Local Alignment Point Calibration Method in Die Inspection

Abstract

A calibration method for calibrating the position error in the point of interest induced from the stage of the defect inspection tool is achieved by controlling the deflectors directly. The position error in the point of interest is obtained from the design layout database.

Claims

1-24. (canceled)

25. A method comprising: scanning an image of a first point of interest using a first deflector of a charged particle beam tool; and scanning an image of a second point of interest, wherein the second point of interest is greater than a first threshold distance away from the first point of interest.

26. The method of claim 25, further comprising: determining an offset induced from the charged particle beam tool; and calibrating an imaging position of the second point of interest based on the offset.

27. The method of claim 25, wherein the image of the second point of interest is scanned without moving a stage of the charged particle beam tool.

28. The method of claim 25, wherein scanning the image of the second point of interest comprises using the first deflector and a second deflector of the charged particle beam tool.

29. The method of claim 26, further comprising: inspecting the second point of interest with the offset applied to the first deflector of the charged particle beam tool.

30. The method of claim 26, further comprising: identifying a first point of interest alignment pattern within the first point of interest; and determining the offset by comparing the first point of interest to a design layout database.

31. The method of claim 26, wherein the charged particle beam tool is a SORIL ebeam tool.

32. A method of using a charged particle beam tool having a first deflector configured to scan a charged particle beam to a large field of view and a second deflector configured to scan the charged particle beam to a field of view smaller than the large field of view, the method comprising: obtaining a first field of view of a charged particle beam tool; inspecting a first point of interest in the first field of view; and controlling the first deflector and the second deflector to inspect a second point of interest in the first field of view.

33. The method of claim 32, wherein the first field of view is a large field of view.

34. The method of claim 32, further comprising: aligning the first point of interest in the first field of view to a design layout database prior to the first point of interest being inspected, such that a first point of interest base position error of the charged particle tool is obtained, wherein the first point of interest base position error of the charged particle tool is offset by controlling a deflector.

35. The method of claim 34, further comprising: determining whether a distance between the second point of interest and the first point of interest is larger than a threshold; and in response to determining that the distance is larger than the threshold, aligning the second point of interest to a design layout database prior to the second point of interest being inspected, such that a second point of interest base position error of the charged particle beam tool is obtained.

36. The method of claim 35, wherein the second point of interest base position error of the charged particle beam tool is offset by controlling a deflector.

37. The method of claim 34, wherein the first point of interest aligning comprises comparing a pattern in the first point of interest to the design layout database and determining the first point of interest base position error.

38. The method of claim 35, wherein the first point of interest base position error and the second point of interest base position error are induced from stage moving error.

39. A charged particle beam tool comprising: a charged particle source configured to generate a charged particle beam; an objective lens system including a first group of deflectors configured to scan the charged particle beam to a large field of view; and a controller that includes circuitry to control the first group of deflectors to scan a first point of interest within the large field of view, and control other deflectors of the objective lens system to deflect the charged particle beam to another point of interest within the large field of view.

40. The charged particle beam tool of claim 39, wherein the objective lens system further comprises: a second group of deflectors configured to scan the charged particle beam to a field of view smaller than the large field of view, wherein the controller includes circuitry to control the first group of deflectors and the second group of deflectors to deflect the charged particle beam to the other point of interest.

41. The charged particle beam tool of claim 39, wherein the charged particle beam tool is a SORIL ebeam tool.

42. The charged particle beam tool of claim 39, further comprising: a detector configured to detect charged particles from the charged particle beam emanated from a sample.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The present invention will be readily understood by the following detailed description in conjunction with the accompanying drawings, wherein like reference numerals designate like structural elements, and in which:

[0035] FIG. 1 shows a diagram of a wafer comprising several dice;

[0036] FIG. 2 shows a diagram of four dice in the wafer;

[0037] FIG. 3 illustrates a diagram of an ebeam inspection tool;

[0038] FIG. 4 illustrates the SORIL system of the ebeam inspection tool;

[0039] FIG. 5 illustrates the diagram of the dimensions of the die, the field of view insides the die, and the point of interest 70 insides the field of view;

[0040] FIG. 6 shows the diagram of the care area defined by users;

[0041] FIG. 7 illustrates the flowchart of the calibration of one embodiment of the present invention;

[0042] FIG. 8 illustrates the first embodiment of the die base calibration method in accordance with the present invention;

[0043] FIG. 9 illustrates the second embodiment of the die base calibration method in accordance with the present invention;

[0044] FIG. 10 shows the third embodiment of the die base calibration method in accordance with the present invention;

[0045] FIG. 11 shows the first embodiment of the field of view base alignment in accordance with the present invention;

[0046] FIG. 12 illustrates the second embodiment of the field of view base alignment in accordance with the present invention;

[0047] FIG. 13 shows an embodiment of the point of interest base alignment method in accordance with the present invention;

[0048] FIG. 14 illustrates another flowchart of the calibration of another embodiment of the present invention; and

[0049] FIG. 15 shows the diagram of the field of view in the ebeam inspection tool and in the CD SEM or review SEM.

DETAILED DESCRIPTION OF EXAMPLE EMBODIMENTS

[0050] Various example embodiments of the present invention will now be described more fully with reference to the accompanying drawings in which some example embodiments of the invention are shown. In the drawings, the thicknesses of layers and regions may be exaggerated for clarity.

[0051] Detailed illustrative embodiments of the present invention are disclosed herein. However, specific structural and functional details disclosed herein are merely representative for purposes of describing example embodiments of the present invention. This invention, however, may be embodied in many alternate forms and should not be construed as limited to only the embodiments set forth herein.

[0052] Accordingly, while example embodiments of the invention are capable of various modifications and alternative forms, embodiments thereof are shown by way of example in the drawings and will herein be described in detail. It should be understood, however, that there is no intent to limit example embodiments of the invention to the particular forms disclosed, but on the contrary, example embodiments of the invention are to cover all modifications, equivalents, and alternatives falling within the scope of the invention. Like numbers refer to like elements throughout the description of the figures.

[0053] Regarding to the present defect inspection tool, which can be a charged particle beam inspection tool, such as review SEM, CD SEM, or ebeam inspection tool, the significant error of the stage movement still exist due to the different orders between the accuracy and the inspection resolution, even if the wafer level calibration of the stage has been finished. Thus, the calibration of the stage before inspecting defects is essential.

[0054] For a CD SEM, since only few dice will be scanned, the position error of the stage will not severely impact the throughput. For a review SEM, parts of the dice are scanned or analyzed, so the throughput is not influenced seriously as well. Accordingly, the repeated calibration procedures can be suffered in the CD SEM or the review SEM.

[0055] In contrast, the positions of the defects cannot be predicted by the ebeam inspection tool due to they're unknown before inspection, and therefore the repeated calibration procedures in the ebeam inspection tool are inevitable for the accuracy and correctness. In this way, the throughput must be reduced significantly, and the long-term errors of the stage position are still accumulated.

[0056] In the present invention, a field of view (FOV) of the ebeam inspection tool has the dimension of 100 to 300 micrometers.

[0057] In the present invention, a large FOV means the dimension of the ebeam inspection tool is ten times larger than that of the conventional CD SEM or the review SEM.

[0058] In the present invention, a point of interest has the dimension of sub-micrometers.

[0059] In the present invention, said near means that a distance is between two points inside a FOV or several FOVs without calibration.

[0060] In the present invention, local alignment means that the calibration is processed in a point of interest within the FOV.

[0061] In the present invention, said far means that a distance between two FOVs is larger than a predetermined threshold value, such that another stage calibration has to be processed.

[0062] First, please refer to FIG.1 and FIG.2. The FIG. 1 shows a diagram of a wafer comprising several dice, and the FIG. 2 shows a diagram of four dice on the wafer.

[0063] As shown in FIG. 1, several dice 12 on the wafer 10, and a scribe line 13 is formed between every two dice 12. When a back-end process of the dice 12 is finished, the dice 12 need to be cut along the scribe line 13. In order to test the electricity of each die 12, a test circuit is fabricated in each scribe line 13, which can be electrically contacted by a four-point probe method to determine whether the electricity of the die is normal or not. Furthermore, the FIG. 2 shows the enlarged diagram of the dice 12 on the wafer 10, which has patterns. An alignment mark 14 is located in the scribe line 13 and is near to one corner of each die 12, which is applied to calibrate a stage position of a semiconductor processing tool or a defect inspection tool.

[0064] Please refer to the FIG. 3 and the FIG. 4. The FIG. 3 shows a diagram of an ebeam inspection tool, and the FIG. 4 shows a diagram of a swing objective retarding immersion lens (SORIL) system of the ebeam inspection tool.

[0065] As shown in the FIG. 3, the ebeam inspection tool 200 comprises an electron gun 210 and a SORIL system 240. The electron gun 210 comprises an electron source 212, suppressor 214, an anode 216, a set of apertures 218, and a condenser 220. The electron source 212 can be a Schottky emitter. More specifically, the electron source 212 includes a ceramic substrate, two electrodes, a tungsten filament, and a tungsten pin. The two electrodes are fixed in parallel to the ceramic substrate, and the other sides of the two electrodes are respectively connected to two ends of the tungsten filament. The tungsten is slightly bended to form a tip for placing the tungsten pin. Next, a ZrO2 is coated on the surface of the tungsten pin, and is heated to 1300 C. so as to be melted and cover the tungsten pin but uncover the pinpoint of the tungsten pin. The melted ZrO2 can make the work function of the tungsten lowered and decrease the energy barrier of the emitted electron, and thus the electron beam 202 is emitted efficiently. Then, by applying negative electricity to the suppressor 214, the electron beam 202 is suppressed. Accordingly, the electron beam having the large spread angle is suppressed to the primary electron beam 202, and thus the brightness of the electron beam 202 is enhanced.

[0066] By the positive charge of the anode 216, the electron beam 202 can be extracted, and then the Coulomb's compulsive force of the electron beam 202 may be controlled by using the tunable aperture 218 which has different aperture sizes for eliminating the unnecessary electron beam outside of the aperture. In order to condense the electron beam 202, the condenser 220 is applied to the electron beam 202, which also provides magnification. The condenser 220 shown in the FIG. 3 is an electrostatic lens which can condense the electron beam 202. On the other hand, the condenser 220 can be also a magnetic lens.

[0067] The FIG. 4 illustrates the SORIL system 240 which comprises a blanker 248, a set of apertures 242, a detector 244, four sets of deflectors 250, 252, 254, and 256, a pair of coils 262, a yoke 260, a filter 246, and an electrode 270. The electrode 270 is used to retard and deflect the electron beam 202, and farther has electrostatic lens function due to the combination of upper pole piece and sample 300. Besides, the coil 262 and the yoke 260 are configured to the magnetic objective lens.

[0068] The electron beam 202, described above, is generated by heating the electron pin and applying the electric field to anode 216, so that, in order to stabilize the electron beam 202, there must be a long time for heating the electron pin. For a user end, it is surely time consuming and inconvenient. Hence, the blanker 248 is applied to the condensed electron beam 202 for temporally deflecting the electron beam 202 away from the sample rather than turning off it.

[0069] The deflectors 250 and 256 are applied to scan the electron beam 202 to a large field of view, and the deflectors 252 and 254 are used for scanning the electron beam 202 to a small field of view. All the deflectors 250, 252, 254, and 256 can control the scanning direction of the electron beam 202. The deflectors 250, 252, 254, and 256 can be electrostatic deflectors or magnetic deflectors. The opening of the yoke 260 is faced to the sample 300, which immerses the magnetic field into the sample 300. On the other hand, the electrode 270 is placed beneath the opening of the yoke 260, and therefore the sample 300 will not be damaged. In order to correct the chromatic aberration of the electron beam 202, the retarder 270, the sample 300, and the upper pole piece form a lens to eliminate the chromatic aberration of the electron beam 202.

[0070] Besides, when the electron beam 202 bombards into the sample 300, a secondary electron will be emanated from the surface of the sample 300. Next the secondary electron is directed to the detector 244 by the filter 246.

[0071] The FIG. 5 illustrates the diagram of the dimensions of the die, the field of view insides the die, and the point of interest 70 insides the field of view.

[0072] The die comprises several fields of view 50, 52, 54, and each field of view comprises several points of interest 70. The die base calibration method can be classified into three types. The first die base calibration method is to identify the position of an alignment mark 14 located in the scribe line 13 adjacent to the die 12, and to calibrate the stage position according to the position of the alignment mark.

[0073] The second type of the die base calibration method is to identify a specific pattern 20 in the die 12, to image two images 40 and 42 with different positions inside the die 12 by using the specific pattern 20, to calculate the distance between two images 40 and 42, to compare the distance with the design layout pattern to obtain a die base position offset of the stage, and to calibrate the stage error according to the die base position offset of the stage 280.

[0074] Another die base calibration method is to identify the specific pattern 20 inside the die, to image an image 40 or 42 according to the specific pattern 20, to compare the position of the specific pattern 20 in the image 40 or 42 to the design layout database to obtain an die base position offset of the stage 280, and to calibrate the stage error according to the die base position offset of the stage 280.

[0075] The design layout database, for example, can be the graphic database system (GDS) or the open artwork system interchange standard (OASIS), which shows the circuit layout.

[0076] The GDS or OASIS is a data format, which presents different types of the integrated circuit. The difference between the GDS and the OASIS is that the GDS can show 32 bits of the integer coordinates and the OASIS can show the variable integer coordinates which can intend to 64 bits.

[0077] In the FIG. 5, the field of view 60 is the correct one if there's no stage position error. However, the stage error inevitably exists during the stage movement, so the shifted field of view 50 is scanned and the field of view base alignment needs to be processed after the die base calibration is finished. About the field of view base alignment, there're two types of methods. One type of field of view base alignment method is to identify a pattern 30 inside the field of view 50, to image two images 32 and 34 with different positions according to the pattern 30 in the field of view 50, to calculate the distance between two patterns 30 insides the two images 32 and 34, to compare the distance to the design layout database, and to obtain a field of view base position offset of the stage 280.

[0078] The other type of the field of view base alignment method is to identify the position of the pattern 30 inside the field of view 32 or 34, to compare the position of the pattern 30 to the design layout database, and to obtain the field of view base position offset of the stage 280.

[0079] Please refer to the FIG. 6, which shows the diagram of the care area 58 defined by users. When the field of view base alignment is finished, users determine whether the field of view base offset has to be calibrated or not based on the first field of view 50 within the care area 58. If the first field of view 50 is out of the care area 58, the field of view base offset of the stage position should be corrected.

[0080] In order to obtain more precise position of the stage, a point of interest base alignment is further processed after the field of view base alignment is finished. As FIG. 5 shows, the point of interest 80 is correct if there's no position error of the stage. Actually, there's still a position error of the stage during the stage movement, and the real point of interest 70 is scanned by users, so the point of interest base alignment needs to be processed. The point of interest base alignment method is to identify a pattern 701 in the point of interest, to image an image 701 according to the pattern 701, and to compare the position of the pattern 701 to the design layout database to obtain a point of interest base position offset of the stage 280. The dimension of the field of view is hundreds of micrometers, but the dimension of one point of interest inside the field of view is only sub-micrometers, and thus the point of interest in the field of view can be regarded as a point. Since the dimension of the point of interest is too tiny to calibrate the point of interest base position error of the stage by controlling the stage 280, the deflectors 250, 252, 254, and 256 are used to calibrate the point of interest base position error of the stage 280 in the invention. The response time of the deflectors is shorter than that of the stage 280, so the calibration time is significantly shortened. If the deflector is electrostatic type, the deflection is to control the voltage applied to the electrostatic deflector. On the other hand, if the deflector is magnetic type, the deflection is to control the current applied to the magnetic deflector. The deflectors can refer to the deflectors 250, 252, 254, and 256 in the FIG. 3 and FIG. 4.

[0081] Moreover, another method for calibrating the stage error is described as follow, which includes the steps of sequentially processing die base calibration, field of view base calibration, and point of interest base calibration. If the whole process is finished, the calibration is completed. However, if the whole process is not finished, another point of interest is determined whether the distance between it and the original point of interest is larger than a first threshold value or not. If the distance between another point of interest and the original point of interest is not larger than the first threshold value, the point of interest is inspected without processing the local alignment. But, if the distance between the another point of interest and the original point of interest is larger than the first threshold value, that another point of interest is further determined whether or not the distance between the field of view having that another point of interest and the original field of view is larger than a second threshold value defined by users. The field of view base alignment is processed again if the distance between the field of view comprising that another point of interest and the original field of view is larger than the second threshold value. On the contrary, the point of interest base alignment is processed again if the distance between the field of view comprising the uninspected point of interest and the first field of view is not larger than the second threshold and larger than the first threshold value.

[0082] The FIG. 7 illustrates the flowchart of the calibration of the present invention. First, the step SI is processed, which is the die base calibration for calibrating the die base position offset of the stage. The die base calibration includes the die base alignment and the die base correction, and the die base alignment can be referred above.

[0083] Then, the step S2 is processed, which is field of view base alignment of the stage for obtaining the field of view base position offset of the stage. Since the dimension of the field of view with hundreds micrometers is much smaller than that of the die, which is two centimeters, the stage position can be calibrated more precisely by the field of view base alignment. The detailed field of view base alignment will be described later. In step S3, the field of view base position offset of the stage is determined whether it is calibrated or not according to that the field of view base position offset of the stage is in the care area or not. More specifically, if the field of view base position offset of the stage does not make the field of view out of the care area, the step S4 is processed, which means that the point of interest base alignment can be processed without correcting the field of view base position offset of the stage. On the contrary, if the field of view base position offset make the field of view out of the care area, then the step S5 is processed, which implies a step of calibrating the field of view base position offset of the stage. And the step S3 is processed again after the step S5 is finished till the field of view base position offset make the field of view within the care area.

[0084] The step S4 mentioned above is to obtain the point of interest base position offset of the stage. The point of interest base alignment is to capture one image inside the field of view according to a pattern, and to compare the position of pattern in the image to the design layout database to acquire the point of interest base position offset of the stage. The detailed point of interest base alignment will be described later in the FIG. 13.

[0085] Next, the step S6 is processed, which calibrates the point of interest base position offset of the stage by controlling the deflection angle and/or the deflection direction of the deflectors due to the point of interest base position offset of the stage is much smaller than the position accuracy of stage. After the step S6 is finished, the step S7 is processed, which means that the point of interest is inspected.

[0086] When the step S7 is finished, the step S8 is processed. The step S8 is to determine whether the flow is end or not. If the flow is finished, the calibration procedure is finished which is shown in the step S9. If the flow is not finished, the step S10 is processed, which means that another point of interest is determined whether the distance between it and the previous point of interest is larger than a first threshold value or not. If the distance between that another point of interest and the previous point of interest is not larger than the first threshold value, the step S7 is processed again. On the contrary, if the distance between that another point of interest and the previous point of interest is larger than the first threshold value, then the step S11 is processed, which means the distance between the field of view having that another point of interest and the first field of view is determined whether it is larger than the second threshold or not. When the distance between the field of view comprising that another point of interest and the first field of view is larger than the second threshold, the steps S2-S8 is processed again till the whole flow is finished. On the contrary, when the distance between the field of view comprising that another point of interest and the previous point of interest is not larger than the second threshold and larger than the first threshold value, then steps S4-S8 are processed again till the whole flow is finished.

[0087] The definition of near, regarding as the distance between two points of interest is smaller than the first threshold value and depending on the accuracy of the stage, is decided by users. The first threshold, for example, can be about the length of 10 to 20 dimensions of the field of views, which is not limited in the present invention. The definition of far, in a preferred embodiment, can be regarded as the distance between the field of view having that another point of interest and the first field of view is larger than the second threshold value. The second threshold value, for example, can be about the length of the dimensions of 100 fields of view. When the distance between two fields of view is far, the field of view base calibration should be processed again. Moreover, if two points of interest are near, the two points of interest may be located in the same field of view or located in different fields of view.

[0088] The FIG. 8 illustrates the first embodiment of the die base calibration method. First, in the step SI a, the coordinates of the alignment mark in the scribe line near the die is identified. Finally, in the step S1b, the die base position offset of the stage is calibrated according to the position of the alignment mark.

[0089] The FIG. 9 illustrates the second embodiment of the die base calibration method. First, in the step S'la, the pattern is identified to capture two images in the die. Next, in the step S'1b, the two images are compared to obtain a result. In the step S'1c, the die base position offset of the stage is calculated according to the comparison result. Finally, in the step S'1d, the die base position offset of the stage is calibrated. More specifically, the calibration method of the FIG. 9, which is achieved by capturing two images with different positions, calculating the distance between the two images, and comparing the distance with the design layout database to acquire the die base position offset of the stage, is different from that of the FIG. 8 which is achieved by calculating the die base position offset of the stage based on the coordinates of the alignment mark. Besides, since one step of calibration method in the FIG. 9 involves the design layout database, the die base position offset of the stage is more accurate.

[0090] The FIG. 10 shows the third embodiment of the die base calibration method. First, in the step S1a, one image in the die is obtained by identifying the pattern in the die. In the step S1b, the position of the pattern in the image is compared with the design layout database. In the step S1c, the die base position offset of the stage is calculated by the comparison result. Finally, in the step S1d, the die base position offset of the stage is calibrated. Briefly, the calibration method in the FIG. 10 is to compare the pattern in the die to the design layout database and to obtain the die base position offset of the stage. The calibration method in the FIG. 9 is to compare the distance between two images to the design layout database and to obtain the die base position offset of the stage, and the calibration method in the FIG. 8 is to calculate the die base position offset of the stage according to the position of the alignment mark. Accordingly, the calibration method of the FIG. 10 is different from that of the FIG. 8 or FIG. 9, and also, the calibration method in the FIG. 10 is more accurate than that of the FIG. 8 due to the comparison by using the design layout database. On the other hand, because the calibration method in the FIG. 10 needs to capture only one image rather than capturing two images described in the FIG. 9, the calibration method in the FIG. 10 is quietly simple.

[0091] Refer to FIG. 11, which shows the first embodiment of the field of view base alignment. First, in the step S21, the image is captured by identifying the pattern in the field of view. In the step S22, another image in the field of view is captured by the pattern. The step S23 is processed to calculate the distance between the two images. Finally, the step S24 is processed to compare the distance to the design layout database to obtain the field of view base position offset of the stage. The field of view base alignment can be more precise because of the small dimension of the field of view with hundreds of micrometers.

[0092] Refer to the FIG. 12, which illustrates the second embodiment of the field of view base alignment. First, in the step S'21, the image is captured by identifying the pattern in the field of view. Finally, in the step S'22, the field of view base position offset of the stage is acquired by comparing the pattern position in the image to the design layout database. Briefly, the alignment method of the FIG. 12 is to compare the position of the image in the field of view to the design layout database, which is different from that of the FIG. 11 which calculates the distance between two images in the field of view and comparing the distance to the design layout database. Hence, the alignment method in the FIG. 11 captures two images and the alignment method in the FIG. 12 only captures one image. Thus, the alignment method of the FIG. 12 is much simple than that of the FIG. 11. Also, the field of view base alignment is more precise than die base alignment due to the smaller dimension of the field of view.

[0093] Refer to the FIG. 13, which shows the point of interest alignment method. First, in the step S41, the image in the point of interest is identified. Finally, the step S42 is to compare the pattern to the design layout database to obtain the point of interest base position offset of the stage. The point of interest is smaller than the field of view, so the point of interest base alignment is more accurate than the field of view base alignment. Besides, the deflection speed of the deflector is much faster than the speed of the stage movement, and thus the calibration time can be significantly reduced by the point of interest base alignment.

[0094] Next, refer to FIG. 14, illustrating another flowchart of the calibration of the present invention.

[0095] First, the step S'1, which shows the die base calibration of the stage, is processed. The die base calibration method comprises three different types. The first type of the die base calibration is to identify the position of the alignment mark located in the scribe line adjacent to the die, and to calibrate the stage position according to the position of the alignment mark. The second die base calibration method is to calculate the distance between two images in the die, to compare the distance to the design layout database to acquire the die base position offset of the stage, and to calibrate the stage error according to the die base position offset of the stage. The third die base calibration method is to compare the position of the pattern in the image to the design layout database to obtain the die base position offset of the stage, and to calibrate the stage error according to the die base position offset of the stage.

[0096] In the step S'2, illustrating the field of view base calibration for calibrating the field of view base offset of the stage. The field of view base calibration includes the field of view base alignment and correction. The field of view alignment method can be divided into two types. One type of the field of view base alignment sequentially comprises the steps of identifying the pattern in the field of view, capturing the image by the pattern, capturing another image by the same pattern, calculating the distance between the two patterns in the two images, and comparing the distance to the design layout database to obtain the field of view base position offset of the stage. The other type of the field of view base alignment method is to identify the position of the pattern inside the field of view, to compare the position coordinate of the pattern to the design layout database, and to obtain the field of view base position offset of the stage. Since dimension of the field of view is much smaller than that of the die, which is only hundreds micrometers, the field of view calibration can be more precise.

[0097] Next, the step S'3 illustrates the point of interest base alignment. The point of interest alignment is to compare the pattern position in the image obtained in the point of interest to the design layout database to obtain the point of interest base position offset of the stage. The dimension of the point of interest is tinier than that of the field of view, and therefore the accuracy of the point of interest base alignment is enhanced rather than the field of view base alignment or the die base alignment. Then, the step S'4 which shows the step of inspecting the point of interest is processed.

[0098] After inspecting the point of interest, the step S'5 is processed. The step S'5 is to determine whether the flow is finished or not. If the flow is finished, then the calibration procedure is finished which the step S'6 shows. On the other hand, if the flow is not finished, then the step S'7 is processed, which means that another point of interest is determined whether the distance between it and the point of interest is smaller than the first threshold or not. When the distance between that another point of interest and the previous point of interest is smaller than the first threshold value, then the step S'4 is processed again. In this way, the calibration time of the deflector and the stage movement can be saved, and thus the throughput can be highly increased. If that another point of interest is not close to the point of interest, the step S'8 is processed. The step S'8 is to determine whether the distance between the field of view having that another point of interest and the first field of view is larger than the second threshold or not. If the distance between the field of view having that another point of interest and the first field of view is larger than the second threshold, the step S'2 should be processed again, which means the field of view base calibration is processed again. On the contrary, if the distance between the field of view having that another point of interest and the first field of view is smaller than the second threshold value, the step S'3 is processed again, which means the point of interest base alignment is processed again.

[0099] Finally, refer to the FIG. 15, which reveals the diagram of the field of view in the ebeam inspection tool and in the CD SEM/Review SEM. According to the FIG. 15, the beam current of the ebeam inspection tool is thousand times larger than that of the CD SEM or the Review SEM, and the field of view 50 of the ebeam inspection tool is hundreds times wider than the field of view 90 of the CD SEM or the Review SEM also. The CD SEM is applied to measure the critical dimension of the target dice, and the review SEM is used to review the suspected dice. However, the ebeam inspection tool is used to inspect all the dice in the wafer, which has different functions from the CD SEM and the review SEM. Therefore, the error of the stage position can be suffered in the CD SEM or the review SEM, but can't in the ebeam inspection tool. By using the calibration method of the present invention, the error incurred by the stage movement is corrected precisely, the complex and repeated alignment and calibration procedures in the prior art are significantly reduced, and the throughput is increased.

[0100] Furthermore, processing the vector scan by using the point of interest base calibration is much useful. More specifically, the vector scan is to scan a patch image in the point of interest rather than scan the full field of view. The vector scan is achieved by controlling the deflection direction and/or angle of the deflector. Besides, if the two points of interest are close, the stage does not need to be moved so as to increase the inspection throughput.

[0101] By using the point of interest base and the field of view base alignment, the position error induced by the stage can be offset more precisely due to the tiny dimension of the point of interest and the field of view. Besides, if the points of interest are close or the field of view is within the care area, the repeated point of interest base alignment procedures or the repeated field of view base alignment procedures can be neglected, and thus the alignment duration is reduced, and the throughput is accordingly enhanced.

[0102] Although the present invention has been explained in relation to its preferred embodiment, it is to be understood that other modifications and variation can be made without departing the spirit and scope of the invention as hereafter claimed.