JP2011013227A - Method and device for inspecting defect - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method and device for inspecting defects for easily determining the optimum imaging condition and threshold to detect the defects with high sensitivity.SOLUTION: A simulation image of a pattern including the defects, and a simulation image of a pattern not including the defects are generated at first based on the plurality of imaging conditions, then, a difference signal between both in each of the plurality of imaging conditions is calculated, and the difference signal is displayed at last on a display in each of the plurality of imaging conditions. The difference signal between the simulation images is calculated as to the plurality of predicted fatal defects to display the difference signal the defect by the defect.

Description

The present invention relates to a defect inspection method and apparatus suitable for use in defect inspection of patterns such as wafers and masks in the manufacture of semiconductor devices. Relates to the device.

  The main manufacturing process of a semiconductor device is roughly divided into a substrate process and a wiring process. In the substrate process, isolation formation, well formation, gate insulating film formation, gate electrode formation, source / drain formation, capacitor structure formation, interlayer insulating film formation, and planarization are performed. In the wiring process, contact plug formation, interlayer insulating film formation, planarization, and metal electrode wiring formation are repeated for multilayer wiring, and finally a passivation film is formed. An inspection process is provided at the main point of the manufacturing process, and a wafer in the middle of processing is extracted to inspect pattern defects. Here, the pattern defect is a general term for a pattern shape abnormality (short circuit, disconnection, etc.), surface foreign matter, etching residue, scratch, and the like. The purpose of pattern defect inspection is to first detect abnormalities in the manufacturing process at an early stage, and secondly to identify the defect generation process and its cause, and high detection sensitivity is required as devices become finer. Yes.

  Now, several hundred devices (also called chips) having the same pattern are produced on one wafer. In the memory portion of the device, a large number of cells having a repetitive pattern are formed. Therefore, in pattern defect inspection, a method of comparing pattern images between adjacent chips or adjacent cells is used.

  As the pattern defect inspection, an optical inspection that optically images a wafer pattern is used. Optical inspection has a relatively high throughput and is mainstream in mass production lines.

  In the optical inspection, there are a plurality of imaging conditions such as a wavelength band, a polarization state, a shape and coherence of an illumination light source, and characteristics of a spatial frequency filter. Since these imaging conditions greatly affect the contrast and brightness of the image, it is necessary to set optimum conditions in order to obtain high detection sensitivity. For example, a test inspection is performed while automatically changing a plurality of imaging conditions, and a series of images is accumulated. For each condition, a list of defect detection information (image, contrast, etc.) is displayed. The user can select an optimum condition based on the test inspection result. An example of such a method is described in Patent Document 1, for example.

  On the other hand, in the defect determination after pattern imaging, the difference signal between the detected image and the reference image is compared with a preset threshold value. When the difference signal is larger than the threshold value, it is determined as a fatal defect, and when the difference signal is smaller than the threshold value, it is determined as a pseudo defect or a non-fatal defect.

  Therefore, it is necessary to set a threshold value so that the detection rate of fatal defects is high and the detection rate of pseudo defects and non-fatal defects is low. Here, a fatal defect is a defect that causes a device failure, a non-fatal defect is a defect that is acceptable as a device, and a pseudo defect is noise caused by various causes. For example, a test inspection is performed to create a frequency distribution of the number of detections with respect to the level of the difference signal. Next, an approximate curve of the frequency distribution is obtained, and the level of the difference signal at which the approximate frequency value is 0 is set as the optimum threshold value. Thereby, a normal pattern is not detected and only a defect can be detected. An example of such a method is described in Patent Document 2, for example.

JP 2002-303586 A JP-A-5-47886

  In the above-described conventional technology, the imaging condition and the threshold are determined based on the result of the test inspection. However, in fact, the fatal defect may be different in type, shape, and size from the defect detected by the test inspection. For this reason, the technique disclosed in Patent Document 1 has a problem that the imaging conditions determined by the user are not necessarily optimal for detecting a fatal defect. In the technique of Patent Document 2, no consideration is given to the lethality of defects. Therefore, there are many false detections of non-fatal defects, which sometimes hinders inspection data analysis and defect review / classification.

  An object of the present invention is to provide a defect inspection method and a defect inspection apparatus capable of easily determining optimum imaging conditions and threshold values so that highly sensitive defect detection is possible.

  According to the present invention, first, a simulation image of a pattern including a defect and a simulation image of a pattern not including a defect are generated based on a plurality of imaging conditions, and then a difference signal between the two is calculated for each of the plurality of imaging conditions. Finally, a differential signal is displayed on the display device for each of a plurality of imaging conditions.

  According to the present invention, a simulation image of a pattern including a defect is generated for each of a plurality of defects according to the input imaging condition, and then a simulation image of a pattern not including the defect is generated. A difference signal between each simulation image of a pattern including a plurality of defects and a simulation image of a pattern not including a defect is calculated, and the difference signal is displayed by a graph for each of the plurality of defects.

  According to the present invention, it is possible to easily determine optimal imaging conditions and threshold values so that highly sensitive defect detection is possible.

It is a figure which shows schematic structure of the optical pattern inspection apparatus by this invention. It is a figure which shows the flow of the determination method of the imaging condition in the optical pattern inspection method by this invention. It is a figure which shows the example of the simulation image of the pattern containing a defect. It is a figure which shows the example of the simulation image of the pattern which does not contain a defect. It is a figure which shows the example of the difference signal of a simulation image. It is a figure which shows the example of the relationship between a defect and a difference signal.

  As an embodiment of the present invention, an optical inspection apparatus for a semiconductor wafer pattern will be described. FIG. 1 shows a schematic configuration of an optical inspection apparatus according to the present invention. The optical inspection apparatus of this example includes a stage 2 on which a wafer 1 is mounted, an optical system 3, an image sensor 4, an AD converter 5, an image storage unit 6, an image processing unit 7, a control unit 8, a user interface unit 9, and An image simulation unit 10 is included. The optical system 3 includes a light source 30, a wavelength selection filter 31, a condenser lens 32, a polarization beam splitter 33, a polarization control unit 34, and an objective lens 35.

  The user interface unit 9 includes input means such as a keyboard and a mouse and display means such as a display device.

  The light from the light source 30 passes through the wavelength selection filter 31 and is collected by the condenser lens 32. By appropriately selecting the wavelength selection filter 31, a desired wavelength band can be obtained in consideration of the reflectance of the material constituting the wafer 1. The aperture stop surface of the condenser lens 32 is provided with an aperture (not shown). By appropriately selecting the aperture, the shape of the illumination light source can be set to a circle, an annular zone, or the like, or illumination coherence. Can be controlled.

  The light from the condenser lens 32 is then reflected by the polarization beam splitter 33, becomes linearly polarized light, passes through the polarization controller 34, and becomes elliptically polarized light. As described in Japanese Patent Application Laid-Open No. 2000-155099, the polarization control unit 34 includes a half-wave plate and a quarter-wave plate, and an elliptical shape is obtained by adjusting the direction of each wave plate. The major axis direction and ellipticity can be controlled. Further, the light from the polarization control unit 34 enters the wafer 1 through the objective lens 35. The light reflected and diffracted by the wafer 1 passes through the objective lens 35 and the polarization controller 34 again, passes through the polarization beam splitter 33, and travels toward the image sensor 4. A spatial frequency filter (not shown) can be inserted into the exit pupil plane of the objective lens 35, and the intensity distribution of the diffraction image can be controlled.

  An image detected by the image sensor 4 is converted into a digital signal by the A / D converter 5 and recorded in the image storage unit 6. On the other hand, the image storage unit 6 records a reference image obtained in an area adjacent to the detected image area and having the same pattern. The detected image and the reference image are transmitted to the image processing unit 7, and after performing necessary corrections, a difference signal between them is calculated. The difference signal is compared with a threshold value set in advance by the user to detect a defect. For example, a defect smaller than the threshold is determined as a non-fatal defect or a pseudo defect, and a defect larger than the threshold is determined as a fatal defect. A method for obtaining the threshold will be described below. The processing in the image simulation unit 10 will be described in detail below.

  Next, a method for determining the optimal imaging condition for accurately determining whether or not a defect is a fatal defect will be described with reference to FIG. First, in step S1, the user, via the user interface unit 9, information on the wafer (material, film thickness, pattern shape, etc.), information on expected fatal defects (position, type, shape, size, etc.), Input a plurality of imaging conditions. Here, the imaging conditions include the wavelength band, the shape and coherence of the illumination light source, the characteristics of the polarization controller, the characteristics of the spatial frequency filter, and the like.

  In step S2, the image simulation unit 10 performs numerical calculation using the input data, and generates a simulation image for each imaging condition. The specific calculation method is as follows. First, the structure and refractive index of the wafer including the defect, the wavelength of the illumination light, the incident angle, the polarization state, etc. are used as input data, and the Maxwell equation is solved by, for example, the time domain difference method to obtain the near electric field near the wafer surface. Next, using this near electric field as input data, imaging calculation is performed in consideration of the numerical aperture and magnification of the objective lens, the characteristics of the polarization controller, the characteristics of the spatial frequency filter, etc., and the light intensity on the image sensor is obtained. Then, the light intensity is integrated for each pixel of the image sensor, and a simulation image (detected image) of a pattern including a defect is generated and output. Similarly, a simulation image (reference image) of a pattern not including a defect is also generated and output. Methods for generating simulation images using input data are known and will not be described in further detail here.

  In step S <b> 3, the control unit 8 calculates a difference signal between a simulation image (detected image) of a pattern including a defect and a simulation image (reference image) of a pattern not including a defect for each imaging condition.

  In step S4, the user interface unit 9 displays a list of imaging condition and defect difference signals. In step S5, the user can select an optimal imaging condition for the expected fatal defect while viewing the relationship between the displayed imaging condition and the difference signal between the simulation images.

  Next, an example of a method for determining optimum imaging conditions and threshold values will be described with reference to FIGS. 3, 4 and 5. Here, it is assumed that the pattern formed on the wafer is a line and space pattern. FIG. 3 shows a simulation image (detected image) of a pattern including a defect calculated based on three types of imaging conditions A, B, and C. The vertical axis represents pixel signal intensity, and the horizontal axis represents pixel coordinates. Here, the three types of imaging conditions A, B, and C correspond to the major axis direction of elliptically polarized light. The defect is arranged in the center line (position where the coordinate value of the horizontal axis is approximately 150). It is the imaging condition B that the intensity of the center line decreases most.

  FIG. 4 is a simulation image (reference image) of a pattern not including a defect calculated based on three types of imaging conditions A, B, and C. The vertical axis represents pixel signal intensity, and the horizontal axis represents pixel coordinates. The contrast increases in the order of imaging conditions C, A, and B.

  FIG. 5 shows difference signals between the simulation image of the pattern including the defect in FIG. 3 and the simulation image of the pattern not including the defect in FIG. 4 for three types of imaging conditions A, B, and C. Here, the difference signal of the simulation image is normalized by the maximum value of the intensity of the reference image. The vertical axis in FIG. 5 represents the intensity of the normalized difference signal, and the horizontal axis represents the pixel coordinates. The user may compare the difference signals of the three simulation images and select the imaging condition B having the largest difference signal. Thereby, highly sensitive defect detection becomes possible.

  FIG. 6 shows a relationship between a defect obtained based on a predetermined imaging condition and a difference signal. Here, a case is shown in which the user selects the imaging condition B from the display result of FIG. 5 and obtains a difference signal for five fatal defects abcde based thereon. First, a simulation image of a pattern including a defect is generated for each of the five defects according to the imaging condition B, and then a simulation image of a pattern not including the defect is generated according to the imaging condition B. A difference signal between each of the simulation images of the pattern including the five defects and the simulation image of the pattern not including the defect is obtained and normalized. If the difference signal is represented by a bar graph for each defect, the graph of FIG. 6 is obtained.

  The user can see from the graph of FIG. 6 that c is the fatal defect with the smallest difference signal. Therefore, the value of the difference signal of the defect c or a value slightly smaller than that may be set as the threshold value X. Using the threshold value X set in this way, defects actually detected by the optical inspection apparatus are discriminated. Thus, in this example, it is possible to reliably detect a fatal defect and reduce false detection of a non-fatal defect.

  Here, the pattern inspection using light has been described for a semiconductor wafer, but the present invention can also be applied to a pattern inspection using an electron beam. The present invention can also be applied to pattern inspection of a mask or a liquid crystal device in a semiconductor lithography process.

  The example of the present invention has been described above, but the present invention is not limited to the above-described example, and various modifications can be made by those skilled in the art within the scope of the invention described in the claims. It will be understood.

DESCRIPTION OF SYMBOLS 1 ... Wafer, 2 ... Stage, 3 ... Optical system, 4 ... Image sensor, 5 ... A / D converter, 6 ...
Image storage unit, 7 ... Image processing unit, 8 ... Control unit, 9 ... User interface unit, 10 ... Image simulation unit, 30 ... Light source, 31 ... Wavelength selection filter, 32 ... Condenser lens,
33 ... Polarizing beam splitter, 34 ... Polarization controller, 35 ... Objective lens

Claims (4)

In a defect inspection method for detecting a defect generated in a target pattern by comparing a difference signal between a detected image and a reference image with a threshold value,
A first simulation detection image generating step for generating a plurality of simulation detection images including a defect based on each of a plurality of imaging conditions according to a target pattern including an expected defect;
A simulation reference image generation step for generating a plurality of simulation reference images not including a defect based on each of the plurality of imaging conditions according to the target pattern including no defect,
For each of the plurality of imaging conditions, a difference signal between the simulation detection image including the defect generated by the first simulation detection image generation step and the simulation reference image not including the defect generated by the simulation reference image generation step is calculated. A first difference signal calculation step;
A first display step for displaying a difference signal of each of the plurality of imaging conditions calculated in the first difference signal calculation step;
An imaging condition selection step of selecting one imaging condition from the plurality of imaging conditions based on the display in the first display step;
A second simulation detection image generation step for generating a simulation detection image based on the selected one imaging condition in relation to each target pattern including each of a plurality of expected defects;
For each of the plurality of expected defects, a difference signal between the simulation detection image generated by the second simulation detection image generation step and the simulation reference image of the selected one imaging condition generated by the simulation reference image generation step A second difference signal calculation step for calculating
A second display step for displaying a difference signal of each of the plurality of expected defects calculated by the second difference signal calculation step;
Threshold value for determining whether the difference signal displayed in the second display step is a fatal defect with a value of the smallest difference signal or a value slightly smaller than the value of the smallest difference signal Defect inspection method including a selection step to select as. The defect inspection method according to claim 1, wherein the second display step displays a difference signal of each of the plurality of expected defects by a graph. In a defect inspection apparatus that detects a defect generated in a target pattern by comparing a difference signal between a detected image and a reference image with a threshold value for determining whether or not the defect is a fatal defect,
An input unit that specifies one imaging condition from among a plurality of imaging conditions and sets the threshold value;
Image simulation that generates a simulation detection image that includes a defect according to an imaging condition and includes a defect according to an imaging condition, and generates a simulation detection image that does not include a defect according to the imaging condition according to an object pattern that does not include a defect And
A first calculation unit that calculates a difference signal between a simulation detection image including a defect generated by the image simulation unit and a simulation reference image not including the defect;
A display unit for displaying the difference signal calculated by the first calculation unit;
A second arithmetic unit that detects the fatal defect;
The display unit
Displaying a difference signal for each imaging condition calculated by the first calculation unit for designating one imaging condition by the input unit;
The input unit is
According to the designation of the one imaging condition, out of the difference signals for each defect according to the one imaging condition, a value of a minimum difference signal or a value slightly smaller than the value of the minimum difference signal is determined to be a fatal defect. Can be selected as a threshold for determining whether
The defect inspecting apparatus, wherein the second arithmetic unit detects the fatal defect based on a threshold selected by the input unit. 4. The defect inspection apparatus according to claim 3, wherein the display unit displays a difference signal for each defect in a graph.