JPS62157760A - Precision processing machine - Google Patents

Abstract

PURPOSE:To improve machining precision, by a method wherein the position of a member to be detected is corrected and controlled from a processing result right before storage each time, subsequent to second machining on the same member to be machined, machining is effected. CONSTITUTION:A deviation detecting means 29 guides light bundle from a light source 30 to the pattern of a member 26 to be detected through a half mirror 31, and is integrally formed so that the microphoto image of a pattern 1 is projected on a line sensor 5. An output from the line sensor 5 is guided to a sample holding circuit 32, is further converted into a digital mode through an A/D converter 33 to guide the results to a memory 34, and the results are orderly stored in given address positions by means of an address signal from a computing and control circuit 35. A machining result right before it is stored in the memory 34 once is inputted to the computing and control circuit 35 for computation, and this computation corrects and controls the position of the member 26 to be detected, resulting in the possibility to effect extremely high-precision machining.

Description

[Detailed Description of the Invention] [Industrial Application Field] This invention places a workpiece on a mounting table, and controls the movement of this mounting table to move the processing part of the workpiece outside the mounting table. A processing machine that performs processing using a fixed processing tool can always perform highly accurate processing without being affected by the relative position error between the workpiece and the processing tool, which occurs due to a 5W increase due to continuous processing. It is related to a processing machine that can be used.

[Conventional technology]

An object to be processed, such as a semiconductor wafer, is
In a cutting machine that is fixed to a mounting table that can be moved separately in the Y-axis direction and the θ direction, and cuts into rectangular chips, for example,
The mounting table is moved in the
Alternatively, it is necessary to position the workpiece by highly accurate driving in the Z-axis direction.

Conventional methods for this purpose include, for example, Japanese Patent Application Laid-Open No. 59-1766.
As described in Publication No. 10, cutting is performed by capturing a two-dimensional image of a region including the position to be cut on the workpiece, and differentiating the image output with respect to the X and Y axis directions of the two-dimensional image. Positioning is performed by detecting the presence of a street formed at the desired location.

Furthermore, in Japanese Patent Publication No. 59-43820, in order to improve the position matching accuracy of a first object having a first identification pattern and a second object having a second identification pattern, each identification pattern is A light image formed optically in the X-axis direction and the Y-axis direction is projected onto each image sensor that receives light, and the image signals in the X-axis direction and the Y-axis direction obtained thereby are projected.
Based on the video signal in the axial direction, the relative deviations of the first object and the second object in the X-axis direction and the Y-axis direction are determined, and the two objects are moved relative to each other so that the deviations are eliminated. A position alignment device is shown for alignment purposes.

[Problem that the invention seeks to solve]

According to the above-mentioned conventional technology, it is possible to align a predetermined position of a workpiece such as a semiconductor wafer with a blade of a processing tool with relatively high precision. However, since all of the methods described in the above publications process video signals obtained by photoelectrically converting the workpiece into a two-dimensional image, it is difficult to compare the signal processing to obtain deviation information. Not only does it take a long time, but since a two-dimensional image is detected, the resolution is limited, making it difficult to obtain sufficient position detection accuracy for the workpiece.

Therefore, in a processing machine equipped with such a positioning device based on the conventional technology, the positioning accuracy cannot be said to be sufficient.
Furthermore, the time required to calculate the amount of deviation of the workpiece relative to the processing tool is problematic because it cannot be done in a short time because two-dimensional image signals are processed. Moreover, it has the disadvantage that no correction effect can be expected for alignment errors caused by temperature rise, which is a problem during long-term continuous machining.

An object of the present invention is to provide a processing machine that eliminates the problems and drawbacks of conventional processing machines as described above.

[Means for solving problems]

The precision processing machine according to the present invention is characterized in that the workpiece is placed on a mounting table configured to be able to displace the processing tool in at least one of the X-Y coordinate axes, and the mounting table is displaceable. In a processing machine that positions the relative position between a processing tool provided at a specific position and the object to be processed and performs processing using the processing tool, the shape allows position information corresponding to the amount of deviation at least in the axial direction to be obtained. A deflection detection device configured by arranging a line sensor in such a relationship that a one-dimensional image signal of the pattern in at least the axial direction is obtained when a test object having a deflection detection pattern formed in the above is placed on the mounting table. and positional information of the pattern based on the specific position and the pattern based on the position of the processed end after processing by processing and calculating the one-dimensional image signal obtained by the deviation detection means. The position information of the pattern position information of the former is calculated by referring to the known standard position information corresponding to the pattern position information of the former, and the subsequent machining of the workpiece is performed by referring to the pattern position information of the latter. The relative position error between the processing tool and the workpiece due to the displacement of the processing tool is calculated by comparing the former pattern position information obtained for the processing position prior to the processing, and the deviation amount and relative It consists of an arithmetic and control circuit configured to obtain a control signal for positional errors, and drives the mounting table with a control signal corresponding to the amount of deviation each time the same workpiece is processed from the second time onward. The positioning position according to the amount of deviation is corrected by controlling and feeding back a control signal corresponding to the relative position error obtained from the immediately previous processed position to the mounting table drive system. It is a precision processing machine.

[Effect]

FIG. 1(A) shows that the workpiece (
Hereinafter, for convenience, this will be referred to as the "tested member." ), a basic form 1 of a deviation detection pattern (hereinafter referred to as "pattern") formed in advance, and a line of deviation detection means fixedly provided at a predetermined position outside the mounting table in response to the displacement of the mounting table. The relationship with sensor 5 is shown.

In the same figure, position 2 of the reference X-axis indicated by the orthogonal coordinate axes of the test member on which the pattern 1 is formed at a predetermined position is a fixed specific position outside the mounting table, and the broken line of the test member is set at a fixed specific position outside the mounting table. Assuming that the positioning line 3 shown as a straight line is aligned, the first stage of detecting the amount of deviation of the member to be inspected at a specific position in the device of the present invention will be described.

The pattern 1 has a shape in which the width in the Y-axis direction changes axially symmetrically about the central axis 4 in the X-axis direction. With respect to this optical image of pattern 1, K
The detection signal from the line sensor 5 drives the line sensor 5 when the test member is placed on the mounting table so that the line sensor 5 such as a CCD has the positional relationship as shown in the figure. It is taken out at the cycle of clock pulse C. If this is shown in analog form, it can be shown in an output waveform diagram as shown in FIG. 1(B).

This output waveform corresponds to a straight line figure in the Y-axis direction at the position of the line sensor 5 of pattern 1, and includes the pattern width at that position and position information n in the Y-axis direction. That is, the positional information nA of the central axis 4 in the Y-axis direction corresponding to the distance from the specific position 2 set on the reference X-axis to the central axis 4 of the pattern 1 can be easily obtained from the following equation (2). I can do it.

ny = ((11++12)/2...
・・・・・・・・・・・・(1) Note that n and n2 are
This is a count value obtained by counting clock pulses from the rising edge to the falling edge of pattern 1 with specific position 2 as the origin.

Also, by calculating the following equation (2) as necessary,
It is also possible to easily calculate the positional information n in which the Y-axis of the X-axis direction where the line sensor 5 intersects the pattern 1 is the specific position 6 of the second chapter.

nx=n2-nl・・・・・・・・・・・・・・・
・・・・・・・・・・・・・・・(2) That is, the pattern 1 shown in FIG. 1(A) is applied to a specific position outside the mounting table.
If the position in the Y-axis direction, which is determined based on the alignment position 6 of the test member indicated by the straight dashed line in , exactly matches the central axis 4 of pattern 1, then the known Reference clock pulse number n. By determining the difference between the calculation result value ny of the equation (1) and the above-mentioned equation (1), the amount of deviation ( no-ny) is obtained.

Furthermore, if necessary, it is also possible to simultaneously calculate the amount of deviation in the X-axis direction with respect to the base MY-axis 6 in the same manner.

Therefore, if the mounting table is controlled to move in the Y-axis direction and/or X-axis direction according to the amount of deviation so that the amount of deviation becomes zero, the position of the member to be inspected should be aligned. position 3
coincides with the specific position 2 outside the mounting table, and the member to be inspected can be positioned at a predetermined correct position.

In addition, each count value of the clock pulse corresponding to the position information based on the specific position of each edge of pattern 1 is 01%.
As a specific means for obtaining n2, for example, two resettable pulse counters are operated simultaneously and synchronously by the clock pulse for driving the line sensor, and one of them is operated as shown in the waveform diagram of FIG. 1(B). The number n of clock pulses in the period up to the first timing t of the output waveform of the line sensor shown in FIG. Configure it to do so.

The two pulse counters configured in this way are shown in Figure 1 (
The line image in the Y-axis direction at a predetermined known and accurate distance from the specific position 2 of A) to the central axis 4 of the pattern 1 is reset at a timing preceded by the number of lock pulses necessary for the line sensor 5 to sequentially take out line images. If operated as follows, each count value nl at timing t and t2
, n2 corresponds to the position information in the Y-axis direction of the upper and lower edges of pattern 1 shown in FIG.
Equation 1) or (2) can be calculated.

In this way, the device of the present invention detects the deviation information of the member to be inspected with respect to a specific position from the one-dimensional image information obtained by the line sensor for the pattern having the above-mentioned shape. Since the amount of deviation can be calculated using the formula, the amount of deviation can be obtained extremely quickly.

In other words, conventionally, the positioning position is detected using two-dimensional information, so if an image sensor with a pixel count of 512 x 512 is used, for example, all of the pixel information is taken into a processing unit and processed. must,
This will require a considerable amount of calculation processing time. In contrast, in the present invention, positioning is performed using one-dimensional information obtained by a line sensor, and it is possible to use an existing line sensor with a pixel count of 2048 to 4096, which: Processing time can be significantly reduced compared to C.

Furthermore, in conventional systems that use secondary information, the total number of pixels is larger compared to systems that use line sensors as in this invention, but when considering line image information, existing line sensors have a large number of pixels. 2048 or 4096.Furthermore, the line sensor has a larger number of pixels, so the resolution is correspondingly higher.Therefore, in this invention, in addition to shortening the processing time, This will further improve alignment accuracy.

After alignment is performed as described above, the aligned location is processed, for example, cut, using a processing tool. However, when continuous processing is performed for a long time, for example, if the processing tool is a rotary cutting wheel, etc. Continuous high-speed rotation causes friction with the test member, resulting in high temperatures, which often causes errors in the relative position with the workpiece. This makes precision machining difficult.

In order to eliminate the influence of such obstacles, the present invention makes the second and subsequent alignments of the same inspected member to the processing tool after alignment and processing as described above to the first alignment. Using the based machining position as a reference position, the pattern formed at a known predetermined position from the workpiece position to be machined from the second time onward is detected again by the deviation detecting means.

FIG. 1(C) shows an example of the waveform of the detection signal. If the cutting is done at the correct position, the processing position will be n from the center of the detection pattern in FIG. 1(A). A cutting signal waveform should be obtained at a distant position, but the position error is Δ
If there is a deviation of y degrees Celsius, the cutting position when there is an error is the first one.
From the center of the detection pattern in figure (A)
It is detected at a position separated by o1Δy. That is, by measuring the distance n 0 / from the center of the detection pattern in FIG. 1(A), the position error Δy can be obtained from the formula %.

The position error information Δy obtained in this manner is fed back to the mounting table drive system.

That is, the first machining position of the test member is measured at a predetermined position with respect to the heavy work position, and the amount of deviation with respect to the processing tool is determined. Processing is performed after positioning by displacing the members. The alignment of each machining position of the above-mentioned member to be inspected after the second machining is performed in the same manner as the first machining, and the pattern formed at a predetermined position from the actual machining position immediately before machining is performed. Find the position information of the pattern corresponding to the distance. This is compared with the pattern position information at the time of alignment.

The error information obtained in this way corresponds to the relative position error between the position of the processing tool and the position of the tested member, and is due to a change in the predetermined position of the processing tool due to some reason, such as a temperature increase due to continuous processing. This occurs due to misalignment, etc.

Since the processing machine of the present invention is configured to feed back the error information obtained as described above to the control system of the mounting table, the relative position error as described above is corrected. Coupled with this function, the inspected member can be processed with high precision.

Note that the positional accuracy in the Y-axis direction corresponding to the cutting position shown in FIG. 1(C) is n3. n3' uses the same means as explained earlier, for example, starts counting clock pulses by setting the timing of the pulse at the cutting position detection position (a known distance N from the pattern detection position) to 0 bits, and starts counting the clock pulses. The count value in the section can be easily determined by using a pulse counter device configured to determine n3s n3'.

〔Example〕

FIG. 2 is a configuration diagram showing an example of an embodiment of the device of the present invention.

In the same figure, reference numeral 21 denotes a mounting table for a member to be tested, which includes a rotary table 22, a Y-axis movable table 26 for displacing this rotary table 22 in the Y-axis direction, and a Y-axis movable table 26 that is placed on the rotary table 22. and an X-axis movable table 24 for displacement in the X-axis direction, and each table 22, 23, 24 is configured to be driven individually by a drive signal from a drive circuit 25.

The member to be inspected 26 is placed on the rotary table 22 on the mounting table 21.
0 Place and hold in the specified position.

Reference numeral 27 denotes a processing tool drive unit set at a specific position outside the mounting table 21. For example, in the case of a cutting machine, the rotation and vertical direction of a rotary cutting wheel 28 etc. are controlled by each drive unit guided from the drive circuit 25. , driven by a drive signal. The position of this cutting whetstone 28 corresponds to the above-mentioned specific position, and is configured to be movably controlled in the Z direction by a cutting control signal from the drive circuit 25, so that the member to be inspected 26 can be cut in the X-axis direction.

The specific position is set at a correct position separated by a known distance from a deviation detection means 29, the details of which will be described later, and the deviation detection means 29 is also placed on the rotary table 22. The pattern formed at a predetermined position on the member to be inspected is correctly set at a position where it can be reliably detected.

In this example, it is assumed that the pattern provided on the test member 26 is formed in a form that can be detected optically, for example,
The deviation detection means 29 has a configuration that deals with this.

That is, the deviation detection means 29 is integrally configured to guide the light beam from the light source 30 to the pattern of the test member 26 via the half mirror 61 and to project a microscopic light image of the pattern 1 onto the line sensor 5. has been done. The line sensor 5 is arranged in the relationship previously explained with reference to FIG.
It is fixedly provided at a position outside the mounting table 21 so that it is always projected onto the line sensor 5 when it is held above.

Therefore, for example, if the pattern has the basic shape shown in FIG. 1(A), the detection signal obtained by guiding the output of the line sensor 5 to the sample and hold circuit 32 is as shown in FIG. 1(B). It will be like this.

The output of this sample hold circuit 62 is digitized via the A/D converter 33 and guided to the memory 34 for calculation and processing.
An address signal from the control circuit 65 causes the data to be sequentially stored at predetermined address positions corresponding to the positions of the X-Y coordinate axes.

The arithmetic/control circuit 65 stores the stored address values n1% and n2 necessary for the calculations of equations (1) and (2) in the memory 34.
, and perform calculations on each of these expressions,
Each calculation result value nY, nz is converted to each known reference value nY', nX
’ deviation amount in the Y-axis direction (nY’-ny)
, (nx/-nx) and outputs a control signal for correcting these deviation amounts. Furthermore, when a member to be inspected 26 on which a pattern of a shape to be described later is formed is placed on the rotary table 22, the amount of rotation of the member to be inspected 260 can be calculated and a correction control signal thereof can also be output.

The drive circuit 25 drives the Y-axis movable table 23, the X-axis movable table 24, and the rotary table 22 of the mounting table 21 by the direction and distance corresponding to the respective displacement amounts using these control signals. However, when the amount of deviation becomes zero, each control signal also becomes zero, and each part 22, 23, 24 constituting the mounting table 21 stops at a controlled position.

In this state, the specific position of the test member on the mounting table 21, that is, the position to be aligned with the position of the cutting wheel 28, is determined at the specific position as explained in FIGS. 1(A) and (B). This means that it is displaced by the movement of the mounting table 21 so that it coincides with .

FIG. 3 shows a pattern forming position on a member to be inspected, taking a semiconductor wafer 41 as an example.

Dotted lines 42-1 to 42-(n-1) indicate positions in the Y-axis direction to be cut, and 43-1 to 43-m indicate, for example, IC chips formed on the wafer 41.

In this example, the position of the blade edge (28 in FIG. 2) of the cutting wheel fixedly arranged so that the blade edge coincides with the cutting position is defined as a specific position, and the cutting position 42 aligned in the Y-axis direction of the wafer 41 is at this position. -1 to 42-(n-1) are aligned sequentially. For this purpose, each cutting position 42-1 to 42-(
n-1) to exactly n. The curved deviation detection patterns 1-1 and 1-1 illustrated in FIG.
1 to 1-n and 2n form a pair on both sides, with the central axis 4 parallel to the line of cutting positions 42 1 to 42 (n-1) being a common central axis, and optically It is formed in a readable form.

By using the semiconductor wafer 41 on which the patterns 1-1.1 to 1-n.2n are formed as a test member and carrying out the test according to the configuration of the embodiment shown in FIG. Using pattern 1-1.1, one of the pair of patterns -1.1 and 1-1.2, the amount of deviation in the Y-axis direction and the X-axis direction can be determined as explained earlier. can. Also, each pattern forming a pair, for example 1-1
・By determining the amount of deviation in the Y-axis direction or There is an advantage that the deflection angle can also be easily detected.

In the configuration of the embodiment shown in FIG. 2, the calculation/control circuit 35 calculates such a rotational deviation angle θ, supplies a control signal corresponding to the deviation angle θ to the drive circuit 25, and thereby The rotation angle of the rotary table 22 of the mounting table 21 is controlled to correct the deviation angle θ, and the cutting whetstone 28
The drive of the member to be tested in the Z-axis direction can also be controlled by a command signal from the arithmetic/control circuit 65.

In the first processing of the member to be inspected 26, the member to be inspected 26 is positioned as described above, and the processing tool 28 is driven in the Z direction to perform processing, for example cutting.

Next, the arithmetic/control circuit 35 sends a control signal to the drive circuit 25 to move the mounting table 21 a small distance in the Y-axis direction according to a predetermined program, so that the processed position of the test member 26 and the pattern 1 are detected by the deviation detection means. It is moved to an extent that can be detected by the line sensor 5. In this state, obtain the output of the line sensor 5 again (Fig. 1 (C)) and count the position information n 3/ of the edge portion with the pulse timing of the cutting position detection position as O bit, as in the previous case. and from there N
The cutting position n0' is obtained by detecting a pattern by moving 0, counting n, . The calculation result value n0/ obtained in this way is based on the cutting tool position and pattern corresponding to the dimensions up to a predetermined position of the pattern formed on the cut piece obtained by actually cutting the member to be inspected. It is position information, and the error value Δy from the correct cutting position
can be obtained.

Next, in order to process the next workpiece position of the part to be inspected, the line processor 5 displaces the mounting table in the Y-axis direction and forms the pattern correctly at a predetermined position relative to the next workpiece position. The part to be inspected is transferred to the detected position, aligned as described above, and pattern position information ny whose origin is the position of the processing tool 26 calculated for the alignment, as well as the pattern position information ny processed immediately before, for example, cutting. The error value Δy measured from the cutting position of the piece is calculated. A control signal corresponding to this calculation is generated from the calculation/control circuit 65, and by controlling the drive circuit 25, the mounting table is controlled according to the comparison result value, and the processing tool 28 of the member to be inspected 26 is controlled.
The position relative to the position is corrected and controlled.

In this way, the present invention corrects and controls the position of the test member with respect to the processing tool by referring to the immediately previous processing result each time the same test member is processed from the second time onward, thereby improving the accuracy during continuous processing. The system is configured to correct each time a deviation in the relative position between the processing tool and the member to be inspected, which tends to occur in processing tools and the like.

Next, modified examples of the deviation detection pattern of the tested member in this invention are shown in FIGS. 4(A), 5, 6, and 7.
Each is shown in the figure.

It should be noted that each of these modified examples is optimal for obtaining the effects of the present invention when applied when the mounting error of the test member on the mounting table 21 is large.

FIG. 4(A) shows a shape in which two wedge-shaped patterns are combined in opposite directions, with a positional relationship diagram of the projected image on the line sensor 5, similar to FIG. 1(A). .

That is, each pattern 44-1, 44-2 has two central axes 45-1 parallel to the Y-axis direction and separated from each other in the Y-axis direction.
．． 45-2 are each axially symmetrical in shape, and each has a wedge shape in which the inclination angle θ of the inclination portion is equal and the inclination direction is opposite.

By detecting such a pattern with the line sensor 5 in the same manner as described with reference to FIG. 1(A), a detection output having a waveform as shown in FIG. 4(B)K can be obtained. As in the case of the rabbit, each of these can be calculated by calculating the following equations (3) and (4) using each count value 01 to n4 obtained by setting the clock pulse at the timing corresponding to the specific position to 0 pit. Calculate the Y-axis position n and the X-axis position n corresponding to the number of pulses n corresponding to the center pulse of pattern 44-1. All you have to do is ask for.

ny = ((n, +n2)/2 + (n3+ n4
)/2)/2...(31n, =((r+2+
r+, )/2-n)/lanθ ・・・・・・・・・
...(4) The pattern example in Fig. 5 is 4 lines arranged in the Y-axis direction.
This is an example formed by two axially symmetric wedge shapes. In this example, the four wedge-shaped patterns 46 to 1 to 46-4 are made to have different sizes in the Y-axis direction, and the gaps between each pattern are all formed to be equal. Not only the amount of deviation but also the amount of rotational deviation of the coordinate axes can be calculated at the same time.

FIG. 6 also shows an example of the shape of a deviation detection pattern that can achieve the same purpose. This example has a shape in which wedge-shaped patterns of equal dimensions are arranged in opposite directions in the Y-axis direction, with the convergence of the mutual gaps sequentially different.
Similar to the example illustrated in FIG. 5, each deviation amount can be detected with extremely high accuracy.

FIG. 7(A) shows other examples of deviation detection patterns, including a sloped pattern 48-1 having the same width and patterns 48-2 to 48-4 parallel to the Y axis and having different widths. Three pairs of patterns combined with any one of the patterns are arranged in the Y-axis direction and formed on the test member, and the strip width of each of these patterns 48-2 to 48-4 is as follows:
This also differs from the width of the inclined pattern 48-1.

That is, when the mounting error of the test member on the mounting table is large, any one of the three patterns is detected by the line sensor 5 of the deviation detection means set at a predetermined position outside the mounting table. Three sets are arranged at appropriate intervals on the Y-axis so that they can be captured.

The lower patterns 48-2, 48-3, 48-4 of each group have a band-like shape centered on each central axis 49 parallel to the 49, it has a band-like shape centered on each central axis 50 making an angle θ. Any pattern in each of these sets, for example, the top pair pattern 48-1.48 as shown.
-2 is detected by the line sensor 5, a detection signal with a waveform as shown in FIG. 7(B) is obtained.

By counting the X-axis, which corresponds to a specific position outside the mounting table, as the origin, as explained using FIGS. 1(A) and (B), clock pulses up to each edge of the detected waveform can be calculated. Count values nI , '2 , n3 , n4
get. Using this, the following equation (1') and the following equation (5)
By calculating the positions Y3 and X in the Y-axis direction,
The axial position XI is calculated and aligned.

Y3'''(r++ +n2)/2 ・・・・・・・・・
・・・・・・・・・・・・・・・・・・・・・α)X
l = ((n3+n, )/2 (nt +n2)/
2)/tanθ-(51 When preparing a plurality of such patterns and detecting any one of them, it is necessary to identify which pattern has been detected by the line sensor 5. The lower pattern widths of the plurality of patterns forming the plurality of patterns are different from each other for identification purposes.

That is, the difference between the count values n and n2 determined from the output of the lie/sensor 5 corresponds to the width of the detected pattern 48-2 in the Y-axis direction.

Therefore, by calculating the difference and comparing it with a predetermined known value corresponding to the width, it is possible to easily detect at which position the pattern has been detected. Therefore, the timing of starting counting n, to n, may be changed depending on the detected pattern.

Even when using the patterns shown in Figure 1 (A), Figure 5, Figure 6, and Figure 7 (A), the positional accuracy in the Y-axis direction cannot be determined with fairly high accuracy. However, in order to prevent detection position errors when using diagonal patterns or a small number of patterns, in order to obtain higher detection position accuracy in the Y-axis direction, multiple It is desirable to use the strip pattern as the deviation detection pattern.

FIG. 8(A) shows an example for precise positioning. In the illustrated example, three strip patterns 51-1.51-2.51-6 are formed at equal intervals in the Y-axis direction.
This is an example in which the patterns are formed as a set of patterns.

A pattern 51-1.51-2.51-6 for precise alignment is used as a deviation detection pattern as shown in FIG. 8(A).
5, 6, or 7 (A) for rough alignment to detect the amount of deviation. Thereby, the mounting table is controlled to move the member to be inspected so that the line sensor 5 of the deviation detection means is located at the projection position of the precision alignment pattern. As a result, a one-dimensional image in the Y-axis direction of the deviation detection pattern for precision alignment shown in FIG. 8(A) is read out by the line processor 5.

That is, the deviation detection pattern for precise alignment is such that when the position Yc of the central axis 52 of the three parallel patterns 51-1 to 51-6 is detected as a predetermined value, the member to be inspected is correctly positioned. is formed on the member to be tested. Then, in a state where the member to be inspected has moved as described above,
It is arranged in a projected relationship with respect to the line sensor 5 as shown in the figure. Usually, it is appropriate to provide them at two locations on both sides of the member to be tested so that the rotational deviation of the member to be tested can be detected at the same time.

FIG. 8(B) shows the relationship between the output waveform and the counted values n, -n6 at the edge portion of the output waveform, which are counted based on a specific position outside the mounting table.

By calculating the following equation (6) from each of these count values n, ~n, the position Yc of the center axis 52 of the extremely accurate and precise positioning deviation detection pattern is determined, and this is used as the reference value. Correct alignment can be achieved by comparing and displacing the member to be tested so that the amount of deviation as a result of the comparison becomes zero.

Yc = ((nt + n2) + 2 (r+3+n
4) + (ns + na) )/8...(
6) Also, as described above, if the deviation detection patterns for precision alignment are provided at both ends of the test member in the X-axis direction, and Yc is determined and compared for each pattern, the test member can be detected. Since the rotational angle deviation of the member can be detected accurately and easily, it is possible to correctly align the rotational angle deviation as well.

In this invention, the shape and structure of the deviation detection pattern of the test member are not limited to those of the above-mentioned embodiments. It goes without saying that any shape is acceptable as long as it provides axial displacement information.

〔Effect of the invention〕

According to this invention, the deviation information of the tested member, for example, the deviation with respect to a specific position of the tested member is obtained from a one-dimensional image signal obtained using a line sensor having high resolution with a pixel count of 2048 to 4096. Since the amount is detected, for example, compared to the conventional method of detecting deviation information from a two-dimensional image signal obtained using an image sensor with 512 x 512 pixels,
Since the calculation processing time is significantly reduced, the alignment speed can be further improved.

Further, since the amount of deviation is obtained by processing the image signal obtained by the high resolution of the line sensor, the calculation accuracy of the amount of deviation is high.

Moreover, when machining multiple locations on the same workpiece, the dimensions of the previously machined piece are referenced to determine the positional deviation between the workpiece position and the processing tool, and this is used to control the position of the workpiece. The system is structured so that the positional deviation is fed back to the drive system of the tool and corrected for each machining after the second machining, and then the machining is performed. Even if an error occurs, it is possible to process the test member with extremely high precision.

[Brief explanation of drawings]

FIG. 1(A) is an explanatory diagram of the relationship between the basic form of the deviation information detection pattern formed on the object to be aligned and the line sensor in the deviation detection means for detecting the pattern in the apparatus of the present invention. Figure 1 (B) shows the output waveform of the line sensor in Figure 1 (A), and Figure 1 (C) shows the pattern obtained by detecting the pattern including the processed end of the processed piece after processing the test member. An explanatory diagram of the output waveform of the obtained line sensor, FIG. 2 is a configuration diagram showing an example of an embodiment of the device of the present invention, and FIG. 3 is a diagram showing the formation of a deviation detection pattern formed on an object to be aligned in this invention. , a front view of a semiconductor wafer showing an example of the position, FIGS. 4(A) and 4(B), another example of the deviation detection pattern and an explanatory diagram of the output waveform of the line sensor in that case, FIG. FIG. 6 is a diagram showing the shape of another deviation detection pattern, FIGS. 7(A) and (B) are diagrams of other deviation detection patterns and the output waveform of the line sensor, and FIG. As shown in Figure 5,
FIG. 7 is a pattern diagram suitable for use in fine positioning when the deviation detection pattern of FIG. 6 or FIG. 7(A) is used as a rough positioning pattern. 1.1-1・1~1-n-n, 44-1~44-2.4
6-1~46-4.47-1~47-4.48-1~4
8-4.51-1 to 51-3...Displacement detection pattern, 2.6...Specific position, 3...Alignment to specific position of test member Position to try, 4.4
5-1.45-2.49.50.52... Central axis of each pattern, 5... Line sensor, 2
1... Placement table, 22... Rotating table, 23... Y-axis movable table, 24...
・X-axis movable table, 25... Drive circuit, 26
...Test member, 27... Processing tool drive section, 28... Cutting wheel, 29... Deflection detection means, 60...・Light source, 31... Half mirror 162... Sample hold circuit, 3
6... A/D converter, 64... Memory, 35... Arithmetic/control circuit, 41...
Semiconductor wafer, 42-1 to 42-(n-1)...
...Position to cut, 46-1 to 43-m...
...IC chip. Patent applicant: Citizen Watch Co., Ltd., NEC Corporation, Sumitomo Special Metals Co., Ltd.: Deflection extraction pattern, 2.6A ma1! 1 (jtl f: Fintenza Figure 1 Figure 2 Figure 2 J-Jashirindai 2q I L Production Shun 22 Kuchi Hitoshi Al 3 OL person 2
Y# Movable table 31 Half mirror
24, N axis working team 7 ゛ ru 26: 类绡 Component 27 Kanjimin Ij part 2If: c77 吋IJmu 3rd figure f-f-f ~ f-7?・2n My moxibustion extraction horn 42-f to 42-(n-f): A device that tries to generate elasticity 43-f -43-7n: Engineering C,
A41 #-Captive heather wafer Fig. 44-f, 44-2 IJR positioning pattern 45-
f, 45-2°individual/)/Z turn instep (fine Fig. 5 Fig. 6 Q x Fig. 7
! ;0: Individual 4Q pattern talent + (l mackerel Figure 8!; f-f -51-3rd and 4th place detection pattern 52, solid Baku->'s V-centered 2t t

Claims (1)

[Claims] A workpiece is placed on a mounting table configured to be displaceable in at least one of the X-Y coordinate axes with respect to the processing tool, and the processing tool is set at a specific position by displacing the mounting table. In a processing machine that positions the object relative to the object and processes it using the processing tool, the object has a deviation detection pattern formed in a shape that allows obtaining positional information corresponding to the amount of deviation at least in the axial direction. a deviation detecting means configured by arranging a line sensor in such a relationship that a one-dimensional image signal of the pattern is obtained at least in the axial direction when the pattern is placed on the mounting table; By processing and calculating one-dimensional image signals, position information of the pattern with respect to the specific position and position information of the pattern with respect to the position of the processed end after processing are obtained, and the former pattern position information is The amount of deviation in the axial direction is calculated with reference to the corresponding known standard position information, and the machining position is determined prior to subsequent machining of the workpiece by referring to the latter pattern position information. By comparing the former pattern position information, a relative position error between the processing tool and the workpiece due to the displacement of the processing tool is calculated, and a control signal for the deviation amount and relative position error is obtained. It consists of the configured arithmetic and control circuit, and drives and controls the mounting table using a control signal corresponding to the above-mentioned deviation amount every time the same workpiece is processed from the second time onward. A precision processing machine characterized in that the positioning position based on the deviation amount is corrected by feeding back a control signal corresponding to the relative position error to the mounting table drive system.