JPS60138921A - Inspecting device of pattern shape - Google Patents

Abstract

PURPOSE:To obtain a superior pattern shape inspecting device by a method wherein the two dimentional deformations of pattern areas set by dividing into blocks of the plural number are corrected in regard to every respective block, and correction of error is enabled to be executed exactly at any pattern deformation. CONSTITUTION:A pattern shape inspecting device 1 has a reticle 2 constructed of an image sensor, a solid-state image pickup element for example, and to check the shape, and a picture input part 3 to output the video signal thereof. Output of the input part 3 is supplied to a shape comparing judgement part 5 after quantized (converted into a binary value) according to a quantization circuit 4. Design pattern data of integrated circuits compiled in a data file 6 are converted previously into format data according to a large size computer 7, and moreover the data thereof are converted into dots according to a dot conversion part 8 to be supplied to the judgement part 5. The judgement part 5 has the function to judge the shape of the reticle 2 according to output of the circuit 4, and to detect the abnormal pattern of the reticle 2 by comparing the shape of the reticle thereof and the design pattern of the integrated circuit supplied from the dot conversion part 8.

Description

DETAILED DESCRIPTION OF THE INVENTION The inventions 1 to 2 relate to pattern forming techniques for semiconductor integrated circuits, photolithography techniques, and the like, and particularly to apparatuses for inspecting pattern shapes formed by these techniques.

[Technical background of the invention and its problems]

In general, photomasks such as reticles used for pattern drawing of integrated circuits (hereinafter, a reticle, which is an example of a photomask), have image performance sufficient to satisfy the ultra-precision required for pattern formation KW of semiconductor integrated circuits. However, a portion of the drawn surface may differ from the set data due to various factors.

For example, for the design shown in Figure 1, no pattern was formed, Figure 2 (a).
), some defects, 1 adhesion, 2 partial enlargement or reduction, and deviations of I-lean occur.Therefore, it is difficult to determine whether the formed pattern was drawn faithfully to the design pattern data. Conventionally, the reticle was
The shape of the reticle can be determined by placing it on a stage, imaging it with an ITV camera, converting the output into binarization, and comparing it with the format data of the integrated circuit design pattern that has been dot-converted in advance. We are conducting an inspection.

However, if the pattern RP designed as shown in Figure 3 has a two-dimensional deformation as shown in the drawn nine-four-turn KPIC diagram, these four-turn RP,
cannot be simply compared. Therefore, the deformation of the entire pattern area is inspected by measuring the deformation error in advance before inspecting the pattern shape, and correcting the design pattern data or the pattern data to be inspected based on this error. That is, as shown in FIG. 4, reference marks M, , Ml, M, are drawn outside the turn area KP to be inspected when drawing the turn area KP.
Set x, y, and θ using the reference mark Ml as the center coordinate, compare it with the design pattern data, and calculate the length error Δ′″x in the X direction,
Calculate the length error ΔY in the y direction and the squareness difference Δθ. When inspecting the pattern shape of the main turn area KP, two-dimensional deformation is corrected based on the error data ΔX, ΔY, and Δθ, and the pattern shape is inspected.

However, in the pattern shape inspection method described above, as shown in FIGS. 5 (JL) to (C),
If the shape is transformed into a parallelogram due to directional expansion/contraction or reduction of perpendicularity, accurate correction can be made, but as shown in Figure 6 (&
) as shown in Figure 6(b), which has been deformed into a trapezoid.
As shown in (0), it is difficult to make accurate corrections when these deformations are superimposed.

The barrel-shaped and coma-shaped distortions shown in FIGS. 6(b) and 6(e) above are likely to occur due to aberrations of the optical lens when the image is transferred through an optical lens and is transferred through a 9/4 turn.

The above-mentioned deformation during pattern formation cannot be avoided, even if the error is large or small, and is a problem with semiconductor integrated circuit technology.
Since the circuit can operate satisfactorily even if the pattern is slightly deformed two-dimensionally, errors can be corrected even when the pattern is deformed as shown in Figures 6(a) to 9(a). A turn-shaped inspection device is desired.

[Purpose of the invention]

This invention was made in view of the above circumstances,
The main objective is to provide an excellent pattern shape inspection device that can accurately correct errors in any pattern deformation.

[Summary of the invention]

That is, in this invention, in order to achieve the above-mentioned corner of the eye, a plurality of reference marks are provided. an image sensor that images the pattern to be inspected formed based on the design pattern; a quantization circuit that quantizes the output of the image sensor at a predetermined level; a correction means for correcting the plurality of probe shapes for each block based on the reference=-2 formed at -7 and the output of the upper dot conversion section, and the dot conversion section based on the output of the correction means. A pattern shape inspection device is constituted by a comparison/judgment section that corrects the output of the quantization circuit and compares the corrected output with the output of the quantization circuit to inspect the pattern shape.

[Embodiments of the invention]

An embodiment of the present invention will be described below with reference to the drawings.

FIG. 7 is a block diagram showing a schematic configuration of the pattern shape inspection device. This pattern shape inspection apparatus J includes an image input section 3 that maps a reticle 2 whose shape is to be checked onto an image sensor, for example, a solid-state image sensor, using an optical system and outputs a video signal thereof. Image input section 3
The output of K is quantized (binarized) by the quantization circuit 4 and then supplied to the shape comparison/judgment section 6.

The shape comparison/judgment section 5 is supplied with a data file 6 which is converted into format data by a large computer 7 and then converted into dots by a dot conversion section 8. The shape comparison/judgment section 5 determines the shape of the reticle 2 based on the output of the quantization circuit 4, and compares this with the design pattern of the integrated circuit supplied from the dot conversion section 8 to determine whether the reticle 2 has an abnormality A? It has a pattern detection function. Reticle 2 detected by shape comparison/judgment section 5
The abnormal pattern is displayed on the abnormal/four-tap display section 9.

Here, the reticle 2 is arranged as shown in FIG.
It is placed on stage 10IIC. X-Y stage 1
A light source 11 and a condensing lens 12 are provided below the light source 1), and the light emitted from the light source 1) passes through a reticle 2 and is focused by a magnifying lens 14 whose focus is adjusted by an automatic focus adjustment mechanism 13. After being exposed to light, for example, a CCD (Chargs Coupled I) provided in the image input section 3
)evice )Solid-state image sensor Ja such as line sensor
After the K incidence, an image at a predetermined position in the lateral direction (hereinafter referred to as the X coordinate) of the reticle 2 is mapped.

X-Y mapped by the optical system composed of the above-mentioned light source 11 etc.
The position of the reticle 2 on the stage 10 is determined by measuring the amount of movement of the X-Y stage 10 in the X-coordinate direction and the Y-coordinate direction by a radar length measurement control section 1'6 that operates in response to a command from the central processing unit 150. It is now detected by the user.

As shown in FIG. 9, this laser length measurement control unit 16 oscillates a radar from a radar transducer 16&, refracts the radar with a beam pender 16b, and connects the beam splitter 16c, beam splitter, beam splitter 16c, X at 16c
The beams are divided into two beams in the X- and Y-directions and sent through a plane mirror interferometer 16d.
After being guided to the Y stage 100 grain mirror 16 and reflected, this reflected beam is transferred to the plane mira ferometer 1.
Interfering with 6d, x-Y each.

is supplied to the receiver 161, and the optical system maps it to the receiver 161. The X-coordinate and Y-coordinate of the X-Y stage 10 in the state where the The positional deviation of the read value in the X and Y coordinates is calculated based on the predetermined X coordinate value and Y coordinate value to be mapped by the optical system, which are supplied from the optical system to the laser length measurement control unit 16 . The amount of deviation between the X-Y reading positions is transmitted to the X-Y stage control 17, and as shown in FIG.
X and Y drive motors 1 via cm 11 d
7e, 111, and the central processing unit 15
With the feeding information of the X-Y stage 10 supplied from the interface 15h of I'm busy doing that. Furthermore, X
-Y The position of the stage 10 is controlled in accordance with the manual feed information Z9.

In this way, the two-dimensional information (mapping) of the reticle 20
As shown in Figure 1, when the x-y stage 10 is moved in the Y direction at a predetermined speed, the solid-state image sensor 3 is moved in synchronization with this.
A is scanned to map the area of the reticle 20 corresponding to the predetermined pit width of the solid-state image sensor 3a, and the same operation is performed with
This is done by moving the x-y stage 10 in such a way that the scan width (4) of the solid-state image pickup device 3a becomes slightly heavier in the direction.

For example, as the solid-state image sensor 3a, a 1728-bit CC
When mapping is performed using a D-line sensor, this CC
If the correction width of the D line sensor is ±50 bits, and the amount of deviation in the X direction during mapping is zero, the scan width is 12th.
As shown in Figure (a) K, it is 50 to 1678 bits.

If the deviation amount is 29 bits in the X direction, C
The corrected scan width of the CD line sensor is 30 to 1
It is 658 bits. Furthermore, as for the amount of deviation in the Y direction, the COD line is determined based on this amount of deviation, as shown in FIG. 2(b).

Furthermore, the amount of deviation of the reading position in the be done. Based on this signal, the sensor control section 3b performs scan control of the solid-state image sensor 3a, as shown in FIG. 13, to correct fluctuation errors caused by the movement of the x-y stage JO. The video signal of the reticle 2 imaged by the solid-state image sensor JaK, which is scan-controlled by the sensor control unit 3b, is supplied from the image input unit 3 to the quantization circuit 4, where it is quantized, and then sent to the serial-parallel conversion transfer unit 20.
The buffer memory 21Vc is supplied through the buffer memory 21Vc.

In the serial-parallel conversion transfer unit 20, the video signal of the quantum-enhanced mapping is input to the system register, and the X-Y
A projection of the reading line in the X direction of the reticle 2 placed on the stage 101fC is formed. The projection signal of this read line is
! As shown in FIG. 14, the image is written into the image buffer memory 21. The projection signal written in the image manual buffer memory 21 can be read out by the central processing unit 15 and displayed on the pattern display section 23 via the display control section 22.

On the other hand, the design pattern data dot-converted by the dot conversion section 8 is stored in a two-block design data buffer memory 24m, in order to accommodate the continuous readout operation of the image input section 3.
24b. The contents written in the design data buffer memories 24m and 24b can be read by the central processing unit 15 and displayed on the pattern display section 26 via the 1-display control section 25.

Also, a design data buffer memory 24a.

The design pattern data written in 24b is supplied to one output end of the exclusive OR circuit 28 via a shift register provided in the read parallel/serial converter 27. The other input terminal VC of the exclusive-or circuit 28 receives the video signal of the mapping of the reticle 2 quantized by the quantization circuit 4 .

Then, this exact-knob-or circuit 28 calculates the exclusive-or of the design pattern data and the reticle 2p mapping video signal, and the calculated value is supplied to the abnormal pattern detection section 29 to detect abnormal locations in the shape of the reticle 2. . The abnormality/mother turn of the m-reticle 2 detected by the abnormal pattern detection unit 29 is determined by the central calculation i position J5.
The pattern is read out by the IIC and displayed on the pattern display section 2:6.

The operation of the quantization circuit 4, the operation of the radar length measurement control unit 16, the operation of the X-Y stage control section 17, the operation of the serial/parallel conversion transfer section 20, and the operation of the read exclude or circuit 28 are as follows. , is performed in synchronization with the operation of the solid-state image sensor 3a.

In the above configuration, the error that occurs when imaging the reticle 2 placed on the It is calculated by detecting the positions of reference marks M11 to M55 arranged in five rows and five columns in the x and Y directions at intervals of XI + Yl. For example, the error at the point Z of -≠P in Fig. 15 is:
It is determined by the following equation (1).

The neck is rounded off by the deformation error at the midpoint Q of the base thread mark.

Note that the number of arrays of reference marks and their intervals are set depending on the complexity of pattern deformation, tolerance for error, etc.

The error obtained in this way (Tsukutan deformation error...
Lean rotation error) to the correction value R of the X component at point Q
xQ and Y-component correction value RY.

is expressed by the following formula.

RXQ = -(Sx+ΔXQ) RYQ ==-ΔYQ However, point 0 is the design coordinate value of the reference mark 22. Point P is
The measured coordinate value Sx is the amount of fluctuation when the X-Y stage 109 moves in the Y direction, and is the sum of the deviation and deviation with respect to the x-coordinate set value 9. (See Figure 13) The correction value RxQ of the X component and the correction value RYQ of the Y component are as follows.
When the reticle 2 is placed on the X-Y stage 10 and the position of the X-Y stage 10 is controlled by the X-Y stage control section 17 and an image is taken at an arbitrary position, the radar length measurement control section 16 As shown in FIG. 16(&) and FIG. 16(b) K, this can be obtained by calculating with the correction control 1g.

Both correction values RXQ * RYQ thus obtained are transferred from the correction control section 18 to the sensor control section 3b of the image input section 3.
As shown in FIGS. 12 (Jl) and (b), as for the correction value RXQ of the It will be automatically corrected by shifting. Further, the correction value RYQ of the Y component is automatically corrected by vertically shifting the yang position of the solid-state image sensor 3a.

The mapping of the reticle 2 placed on the x-y stage 10 in this manner is binarized as an extremely accurate image and written into the image manual buffer memo 921. Then, if necessary, the central processing unit 15 issues a command to display the information on the Shibata-turn display section 23.

On the other hand, the set carp that is recorded in data file 6, '
? ;'l'-n data is dot-converted and written into the design data buffer memories 24a and 24bK.

Therefore, this design data buffer memory 34 a +
24 b and the written information (values of the mapped image and design data) in the imager memory 2 are sent to the central processing unit 1.
5, the exclusive OR circuit 28 performs both calculations, and the abnormal pattern detection section 29 can detect an abnormality/f turn of the reticle 2. This abnormality/mother turn is displayed on the pattern display section 26 and can be confirmed visually.

〔Effect of the invention〕

As explained above, according to the present invention, errors can be corrected accurately in any four-turn deformation.

This provides an excellent pattern shape inspection device that can be used to inspect patterns.

[Brief explanation of the drawing]

Figures 1 and 2 are a first view and a third view of a Noreen shape showing an example of the designed and formed Noreen shape, respectively.
The figure shows the two-dimensional deformation of the pattern area. and FIG. 6 are diagrams for explaining the deformed state of the pattern area, FIG. 7 is a block diagram showing a schematic configuration of a pattern shape inspection apparatus according to an embodiment of the present invention, and FIG. 8 is a diagram showing the same embodiment. 9 is a block diagram showing the configuration of the laser length measurement control section in the same embodiment. FIG. 10 is a block diagram showing the configuration of the X-Y stage control section in the same embodiment. , FIG. 11 is a diagram for explaining the scanning direction of the solid-state image sensor in the same embodiment,
Figure 1.12 is a diagram for explaining the correction of errors in the X component and Y component in the same embodiment, and Figure 13 is a diagram for explaining the fluctuation error due to the movement of the X-Y stage in the Y direction in the same embodiment. To explain the correction method? 14 is a block diagram showing the configuration of a circuit for comparing an image pickup signal and design pattern data in the same embodiment, and FIG. 15 is a diagram for explaining a pattern distortion correction method in the same embodiment. Fig. 16 is a diagram for explaining the method of calculating the correction values of the X and Y components of the reticle in the same embodiment.1111-M55...Reference mark, 8...Dot converter, 3a・・Solid-state image sensor (image sensor), 4
. . . quantization circuit, 5 . . . shape comparison and determination section. Applicant's agent Patent attorney Takehiko Suzue Figure 2 Figure 3

Claims (1)

[Claims] A dot conversion section that converts format data of a design pattern including a plurality of reference marks into dots, and the above design i4.
? an image sensor that captures an image of a nolin to be inspected formed based on the turn; Amount of output of this image sensor at a predetermined level:
f□Fui. □1.8. hot water. AI->Kl□ The two-dimensional deformation of Lean-Rear, which is set to be divided into multiple groups based on the formed reference mark and one output from the dot conversion section, is performed for each group. A correction means for correcting, correcting the output of the dot conversion section based on the output of this correction means, and comparing the correction output of 1 and the □ output of the quantization circuit to inspect the lean shape. A main turn shape inspection device characterized by comprising: a determination section; 1