JPS63168708A - Positioning device for rotation - Google Patents

Abstract

PURPOSE:To attain stable angle detection even if dust or the like exists on a base by providing a picture processing section calculating the tilt of the base from the relative position of each cross point of a grating line. CONSTITUTION:A sensor control section 7 fetches a sensor signal (a) from a linear sensor 6 and sends a synchronizing signal (b) to a driving unit 4. The drive unit 4, according to the synchronizing signal (b), supplies a Y axis driving signal (c) to move the Y axis by a prescribed quantity in response to the resolution of a picture to an XY table 3. A signal from the linear sensor 6 passes a sensor control section 7 as the sensor signal (a), goes to a one line picture signal (d) of the base 1, converted into a binary by a binary picture memory circuit 8 and stored in a picture memory. Then the stored picture is inputted to a picture processing section 9 as a binary picture signal (e), where only the grating pattern such as a dicing line is extracted from the binary picture, the tilt is calculated and the tilt signal (f) is inputted to the drive unit 4. The drive unit 4 outputs a rotation drive signal (g) to drive a turntable 2 by a quantity in response to the tilt signal (f).

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention is applicable to a rotary positioning device, particularly to a sorting device or an inspection device for substrates having a lattice pattern as seen in semiconductor wafers, for example, to wafers after die cutting. The present invention relates to an applicable rotary positioning device.

[Conventional technology]

As a conventional technique, for example, Japanese Patent Application Laid-Open No. 55-12314
There is a method of rotationally positioning a semiconductor wafer as disclosed in Japanese Patent No. 5.

The conventional rotational positioning method involves detecting contrast to the wafer in the Y direction (or X direction) at two points several inches apart in the X direction (or X direction) on a semiconductor wafer, and It is configured to include a part that calculates the maximum linearity of the cross-correlation function of both detection output distributions obtained, and a part that uses the maximum value to calculate the rotation angle of the semiconductor wafer and performs positioning.

Next, a conventional method for rotationally positioning a semiconductor wafer will be described in detail with reference to the drawings.

FIGS. 5 and 6 are explanatory diagrams of a conventional method for rotationally positioning a semiconductor wafer.

In Fig. 5, the position [A, B
The laser beams are irradiated in the Y direction and scanned by several inches. When the scanning results are detected, a distribution having periodicity for each tick is obtained, for example, as shown in FIG. 6. Here, each output distribution f1° is expressed as (1).

In this case, the value of σ is 1 inch, 1 inch of the length in the Y direction of
It is desirable that the value be smaller than /2.

In other words, 1σ]<HLY (LY: length of one chip in the Y direction)
・・・・・・・・・(2) (taking into consideration the measurement point position A in the X direction).

Set B. Under such conditions, the function value K of equation (1)
Find the value σmax that maximizes θ=jan"""(a max / L X )
・=-(3) (θ: Declination angle to be corrected, LX: As2
(distance between the two) and performs rotational positioning.

[Problem that the invention seeks to solve]

The conventional rotational positioning method described above uses a laser beam to scan the wafer at certain intervals, and then compares the waveforms of the two resulting surface profiles to detect the phase shift. Therefore, if there is a defect such as a pattern with strong contrast, dust, or scratches inside the chip, the waveform at that position changes and the two waveforms do not match, making it impossible to perform accurate comparison calculations.

[Means for solving problems]

The rotary positioning device of the present invention comprises: a rotary table for fixing a substrate having a regular lattice pattern such as found in semiconductor wafers; an X-Y table for fixing the rotary cheagles;
@Y table and a drive unit that controls the rotary table; a one-dimensional quantity fixed in parallel to the X direction of the X-Y table in order to image the substrate; and a drive unit that controls the one-dimensional sensor. a light source fixed so that specularly reflected light is incident thereon; and a drive unit that controls the one-dimensional sensor and that drives the unit.

a sensor control unit that sends a synchronization signal to the one-dimensional sensor; a binary image memory circuit that converts the signal of the one-dimensional sensor into a binary image and stores a binary image of the substrate; and an image processing section that calculates the inclination of the substrate from the relative position of each intersection of the lattice method lines.

〔Example〕

Embodiments of the present invention will now be described in detail with reference to the drawings.

FIG. 1 is a block diagram showing one embodiment of the present invention.

The rotary positioning device shown in FIG. 1 includes a rotary table 2 for fixing a substrate 1 having a lattice pattern seen on semiconductor wafers, an X-Y table 3 for fixing the rotary table 2, and a rotary table 2 and an X-Y table 3 for fixing the rotary table 2. A drive unit 4 that controls the Y table 3, a light source 5 that illuminates the top of the substrate 1, a one-dimensional sensor 6 that images the top of the board, a seven-sensor control unit 7 that controls the one-dimensional sensor 6, and a sensor control unit 4 that controls the Y table 3; It is composed of a binary image memory circuit 8 which binarizes the image signal from the unit 7 and stores the binary image, and an image processing unit 9 which calculates the slope from the binary image.

The substrate 1 is illuminated by a light source 5, and the specularly reflected light is captured by a one-dimensional sensor 6. Therefore, it is possible to reliably capture a grid pattern such as a ding line on the substrate 1.

The one-dimensional sensor 6 is fixed parallel to the X-axis of the X@Y@-bull 3, and creates a two-dimensional image by moving the Y-axis.

The sensor=ltn unit 7 takes in the sensor signal a from the one-dimensional sensor 6 and sends a synchronization signal a to the drive unit 4 in synchronization with it. The drive unit 4 sends a Y@drive drive signal to the X-Y table 3 in accordance with the synchronization signal to move the Y axis by a predetermined amount depending on the resolution of the image.

The signal from the missing element cell 6 passes through the sensor control section 7 as a sensor signal a, and becomes an image signal d of one line of the substrate 1.

The image signal d is converted into a binary signal in the binary image memory circuit 8 and stored on the image memory.

As described above, images are sequentially captured by the one-dimensional sensor 6, and a binary image of the substrate is formed in the binary image memory circuit 8.

The image stored in the binary image memory circuit 8 is input to the image processing section 91C as a binary image signal e.

The image processing unit 9 extracts only grid patterns such as dicing lines from the binary image and calculates the slope thereof. The result is input to the drive unit 4 as a tilt signal f.

The drive unit 4 outputs a rotational drive signal g as a signal for rotating the rotary table 2 by a weight corresponding to the tilt signal f.

FIG. 2 is a block diagram showing the configuration of the image processing section 9. As shown in FIG.

The binary image signal e outputted from the binary image memory circuit 8 has an isolated point of one pixel removed by a noise removal port MiO, and is output as a noise removed image signal.

As shown in FIG. 3, the noise-removed image signal includes not only a lattice pattern but also a large amount of noise due to flaws such as scratch marks or pads.

The noise-removed image signal is input to an X-direction labeling circuit 11 and a Y-direction labeling circuit 2.

The X-direction labeling circuit performs a labeling operation of assigning the same number to consecutive pixels in the X-direction of the noise-removed image, and outputs an image i labeled in the xy5 directions. X
The 71i-direction line extraction circuit extracts only the X-direction line component based on the X-direction label image amount and outputs it as an X-direction line signal j.

Similarly, in the X-direction labeling circuit 12, Yy5
Labeling is performed for continuity in the direction, and a Y-direction label image k is output. Similarly, the Y-direction label image is input to the X-direction line extraction circuit 14, and after extracting the X-direction twin component, it is output as a Y-direction line signal l.

The line signals j and l are added together by an OR circuit 15 and input to a thinning circuit 6 as a line image signal m.

The line thinning circuit 16 converts the line image into a line drawing with a width of one pixel, and outputs a line image signal n.

- Image n is input to the intersection detection circuit 17, and the intersection coordinates of each line segment and the label number of that point are output as intersection coordinate signal 0.

The intersection coordinate 0 is input to the straight line approximation circuit 18. The straight line approximation circuit 18 approximates a straight line using the defect coordinates having the inward label of the input intersection coordinate signal 0. in this case,
The approximate straight line is determined by the method of least squares.

Also, at this time, each defect has two labels, one in the X direction and one in the
A straight line image signal P is output.

The straight line image signal P is input to a slope calculation circuit, and by averaging the slopes of the straight lines in the same direction, a highly accurate slope can be obtained, and a slope signal f can be extracted.

FIGS. 4(a) and 4(b) are operation explanatory diagrams for explaining labeling in the Y direction of a binary image.

In the binary image shown in FIG. 4(a), the line component and noise component are expressed as 1 black, and the background component is expressed as 1 white.

First, a binary image is raster scanned in the X direction from the first line q. At this time, label @1” on the first 1 black point.
give. Next, give the label 11" to the adjacent 1 black" point in the same way. If the 5 "black" points are broken, set the label to 12" and label the next 1 black" point. Give it that label when it appears. By sequentially repeating this process, labeling for the line components in the Y direction is completed. At this time, if there is a noise component on the t4] line, that point is also labeled.

Next, from the second line, the label attached to the upper two pixels SK among the eight neighboring pixels shown in FIG. 4 fb) is attached to the center pixel tK, which is the scanning shield point.

At this time, if there are no 2 bells in the upper two pixels S, no labeling is performed. Furthermore, if there are different labels within the upper two pixels S, priority is given to the one with the smaller label value. This process is performed for all pixels, only the labels remaining on the final line U are made valid, and only the pixels having valid labels are extracted by the Y-direction line extraction circuit.

For labeling in the X direction, similar processing is performed using raster scan in the X direction.

[Effects of the Invention] The rotary positioning device of the present invention performs an operation of labeling the same area of a binary image of a substrate having a pattern of the lattice method found in semiconductor wafers, etc., and labels only the line segments of the lattice method. Since the board is being removed, there may be dust, scratches, and scratch marks on the board. Stable angle detection is possible even when there are noise components such as cracks and chips, and when targeting semiconductor wafers, the ding lines are extracted, so valleys and edges can be cut out at the same time. There are 5 benefits that you can do.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a group diagram showing details of the image processing section shown in FIG. 1, and FIG. 3 is an explanatory diagram of a binary image. Figure 4 is (a),
(Operation explanatory diagram for explaining bl labeling, FIGS. 5 and 6 are explanatory diagrams showing conventional examples. 1... Board, 2... Rotating Cheagle, 3
...XIIY Cheagle, 4... Drive unit, 5... Light source, 6... One-dimensional sensor, 7... Center control Part, 8...Binary image memory, 9...U! JJ image processing unit, 10
... River noise removal circuit, IJ ... Y direction labeling circuit, 12 ... X direction line extraction @ road,
13... Y direction labeling circuit, 14...
... Y-direction twin extraction circuit, 15 ... OR circuit, 16 ... Thinning circuit, 17 ... Intersection detection circuit, 18 ... Group/straight line approximation circuit, 19.・・・・・・
Slope true output circuit. a...Sensor signal, b...Synchronization signal,
C・Old...Y-axis movement signal, d...Image signal, e
...Binary image signal, f...Tilt signal,
g... Rotation drive signal, h... Rise removal image signal, i... X direction label image, J...
...X-direction line signal, k...X-direction label image,! ...Y direction line signal, m... group/fin image signal, n... line image signal, 0...
...Intersection coordinate signal, p...Line image signal, q.
・・・-・First line, R・Old・・Second line, S・・
...Top two pixels, t...Center pixel, U...
...Final line. Gi 2 Zuto 3 Figure C Haruka) Certain 4ei ``菟 5 times 4pu Otsu Kan

Claims (1)

[Scope of Claims] A rotary table for fixing a substrate having a regular grid pattern such as that seen on a semiconductor wafer, an X-Y table to which the rotary table is attached, the X-Y table, and a drive unit that controls a rotary table; a one-dimensional sensor installed parallel to the X direction of the X/Y table to image the substrate; and specularly reflected light incident on the one-dimensional sensor. a light source installed to control the one-dimensional sensor; a sensor control unit that controls the one-dimensional sensor and sends a synchronization signal to the drive unit; a binary image memory circuit for storing, and an image processing unit that labels the binary image, extracts the grid lines of the board, and calculates the inclination of the board from the relative position of each intersection of the grid lines. A rotary positioning device comprising: