types of concave/convex portions in an image as shown in FIG. 16, for example, since a shape 1601 has a steep inclination, its in-focus portion has a large differential value, while since a shape 1602 has a moderate inclination, its in-focus portion has a small differential value. Therefore, if the same extraction level is applied to both shapes, the shape 1602 may not be extracted, depending on a selected extraction level. Thus, failing to set an appropriate extraction level produces an unextracted contour portion. Although the example in FIG. 16 shows only two types of concave/convex portions, an actual image has an infinite number of concave/convex portions. It is impossible to set an extraction level by which all of these contour portions are extracted. Since the above example extracts contour portions of each image separately, and no consideration is given to relationships between images whose portions have been extracted, when the extracted portions are combined to produce a composite image without setting an appropriate extraction level, some portions in the composite image may be left indefinite, or portions extracted from two or more images may overlap, as shown in FIG. 17. That is, in the invention disclosed in Japanese Laid-Open Patent Publication No. 5-299048 (1993), it is very difficult to set an extraction level, and in addition, no consideration is given to a method for processing extracted portions between images.

It is an object of the present invention to obtain an image which is focused on all portions of a sample or a certain two-dimensional area of a sample and to provide a charged particle beam apparatus capable of obtaining a two-dimensional image which has no blurred part over an entire sample.

In order to achieve the above object, a charged particle beam apparatus in accordance with the present invention comprises a charged particle source, a scan deflector for scanning a charged particle beam emitted from the charged particle source on a sample, means for changing a focus of the charged particle beam emitted from said charged particle source, a charged particle detector for detecting charged particles obtained at a portion of said sample irradiated with the charged particle beam, and means for composing a two-dimensional image of the sample as viewed from a direction of said charged particle beam source, based on signals on which said charged particle beam is focused, said signals being among signals output from the charged particle detector.

With this configuration, it is possible to select charged particles emitted from a two-dimensional area of a portion in focus from among charged particles obtained from an entire sample, and use the charged particles to form a sample image. That is, since a sample image can be constructed based on charged particles focused on an entire area or a specific two-dimensional area in a beam scan area, it is possible to compose a two-dimensional image that is focused on the charged particle beam scan area or a specific two-dimensional area thereof.

Another mode according to the present invention utilizes differential values or changes in a Sobel value at same coordinates of a plurality of images obtained by changing a focus, and uses a pixel value of the original image of an image which has a maximum value of those values to compose an image. This eliminates setting of unstable parameters as well as overlapping of portions extracted from the same image or more than one image for composition, resulting in composition of a full-focused image.

BRIEF DESCRIPTION OF DRAWINGS

FIG. 1 is a schematic diagram of a configuration of a scanning electron microscope;

FIG. 2 is a graph showing changes in a focus evaluation value as electron lens conditions are changed;

FIG. 3 is a diagram for describing focus deviations, which are a problem to be solved by the present invention;

FIG. 4 is schematic diagram for describing creation of a composite image according to the present invention;

FIG. 5 is a flowchart showing a flow of processes for extracting in-focus portions and creating a composite image according to the present invention;

FIG. 6 is a flowchart showing another flow of processes for extracting in-focus portions and creating a composite image according to the present invention;

FIG. 7 is a flowchart showing still another flow of processes for extracting in-focus portions and creating a composite image according to the present invention;

FIG. 8 is a flowchart showing a flow of processes in which image acquisition, extraction of in-focus portions, and creation of a composite image are performed in parallel according to an embodiment of the present invention;

FIG. 9 is a diagram showing indication examples for displaying composite images on a real time basis according to the present invention;

FIG. 10 is a diagram showing an example of measuring a length using a composite image according to the present invention;

FIG. 11 is a schematic diagram showing a composing process according to the present invention;

FIG. 12 is a schematic diagram showing a method for calculating a height difference between two given points in a composite image according to the present invention;

FIG. 13 is a graph showing a relationship between an excitation current and a focal distance;

FIG. 14 is a diagram showing an indication example for a display device for an apparatus according to an embodiment of the present invention;

FIG. 15 is a diagram showing an example of a GUI screen for an apparatus according to an embodiment of the present invention;

FIG. 16 is a diagram showing a method for detecting a concave/convex contour;

FIG. 17 is a diagram showing composition results in concave/convex contour detection;

FIG. 18 is a schematic diagram showing an image composing process by determining an in-focus degree using a signal of type different from one used for composition detected at the same time;

FIG. 19 is an example of a composite image obtained by characteristic quantity comparison by use of a plurality of different types of signals;

FIG. 20 is a schematic diagram of a configuration of a scanning electron microscope having a plurality of detectors;

FIG. 21 is a diagram showing the principle of a scanning electron microscope suitable for composing an image using sample images obtained for each focus by changing a focus in a stepwise manner;

FIGS. 22(A) to 22(D) are diagrams for describing a line profile for each sample image obtained by changing a focus;

FIG. 23 is a diagram showing the concept of full-focused image composition;

FIG. 24 is a diagram showing an embodiment of full-focused image composition;

FIG. 25 is a diagram showing another embodiment of full-focused image composition;

FIG. 26 is a diagram showing a configuration of a focus determination means;

FIG. 27 is a diagram showing a configuration of a noise determination means;

FIG. 28 is a diagram showing a configuration of a composing means;

FIG. 29 is a flowchart of full-focused image composition;

FIG. 30 is a diagram showing an embodiment of full-focused image composition;

FIG. 31 is a diagram showing a configuration of a preprocessing means 210 for full-focused image composition.

FIG. 32 is a diagram showing another configuration of a preprocessing means 210 for full-focused image composition;

FIG. 33 is a diagram showing an embodiment of full-focused image composition;

FIG. 34 is a diagram showing another embodiment of full-focused image composition;

FIG. 35 is a diagram showing still another embodiment of full-focused image composition;

FIG. 36 is a graph showing relationships between an observation magnification and a focal depth when the same acceleration voltage and different beam resolutions are applied;

FIG. 37 is a graph showing relationships between an observation magnification and a focal depth when the same beam resolution and different acceleration voltages are applied;

FIG. 38 is a diagram showing a schematic configuration of a scanning electron microscope used to describe an embodiment of the present invention;

FIG. 39 is a diagram showing a schematic configuration of another scanning electron microscope used to describe an embodiment of the present invention;

FIG. 40 is a flowchart showing a control flow for acquiring a series of images each having a different focus with the number of images to be acquired specified;

FIG. 41 is a flowchart showing a control flow for acquiring a series of images each having a different focus with a focal depth specified; and

FIG. 42 is a flowchart showing a control flow for acquiring a series of images each having a different focus with a focal range specified.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

Although embodiments according to the present invention will be described below with reference to a scanning electron microscope, which is a charged particle beam apparatus, the invention is not limited to this specific charged particle beam apparatus but can be applied to other charged particle beam apparatuses such as an FIB (Focused Ion Beam) apparatus which scans an ion beam on a sample to obtain a sample image.

FIG. 1 is a diagram showing a scanning electron microscope to which the present invention is applied. This scanning electron microscope incorporates an automatic focus control function. In FIG. 1, reference numerals 101 and 102 denote a sample stage and a sample to be imaged on the sample stage, respectively; 104 denotes a cathode; 105 represents a scanning coil; 106 represents an electron lens; 108 denotes a scanning coil control circuit; and 109 denotes a lens control circuit.

An electron beam 114 is scanned on the sample 102 by the scanning coil 105, and electrons emitted from the sample 102 are detected by a detector 103. A signal S1 from the detector 103 is input to an AD converter 107, which converts the signal into a digital signal S2.

The digital signal S2 is fed to an image processing processor 110 which performs image processing such as differential processing of an image and extraction of characteristic quantities, and sends the results to a control computer 111.

The processed image is also sent to a display device 112 where the image is displayed. A focus control signal S3 from the control computer 111 is input to the lens control circuit 109 so that the lens control circuit can adjust the exciting current for the lens 106 to perform focus control.

Numeral 113 denotes an input means connected to the control computer 111. This scanning electron microscope, as configured above, performs automatic focus control by automatically setting focal conditions of the electron lens to optimum values. Specifically, the scanning electron microscope scans a plurality of frames while changing the electron lens conditions and calculates and evaluates focus evaluation values using the detection signals of secondary electrons and reflected electrons so as to set optimum values to the electron lens conditions. FIG. 2 shows changes in a focus evaluation value as an electron lens condition is varied.

Here, differential values between pixels or the like are used as focus evaluation values. Specifically, the sum of differential values is calculated for each frame captured while changing the lens condition, and a lens condition in which the sum is maximized is regarded as an in-focus condition. In FIG. 2, since the focus evaluation value is at maximum (Fmax) when the exciting current for the electron lens is f, the condition f is determined as an in-focus condition.

A scanning electron microscope having an automatic focus control function such as one described above has the following problems:

First, since a focus evaluation value is calculated by applying a certain set process to a frame image or detection signals, an acquired value is applied to a frame image as a whole, that is, no consideration is given to local in-focus conditions. More specifically, when a sample has concave/convex portions, in-focus conditions are different between the upper surface and the bottom surface. However, the automatic focus control determines an optimum condition for either the upper surface or bottom surface as an optimum value for both, or it calculates a medium condition as an optimum value.

Next, the automatic focus control takes time. As seen from FIG. 2, since a plurality of frame images must be read to find an optimum focus evaluation value (that is, to find the maximum value in FIG. 2), it takes a few or a few tens of seconds to complete the control operation.

Embodiments according to the present invention provide a preferred scanning electron microscope capable of solving especially the above two problems. The configurations of apparatuses embodied according to the present invention will be described in detail below.

First Embodiment

FIG. 3 is a diagram used to describe focus deviations, which are a problem to be solved by the present invention. When a semiconductor sample with a contact hole therein is scanned by a scanning electron microscope, if the electron beam is focused on the surface of the semiconductor sample, the bottom surface of a contact hole having a high aspect ratio becomes out of focus. However, if the electron beam is focused onto the bottom surface of the contact hole, on the other hand, the surface of the sample becomes out of focus. Automatic focus control functions currently incorporated in scanning electron microscopes cannot handle a local focus deviation such as this one, and can only calculate a sample surface or some average position as a focal position.

FIG. 4 is a schematic diagram for describing creation of a composite image according to the present invention. Using the semiconductor sample with contact holes therein described in FIG. 3, two images are captured: one in which a focal position is set on the surface of the semiconductor sample and the other in which a focal position is set on the bottom surface of a contact hole. Then, in-focus portions can be extracted from each image so as to produce a composite image, which is a two-dimensional image focusing on all surfaces of the sample. These two images are registered in, for example, two frame memories.

FIG. 5 is a flowchart showing a flow of processes for extracting in-focus portions and creating a composite image according to the present invention. Two images Ai,j and Bi,j whose focal positions are shifted from each other are captured. The focal positions are shifted by sending the focus control signal 53 from the control computer 111 to the lens control circuit 109 to adjust exciting current for the electron lens 106, as described using FIG. 1. Step 501 creates a differential absolute value image for each of the two captured images (ΔxyAi,j for Ai,j and ΔxyBi,j for Bi,j). A differential absolute value image is created using values acquired by adding the absolute value of the difference between a pixel and another pixel shifted n pixels from the pixel in the x-direction to the absolute value of the difference between the same pixel and another pixel shifted n pixels from the pixel in the y-direction, as indicated by the formula (1).





Before using differential absolute value images created at step 501 as in-focus evaluation references, they are smoothed at step 502 to suppress noise influence. Step 503 determines which image is in focus, and creates a composite image. Here, the in-focus evaluation is performed based on the formula (2). That is, step 503 compares pixel values at same coordinates in the two differential absolute value images smoothed at step 502, and determines that a pixel of an original image which has a larger corresponding pixel value is in focus.



A composite image Ci,j is composed of in-focus portions of Ai,j and Bi,j. FIG. 5 illustrates composition using two images, and composition using n images can be performed by sequentially repeating the same process on a series of image pairs.

FIG. 11 is a schematic diagram showing a composing process according to the present invention. The figure illustrates an example in which pixel values from a Sobel filter are set as in-focus evaluation references. Like image differential, the Sobel filter is used to extract edge information of an image, and when a pixel value processed by a Sobel filter is large, this means that changes in pixel values around the pixel are large. That is, the pixel is in focus and is hardly blurred. Numeral 1101 indicates a plurality of images captured by changing a focus, and 1102 indicates images obtained by processing each image 1101 by use of a Sobel filter. Each of the images 1101 is registered in one of a plurality of prepared frame memories.

Pixels Sg1 through Sg5 at same coordinates in the plurality of images 1102 registered in the frame memories are compared, and of those pixels, a pixel of the largest value is extracted. Supposing that the pixel of the largest value is Sg2, a pixel value g2 of the original image of the pixel Sg2 is projected to a pixel at same coordinates in the composite image. A composite image 1103 is acquired by repeating this process for all coordinates of the image to select pixels of largest values, and arranging them to form a two-dimensional image.

FIG. 6 is a flowchart showing another flow of processes for extracting in-focus portions and creating a composite image according to the present invention. Steps 601 and 602 create differential absolute value images and smooth the images, respectively, as at steps 501 and 502 in FIG. 5. Next, step 603 determines which one of the two images is in focus, and creates a composite image. Here, the in-focus evaluation is performed based on the formula (3). That is, step 603 compares pixel values at same coordinates in the two differential absolute value images smoothed at step 602, and composes a pixel value using corresponding pixel values of the original images at a ratio of the compared pixel values. In this method, comparing with the method shown in FIG. 5, influence of a blur due to focus deviation is added. However, this method is characterized by a smooth transition portion where the in-focus state is switched from image A to image B.


FIG. 7 is a flowchart showing still another flow of processes for extracting in-focus portions and creating a composite image according to the present invention. Steps 701 and 702 create differential absolute value images and smooth the images, respectively, as at steps 501 and 502 in FIG. 5. Next, step 703 determines which one of the two images is in focus, and creates a composite image. Here, the in-focus evaluation is performed based on the formula (4). That is, step 703 compares pixel values at same coordinates in the two differential absolute value images smoothed at step 702, and composes a pixel value using corresponding pixel values of the original images at a weighted ratio of the compared pixel values. This is a method somewhere in between the methods illustrated in FIG. 5 and FIG. 6. If a weight coefficient K is set to 1, this method is equal to the method shown in FIG. 6, while this method approaches the method shown in FIG. 5 if the weight coefficient k is set to a value larger than 1.

 [ΔA]i,j≧[ΔB]i,j




FIG. 6 and FIG. 7 illustrate composition using two images, and composition using n images can be performed by comparing differential absolute values or Sobel-filtered pixel values at same coordinates in n images, and selecting the first and second largest values to apply the same process to them. Furthermore, although in FIG. 6 and FIG. 7, differential absolute values are used as in-focus evaluation quantities, Sobel-filtered pixel values can be used in the same process flow.

In the above configuration, a two-dimensional image whose every area is in focus, taking local in-focus into account, can be composed by a simple calculating means.

Second Embodiment

FIG. 8 is a flowchart showing a flow of processes in which image acquisition, extraction of in-focus portions, and creation of a composite image are performed in parallel according to an embodiment of the present invention. Numerical 813 indicates a process in which a focal position, which can be represented by an exciting current, is varied with time.

Description will be made of processes performed as time elapses, taking processes at steps 801 through 812 as examples. An image A1 at step 801 is captured at time a1. A differential absolute value image ΔA1 of the image A1 is created at step 803 before the next image capture at time a2, and an image A2 at step 802 is captured at time a2. Before the next image capture at time a3, a differential absolute value image ΔA2 is created at step 804, and the differential absolute value imageΔA1 at step 803 is compared with the differential absolute value imageΔA2 at step 804, and at step. 805, an image ΔA1 each of whose pixels is a larger differential absolute value is created. At step 806, to prepare for the next composition, an image S1 each of whose pixels is a corresponding pixel value of the original image of a differential absolute value image having a larger differential absolute value, determined on the basis of the image ΔG1 acquired at step 805, is created. Here, a composite image F1 at step 807 is created based on the image ΔG1 acquired at step 805 using the method illustrated in FIGS. 5 through 7.

The composite image F1 is displayed in a display device 112 shown in FIG. 1. Next, an image A3 at step 808 is captured at time a3. Before the next image capture at time a4, a differential absolute value image ΔA3 is created at step 809, and the differential absolute value image ΔA3 is compared with the image ΔG1 acquired at step 805, and at step 810, an image ΔG2 each of whose pixels is a larger differential absolute value is created. At step 811, to prepare for the next composition, an image S2 each of whose pixels is a corresponding pixel value of the original image of a differential absolute value image having a larger differential absolute value, determined on the basis of the image ΔG2 acquired at step 810, is created. Here, a composite image F2 at step 812 is created based on the image ΔG2 acquired at step 810, and displayed subsequently after the composite image F1. That is, a composite image of the previously captured images is completed and displayed at the time of capturing the next image. By repeating the same processes, image capture, composition processing, and indication can be performed in parallel so as to display composite images in real time. Furthermore, parallel processing such as this makes it possible to reduce the control time it takes to correct a focus, increasing the speed of automatic focus correction control.

Third Embodiment

FIG. 14 is a diagram showing an indication example for a display device 112 for an apparatus according to an embodiment of the present invention. This indication example shows a composite image of a contact hole formed in a semiconductor wafer. An apparatus according to this embodiment has almost the same configuration as that described by use of FIG. 1, and, therefore, the description provided earlier will not be repeated.

Incidentally, this apparatus embodiment is provided with a pointing device (now shown) for moving a cursor 1401 on the display screen of the display device 112. This pointing device is used to select a specific area on the display screen. This apparatus embodiment has the function of replacing an area selected by this pointing device with another image. This function will be described by way of example.

The display device 112 shown in FIG. 14 is displaying an image of a contact hole formed in a semiconductor wafer. The above composition processes have been applied to this sample image. When the cursor 1401 is placed in a center portion 1402 of the contact hole displayed in this display device 112 to select this portion, the portion, which is created by an electron beam having almost the same focal distance as that for the selected point, that is, a selected area in a specific original image among the original images each registered using one of a plurality of different focuses, is replaced by another image. This replacement process is performed based on address data of pixels which are registered in the above specific original image and have an in-focus evaluation value larger than a predetermined value or almost the same in-focus evaluation value as that for the selected point.

With this arrangement, edges of a contact hole can be made distinct. For example, a selected area (in the above example, the center portion 1402 of the hole) may be indicated in black so that it is in clear contrast with the other portions.

This method is effective when edge portions of a contact hole show little changes in their brightness. In a scanning electron microscope used to form a line profile based on image data and measure a pattern length using this line profile, unclear contrast in an edge portion causes an error in edge-position determination performed based on the line profile. Adoption of an apparatus according to this embodiment of the present invention can solve the above technical problem.

Although in the above description, a selected area in an image is replaced with another image, this apparatus may be configured so that the focus of a selected area can be adjusted. Specifically, the cursor 1401 is placed in the center portion 1402 of the contact hole in the composite image displayed in the display device 112 to select the portion.

Then, the selected area in an original image is replaced with pixels which are in a specific original image forming the selected area image and which have an in-focus evaluation value larger than a predetermined value or almost the same in-focus evaluation value as that for the selected point by the cursor 1401. With this arrangement, it is possible to perform an operation in which it looks as if to selectively adjust the focus of a specific portion of a sample image.

In an area whose image is to be replaced, the portion which has almost the same focus as that for the selected point by the cursor 1401, that is, a specific image among images each registered using one of a plurality of focuses, replaces another registered image.

Although in the above description, an image of a portion whose focus is almost equal to that for a selected point is replaced, this should not be construed restrictively; for example, a means for selecting an arbitrary area in a sample image may be provided, and an image in the selected arbitrary area is replaced based on address data of the area.

Although the above description illustrates an example in which an operator manually performs operations while observing the display screen 112, this should not be construed restrictively; for example, an image may be replaced with an image of a specific focus in an automated process.

Fourth Embodiment

FIG. 9 shows indication examples for displaying composite images on a real time basis according to the present invention. An indication example 901 displays images composed one after another, in the display monitor of a workstation, etc. by dividing the screen to accommodate each of the composite images so that the process in which the composite images are produced can be observed by comparing one image with the next. The other indication example 902 displays only the latest composite image from among the images composed one after another in the display monitor. This embodiment also has a function in which it is possible to stop the series of processes from acquisition of an image to its indication, by input from a input means 113 connected to a control computer 111 shown in FIG. 1 when an image having a desired in-focus portion has been found, while observing the series of composite images.

With this arrangement, it is possible to eliminate unnecessary electron beam irradiation that is not related to capturing of an image, and perform automatic focus control of a target portion efficiently and in a short time.

Fifth Embodiment

FIG. 10 shows an example of measuring a length using a composite image according to the present invention. A composite-image creation function according to the present invention is added to a scanning electron microscope having a function of measuring the shape of a semiconductor so that it is possible to measure a shape on a composite image by use of these functions.

Furthermore, it is possible to select an image of a specific focus, selectively read pixels indicating an in-focus evaluation value larger than a predetermined value from the image, and measure a length based on the pixels. This arrangement makes it possible to, for example, selectively read only the bottom portion image of a contact hole and measure a length based on the image so as to eliminate an error in length measurement due to an erroneous judgement of an edge position of the contact hole, resulting in realization of highly accurate length measurement.

Sixth Embodiment

FIG. 12 is a schematic diagram showing a method for determining the height difference between two given points in a composite image according to the present invention using the difference between exciting currents used when the original images of the pixels at the two given points are captured. To find the height difference between two points (pixels) g1 and g2 in a composite image 1201, it is necessary to check in which original images 1202 the corresponding pixels existed. When the point g1 existed in an original image 2 and the point g2 existed in an original image 5, the height difference between the two points g1 and g2 can be calculated from a difference Δd between a focal distance d2 corresponding to the exciting current for the original image 2 and a focal distance d5 corresponding to the exciting current for the original image 5, using a relationship between an exciting current and a focal distance shown in FIG. 13.

FIG. 15 shows a GUI screen (Guide User Interface) example for specifying the points g1 and g2 in a display device. This GUI screen has a cursor 1401 movable by a pointing device and a display column 1501 for length measurement results, therein. If it is arranged such that the points g1 and g2 can be specified by use of the cursor 1401, an operator can, for example, specify the surface of a sample and the bottom surface of a contact hole while observing the image of the contact hole so that the depth of the contact hole can be measured.

According to this embodiment, positions to be set as the points g1 and g2 (reference points for depth-direction measurement) can be accurately specified in a two-dimensional image, which makes it possible to accurately measure a depth-direction dimension of a sample, which is difficult to determine in a two-dimensional image. Since in the example shown in FIG. 15, the point g1 is set to the surface of a sample and the point g2 is set to the bottom of a contact hole, the formation depth of the contact hole can be accurately measured using the sample surface as a reference level.

Although the above description illustrates an example in which the two points g1 and g2 are specified as references for dimensional measurement, this should not be construed restrictively. A point g3 may be specified in addition to the points g1 and g2. Then, a sequence may be incorporated to measure the dimensional difference between the points g1 and g2, and the dimensional difference between the points g1 and g3 so that, for example, the depths of two contact holes can be compared. Since this specific example uses the same g1 as a reference for both contact holes, it is possible to accurately compare the formation depths of the contact holes.

An apparatus according to this embodiment can adopt a deceleration electric field forming technique in which a negative voltage is applied to a sample 102 or a sample stage 101 on which the sample is placed to produce an electric field between the sample or sample stage and an electron lens 106 which is set to a ground potential so as to reduce the energy of an irradiation electron beam when it has reached the sample (not shown).

This technique (hereinafter referred to as retarding technique) attains both reduction of color aberration by passing an electron beam through the electron lens 106 at high acceleration speed and prevention of charge-up by reducing the acceleration speed of the electron beam when it has reached a sample.

In a scanning electron microscope using a retarding technique, a negative voltage is applied to a sample as described above. The focus of an electron beam can also be adjusted by adjusting this applied negative voltage. In an embodiment according to the present invention, a negative voltage applied to a sample may be changed in a stepwise manner, and an image obtained at each step may be stored. In this case, a focal distance can be decided by the magnitude of the applied negative voltage.

As described in detail above, an apparatus according to an embodiment of the present invention can acquire a sample image which is locally in focus.

Seventh Embodiment

FIG. 18 is a schematic diagram showing an image composing process in which an in-focus degree is determined using different types of signals detected at the same time according to an embodiment of the present invention. Different types of signals that can be detected at the same time in a scanning electron microscope are secondary electrons and reflection electrons. A general SEM image uses secondary electrons, but reflection electrons are sometimes used to obtain additional information about a sample. When a full-focused image is composed using reflection electrons, if reflection electron signals are weak and, as a result, the S/N ratio of each reflection electron image having a different focus is low, an image obtained by applying a differential process or a Sobel filter to a reflection electron image sometimes cannot be used for accurately performing in-focus determination. In this case, a secondary electron image is used for in-focus determination, while a reflection electron image is used for image composition. Numerals 1801 and 1802 denote a plurality of reflection electron images and a plurality of secondary electron images, respectively, captured at the same time by changing a focus. Therefore, a point g1 in an image 1801 and a point g1 in an image 1802 have different signal intensities but are located at the same position in a sample. Each of images 1803 is obtained by applying a Sobel filter to one of the secondary electron images 1802. Pixels Sg1 through Sg5 at same coordinates in the plurality of images 1803 are compared, and of these pixels, the largest one is detected. Supposing that pixel is the pixel Sg2, a pixel value g21 of a reflection electron image acquired at the same time with a pixel value g2 of the original image corresponding to the pixel Sg2 is projected to a pixel at the same coordinates in a composite image. A reflection electron composite image 1803 can be created by applying this process to all coordinates of the images.

When use of only one type of signals for characteristic quantity comparison does not produce satisfactory results, another type of signals detected at the same time with the former type may also be used. FIG. 19 shows an example of an image composed by comparing also another type of signals. FIG. 18 shows an example in which secondary electron images are also used to compose a reflection electron image. In FIG. 18, reflection electron images are generally used for in-focus determination, and when use-of reflection electron images for in-focus determination cannot produce satisfactory results, secondary electron images are additionally used. That is, in FIG. 19, an area 1901 is determined by in-focus determination using reflection electron images at the first stage, while an area 1902 is determined by in-focus determination using secondary electron images at the second stage since the area 1902 cannot be determined using reflection electron images at the first stage.

FIG. 20 is a schematic diagram showing a scanning electron microscope having a plurality of detectors according to the present invention. Components indicated by numerals 2001 through 2014 correspond to components indicated by numerals 101 through 114 in FIG. 1. An electron beam 2014 is scanned on a sample 2002 by a scanning coil 2005, and a plurality of different types of electrons, for example secondary electrons and reflection electrons, emitted from the sample 2002 are detected by detectors. Secondary electrons are detected by a detector 2003, while reflection electrons are detected by a detector 2015. A signal Si from the detectors 2003 and 2015 is input to an AD converter 2007, which converts the signal into a digital signal S2.

Eighth Embodiment

A charged particle beam apparatus represented by a scanning electron microscope, or an optical inspection apparatus, which irradiates light such as laser light onto a sample, scans a beam on a target sample to obtain a pattern image of, for example, a semiconductor, an image sensor, or a display element. The embodiment described below relates to a technique suitable for properly scanning a sample regardless of its concave/convex portions to form a sample image, and inspecting the sample based on the sample image, in a charged particle beam apparatus or an optical inspection apparatus.

A beam scanning inspection apparatus such as a scanning electron microscope (hereinafter referred to as SEM) is suitable for measuring or observing patterns formed on a semiconductor wafer, which has been becoming finer. Of SEMs, a length measuring SEM produces a line profile based on, for example, a contrast obtained by irradiation of an electron beam onto a sample, or a signal amount of a secondary signal (secondary electrons and reflection electrons) generated from the sample to measure pattern dimensions based on the line profile.

Since SEMs thinly converge a beam and irradiate it onto a sample, it is necessary to properly focus the beam on the sample. Generally, a beam is focused so that a blur in edges of a sample structure image is minimized over the entire image.

However, semiconductor wafers recently have been multilayered and have become finer, making greater the height difference between the surface of a sample and the upper surface of a pattern formed on the sample or the bottom surface of a contact hole, as well as increasing the aspect ratio. As a result, a problem has arisen that the upper surface of a pattern and the sample surface have different proper beam focal distances.

A scanning electron microscope having an automatic focus control function as disclosed in Japanese Laid-Open Patent Publication No. 6-89687 (1994) uses a technique to change a focus in a stepwise manner and determine a proper excitation condition for an object lens based on a detection signal acquired for each focus. This method, however, can acquire only an average focus over the entire sample image, and has the problem that a portion having a height different from that of the sample surface, such as a pattern, becomes partially out of focus.

This means that it is not possible to accurately measure the formation width of a pattern formed on a semiconductor wafer, etc, and this problem has caused reduced measurement accuracy.

An object of this embodiment is to provide a beam scanning inspection apparatus capable of accurately measuring the formation width of a pattern formed on a semiconductor wafer, etc. by solving the above problem.

To accomplish the above object, a beam scanning inspection apparatus according to this embodiment forms images of a sample based on signals obtained by scanning a beam on the sample, said beam scanning inspection apparatus comprising: a means for changing a focus of said beam in a stepwise manner; a storage means for storing a sample image for each focus changed by said means; and a means for forming a sample image by overlapping the sample images stored in the storage means.

This beam scanning inspection apparatus also forms a line profile based on the overlapped sample image to measure dimensions according to the line profile.

In order to realize high integration density and high operation speed of semiconductor devices, it has been demanded to develop finer patterns formed on a semiconductor wafer, devices having a three-dimensional structure, and multilayered wiring.

On the other hand, development of finer patterns necessitates higher measurement accuracy on the inspection apparatus side, while development of devices having a three-dimensional structure and multilayered wiring further increases the aspect ratio (pattern height/pattern width) of a pattern to be measured. What this trend means to length measuring SEMs is that it is necessary to realize a higher resolution for increasing measurement accuracy and an increased focal depth for enabling observation of high aspect patterns (large-height-difference pattern) at the same time.

However, a resolution R and a focal depth DOF are proportional to each other as indicated by the following formula. A Focal depth decreases (becomes shallower) as a resolution is enhanced (becomes smaller). That is, their effects work against each other.



d: diameter of electron beam, α: half aperture angle of electron beam

Therefore, when a fine and large-height-difference pattern is measured under high resolution conditions (extremely thin electron beam), focusing on the upper surface of the pattern blurs the surface of the substrate, making it impossible to measure pattern edges on the substrate surface (desired pattern width) with high length measurement accuracy.

On the other hand, considering the current focusing technique, it is very difficult to control an electron beam so that it is always focused on the substrate surface.

An object of this apparatus embodiment is to attain both enhancement of resolution and increasing of a focal depth which are mutually contradictory as described above.

To acquire high resolution, an electron beam diameter is decreased by increasing the reduction ratios of a converging lens and an object lens. Generally, the aperture angle (2α) of an electron beam incident on a sample surface increases as the reduction ratio is increased. As the aperture angle (2α) increases, an increase in-the diameter of the electron beam (2α·ΔF) due to a focus deviation (ΔF) becomes larger. Observation with higher resolution is possible with a smaller electron beam diameter d on a focal surface since an electron beam of a smaller diameter is irradiated to an object on the focal surface. On the other hand, the image of an object placed apart from the focal surface, however little it is apart, becomes significantly blurred since the electron beam diameter (d+2α·ΔF) becomes larger.

Considering this problem, in order to attain both a large focal depth and a high resolution, this embodiment stores images captured with a large half aperture angle α and several focal positions matching the height of a pattern, and forms a sample image by overlapping these stored images. The principle is shown in FIGS. 22(A) to 22(D).

FIG. 22(C) shows a line profile obtained when a beam focused on the upper surface of a pattern (in the left in FIG. 22(A)) is scanned across the pattern. The edge profile on the upper surface of the pattern is distinct, while the edge profile on the substrate surface is not distinct. When a beam focused on the substrate surface is used, on the other hand, a line profile shown in FIG. 22(B) is produced in which the edges on the substrate surface is distinct but the edges on the upper surface is not.

Overlapping of line profiles in FIGS. 22(B) and 22(C) produces a line profile having distinct edges both on the upper surface and the substrate surface as indicated by a solid line in FIG. 22(D). A broken line in FIG. 22(D) indicates an ideal line profile obtained when an ideal point beam (all portions in focus) is scanned across the pattern. As seen from the figure, the overlapped line profile indicated by the solid line in FIG. 22(D) is close to the ideal signal intensity distribution, compared with those in FIG. 22(B) and FIG. 22(C). That is, use of a overlapped profile makes it possible to accurately measure the dimension of a large height-difference pattern even with its focal position deviated.

As described above, the effect of an increased focal depth by overlapping images becomes more distinct with higher resolution. Furthermore, with a larger electron beam half aperture angle α, that is, with a smaller electron beam diameter, the image of an object at a focal position becomes clearer, whereas the image of an object apart from the focal position, however little apart, becomes more unclear (only a background contributing to only increasing of brightness of the entire image). This means that an overlapped image has a higher resolution.

Utilizing the above principle, this embodiment comprises: a means for repeating setting of a beam focal position, and formation and capture of a predetermined number of frame images; and a means for overlapping the plurality of frame images acquired by the above means to form a sample image.

With the above arrangement, both a high resolution and an increased focal depth can be attained. As a result, it is possible to accurately measure the dimension of a fine and large-height-difference pattern regardless of accuracy of a focal position.

It should be noted that changing a focal position on a frame basis makes the control operation easy, reducing redundant time in measurement.

An apparatus according to this embodiment will be described in detail below using FIG. 21. Although the apparatus will be described with reference to a length measuring SEM used to measure the dimension of a pattern formed on a semiconductor wafer, etc., this should not be construed restrictively. A scanning ion microscope using an ion beam, or an optical microscope which forms a sample image by scanning laser light on the sample, may be used.

An electron beam 3002 emitted from an electron gun 3001 is thinly converged by a converging lens 3003 and an object lens 3004 after the beam is accelerated, and is focused on a sample wafer 3005. The converging lens is used to control the electron beam current value, while the object lens is used to adjust the focal position.

The electron beam 3002 that has been focused on the wafer is deflected by a deflector 3006 so that it is scanned on the wafer surface two-dimensionally or one-dimensionally. Part of the wafer irradiated with the electron beam 3002, in turn, emits secondary electrons 3007. The secondary electrons 3007 are detected by a secondary electron detector 3008 and converted into an electric signal. It should be noted that even though the following description assumes and thereby explains that the secondary electrons 3007 are detected, reflection elections may be detected instead of, or in addition to the secondary electrons 3007.

The electric signal is subjected to signal processing such as A/D conversion in a signal processing unit 3014. The image signal subjected to signal processing is stored in a memory unit 3015, which is a storage medium, and used to apply intensity modulation or Y-modulation to a display 3009. The display 3009 is scanned in synchronization with scanning of an electron beam 3002 on the wafer surface so that a sample image is formed in the display. When the display is scanned two-dimensionally with intensity modulation applied, an image is displayed, while when the display is scanned one-dimensionally with Y-modulation applied, a line profile is drawn.

The sample image (an image and/or a line profile) is used to measure a pattern dimension as follows:

(1) First, an image is formed, and positioning of a pattern to be measured and focusing are performed.

(2) Next, a line profile is formed by scanning an electron beam across the pattern to be measured one-dimensionally in such a direction that a desired dimension can be acquired.

(3) Then, the pattern edges are determined from the line profile according to a predetermined pattern edge determination algorithm.

(4) Then, the dimension of the pattern to be measured is calculated from the distance between the pattern edges.

(5) Finally, the calculated value acquired is output as a dimensional measurement result.

A threshold method or a linear approximation method is generally used as a pattern edge determination algorithm. In the threshold method, a line profile is cut by a given threshold level, and the intersection points of the line profile and the threshold level are determined as pattern edges.

By setting the threshold level to 50%, a measurement result close to the actual dimension can be acquired. In the linear approximation method, changes in a line profile in pattern edge portions are linearly approximated, and the intersection points of this approximated line and the base line of the line profile are determined as pattern edges.

A pattern measurement accuracy (length measurement accuracy) is inversely proportional to the diameter of an electron beam as a first approximation. Decreasing an electron beam diameter can reduce variations in length measurement values. On the other hand, decreasing an electron beam diameter also reduces an electron beam current. Specifically, an electron beam current is proportional to the square of an electron beam diameter. If a beam current is excessively reduced to decrease an electron beam diameter, the S/N ratio of a sample image will be lowered.

A large reduction in the S/N ratio deteriorates image quality, that is, resolution, and reduces length measurement accuracy. That is, to achieve high length measurement accuracy, it is necessary to attain both a reduced electron beam diameter and a sufficient S/N ratio (which lead to high resolution).

To satisfy these requirements, this embodiment employs a method in which a plurality of frames are overlapped to form a sample image. The processes employed in this method are performed in the following order.

(1) A sample image is formed using an electron beam of a small diameter. (This is called a frame image. The image quality is not good because the S/N ratio is low.)

(2) Formation of a frame image is repeated a plurality of times, for example, 20 times.)

(3) Twenty frame images obtained in (2) are overlapped to form a sample image of a sufficient resolution (an image which is captured with a small electron beam diameter and has a high S/N ratio).

(4) The sample image acquired in (3) is processed to calculate pattern dimensions.

In this embodiment, the following procedure is used to measure a pattern dimension.

A wafer 3005 to be measured is extracted from a wafer cassette 3010 and is prealigned. Prealignment is an operation performed to orient the direction of a wafer using an orientation flat and a notch formed on the wafer as references.

After prealignment, a wafer number formed on the wafer 3005 is read by a wafer number reader (not shown). A wafer number is specific to each wafer. A recipe previously registered for this wafer is read using the read wafer number as a key. After this, operations are performed according to this recipe automatically or semi-automatically.

After the recipe is read, the wafer 3005 is transferred onto a X-Y stage 3012 in a sample chamber 3011, which is kept vacuous, and is loaded there. The wafer 3005 fitted on the X-Y stage 3012 is aligned using an optical microscope 3013 fitted on the upper surface of the sample chamber 3011 and an alignment pattern formed on the wafer 3005.

The alignment is performed using the alignment pattern formed on the wafer to correct the position coordinate system on the X-Y stage and the pattern position coordinate system in the wafer. An optical microscope image acquired by magnifying the alignment pattern a few hundreds times is compared with an alignment pattern reference image registered in the memory unit 3015, and correct the stage position coordinates to exactly align the visual field of the optical microscope image with that of the reference image.

After the alignment, the visual field is moved to the position of a predetermined pattern to be measured by use of the X-Y stage 3012 and a deflection coil 3006, and the pattern to be measured is positioned. Then, after the focal position is set with the current visual field position, a sample image of the pattern to be measured is formed using the frame image overlapping method as described above and stored in the memory unit 3015.

Focusing is performed by generally observing the signal intensity distribution (a line profile) of a sample image, and adjusting the focus to a position at which the line profile is considered to be most distinct. Then, using a sample image read from the memory unit 3015, a dimension measurement unit 3016 calculates the dimension of the pattern to be measured according to a predetermined pattern edge determination algorithm.

An apparatus according to this embodiment performs the following processes at the time of forming a sample image. Utilizing the characteristic that the focal distance of a magnetic lens is almost inversely proportional to the square of the exciting current, a control unit 3017 in FIG. 21 changes the focal position of an electron beam for each predetermined number of frames by controlling the exciting current for the object lens 3004.

For example, within a predetermined range of focal distance set so that it includes a focal position acquired using the above-mentioned focusing method, the exciting current is changed from the largest value to a smaller value by steps (that is, the focal position is changed from the neighborhood of the upper surface of a pattern to the substrate surface by steps) to acquire and store frame images like four frames, six frames, eight frames, ten frames, twelve frames, and so on.

Here, the number of frames captured at one step is increased as the substrate is approached, considering the fact that the secondary electron signal amount is reduced at positions closer to the substrate since electrons emitted from the substrate are obstructed by the side walls of a pattern. This makes it possible to acquire an image having uniform brightness over the entire sample and a high resolution.

Next, the stored frame images are read out and overlapped to form a sample image after they are passed through a high pass filter to cut off low-space-frequency components. Frame images captured at positions upwardly away from the pattern surface or downwardly away from the substrate surface have almost uniform brightness (image composed of only low-space-frequency components) since the electron beam comes greatly out of focus there, and, therefore, overlapping the frame images with their low-space-frequency components cut off does not deteriorate the resolution of the sample image.

These frame image acquisition conditions are registered in the memory unit 3015 as a recipe. A recipe specifies a measuring procedure or measuring conditions to automatically or semi-automatically perform measuring operations.

Incidentally, cutoff of the low-space-frequency components may be performed before forming or storing frame images, instead of being performed as a preprocess for overlapping of frame images. Either way, a sample image having more distinct pattern edges can be acquired since the low-space-frequency components are cut off by a high pass filter, making it possible to remove signal components originated from objects apart from the focal position.

A line profile is formed based on this sample image. Then, pattern lengths are measured based on the distance between edges of the line profile. In this apparatus embodiment, these processes are performed by the control unit 3017.

Then, the operations after alignment are repeated for each of predetermin