JPS6236242A - Wafer processing device - Google Patents

Abstract

PURPOSE:To enhance the productivity by accommodating information about actual wafer position in No.2 memory in correspondence to the value in No.1 memory, in which the information about standard wafer position is accommodated, and by setting or removing the wafer according to the actual position information. CONSTITUTION:A driving part AD for setting and removal of wafers, which is movable vertically and rotatable, is installed between a plurality of wafer cassettes WK (WK3, WK4). The information about standard position for wafer WFi is stored previously in CPU. The beam from a laser source LZ in said driving part for setting and removal AD passes through a half mirror HM and between wafers WFi, and is total reflected by a mirror FM to proceed reversely for a sensor PS. Every time a wafer WFi is encountered during sink of said driving part AD, the beam is blocked, and actual position information in correspondence to the standard position information is stored in the CPU. A finger FG is intruded into the cassette WK so that takeout and accommodation of wafer WFi is made in accordance with the actual position information. This will enlarge the wafer processing capacity.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field] The present invention relates to a wafer processing apparatus used for processing wafers for manufacturing high-density integrated circuits, ie, circuit baking, surface inspection, chip operation testing, and the like.

[Prior Art] Conventionally, this type of apparatus has had drawbacks in its accuracy, productivity, occupied area, etc., and in particular, the processing speed when taking out and storing wafers was slow.

[Objective] The present invention solves the above-mentioned difficulties, and provides a wafer processing apparatus that has features such as ultra-precision, high productivity, and miniaturization, and has improved processing speed especially when taking out and storing wafers. The purpose is to

[Embodiment] Fig. 1 shows the entire structure of a uni hub according to an embodiment of the present invention.
Figure 2 shows a schematic diagram of the body. In the figure, WK1 to WK4 are four wafer cassettes that store a large number of wafers WF, and are stacked in two stages with their respective openings facing each other. WKDI, WK
D2 is a drive unit □ for vertically moving each of the two stages of wafer cassettes via the support plate KS, FG is a finger for mounting wafers, and AMI and AM2 are arms. G
l, Gl.

Q and 3 are axes, arms AM1, AM2 and finger F
G is configured so that it can be moved telescopically. AD is a drive unit for telescopically moving the arm and finger. Further, this drive unit ΔD is configured to be capable of vertical movement and rotation by mounting the arm and finger.

The wafer WF carried out from the wafer cassette WK by the arms AM1, AM2 and fingers FG is centered by the centering hand C)I and placed on the ×Y stage ST.
The wafer is transferred onto the upper wafer chuck WC. The transferred wafer WF is repositioned by the electrostatic old flaw type sensor QS. At this time, the positioning is performed manually or automatically by visual observation using a television (TV) camera TVK and a television monitor TVD.

The wafer WF, which has been accurately positioned on the wafer chuck WC by the centering hand CH, is set directly below the microscope OP by moving the XY stage ST, and the positions of the chip terminals and probe terminals (not shown) on the wafer WF are determined by visual observation or automatically. A match is made. Thereafter, a predetermined chip test operation is performed, and after the test, the wafer WF is returned to its original position in the wafer cassette WK by the fingers FG1 arms AM1 and AM2. By repeating the above operations, the chip test operation is sequentially performed on each wafer WF within the wafer cullet WK.

Second. FIG. 3 is a side view and a top view of the main part of FIG. 1, showing how the wafer WF is taken out and stored from the wafer cassette WK. However, finger FG and arm AM
1. AM2 has been rotated 90° clockwise with respect to FIG. In the figure, for example, wafer cassette WK4 is raised together with wafer cassette I-WK3 to a predetermined wafer delivery and collection position by wafer cassette drive section WKD2.

Finger FG and arm Av1. AM2 is expanded and contracted by a pulse motor SM built in the arm expansion/contraction drive unit AD to take out or store wafers WFI to WF4 and the like.

The arm telescopic drive unit AD is moved up and down together with the base KD by the movable plate EB of the elevating mechanism I'D, and the wafers WF1 to WF4 are moved up and down together with the base KD.
A specific wafer is selected from among the following. MO is a rotation drive unit for rotating the arm extension/contraction drive unit AD along with the base KD, and performs a wafer transfer operation between the wafer cassette WK and the wafer chuck WC.

LZ is a semiconductor laser source, HM is a half mirror that irradiates the laser beam LW from the semiconductor laser source LZ toward the wafer cassette side, and the laser beam is reflected by a total reflection mirror FM (FMR or FML) provided at the outer edge of the wafer cassette. The laser beam LW is transmitted and transmitted to the semiconductor laser sensor PS.

A semiconductor laser sensor PS inside the protective case SB detects this reflected laser light.

Since this apparatus is configured as described above, the wafer W
The current position of F can be reliably known, and malfunctions such as collision between finger FG and wafer WF can be prevented.

Furthermore, since a semiconductor laser source is used as a light source, the accuracy of wafer position detection is significantly improved compared to a light emitting diode or the like. Furthermore, since the laser beam can be blocked according to the rotation of the drive unit AD, safe operation can be achieved. That is, the laser beam blocking plate LS is integrated with the plate EB, and the plate LS is provided with a hole LH, and the laser beam passes through the hole LH at the position shown in FIG. 2, and there is no hole at the position shown in FIG. It is blocked because of this.

This is preferable because the laser beam is emitted only at necessary positions and not at unnecessary positions, ensuring safety during maintenance and inspection.

Furthermore, since the semiconductor laser source LZ is arranged so that the light emission direction is directed upward as shown in FIG. Pulse motor SM for expansion and contraction
This is preferable as it allows coexistence with Also half mirror-1-
IM and the semiconductor laser sensor PS are arranged at a position opposite to the direction in which the fingers FG extend, as shown in FIG. 2.3.

This is preferable because the optical path of the sensor PS is not blocked by the arm and the extension and contraction of the arm is not obstructed.

This telescopic arm has a vacuum tube TtJ for wafer suction.
are aligned by a method such as penetration, groove insertion, or adhesion. As a result, the arm, which has a large movement range, and the tube move around in conjunction with each other, so the tube and arm do not become entangled, and the overall space can be made smaller.

In FIG. 2, KS4 U and KS4 L are wafer uppermost (lower) position information generating means each consisting of, for example, a protrusion, and LE and PH in the lifting mechanism HD detect the position of the movable plate EB, and are used to transfer or collect the wafer. It is a light emitting/receiving element that determines the position.

The finger FG has a simple configuration of two upper and lower plates FGU and FGL glued together as shown in FIGS. 4A, B, and C. The upper plate FGU is provided with through holes U1 to U10, and the lower plate FG is provided with through holes U1 to U10.
L has a groove L11. L'1 through holes L2 to L7 are provided.

And groove L1 and hole ui, U2.

A vacuum suction chamber is formed with U3. That is, the groove 11 is the hole U.
10. Through the tube holder TH J, -BU TU
is connected and exhausted. When a wafer is placed in the upper part of the suction chamber formed as shown in FIG. 4B, the wafer is suctioned and sealed, and the wafer is suctioned. Also, hole jJ8. l
J9. L6. L7 etc. have bins SR and SL as shown in Figure 4C.
are fitted. Bins SR and SL contain wafers W as shown in Figure 3.
A positioning member for positioning the end face of F, 4 inches,
For wafers with different diameters such as 6 inches and 8 inches, select one of holes U4, U6, and tJ8 to attach pin SL.
mate. Similarly, pin SR has holes us and U7.

Select U9 and fit it. This is preferable because the pin positions can be determined optimally for wafers with different diameters.

As shown in FIG. 2, the upper wafer cassettes WK1 and WK3 are constructed so that, for example, support plates KS3 and KS3B can be bent outward. As a result, the lower cassettes WK2, W
This is preferable because the K4 replacement operation becomes easy. Note that it is not necessary to provide two mirrors FM at the rear of the cassette, one above the other, and only one mirror with a length that corresponds to a single cassette at a position corresponding to the wafer removal and storage position (the upper position of the cassette in Figure 1). Just be prepared.

The aforementioned sensor PS is the wafer cassette identification code information KC.
3. KO2 etc. can also be read.

That is, KO2, 04, etc. consist of codes representing logic r1J, rol, and the sensor SP moves vertically and/or horizontally to read the code information such as 03, etc., thereby knowing the cassette identification number. .

FIG. 5 shows an example of a wafer WF used for circuit burning, inspection, etc. In the figure, WCN is code information for identifying a wafer, and a wafer number, wafer cassette number, lot number, etc. are prepared. Various methods are used to write this information, such as exposure printing from the reticle original plate, laser printing, ink, etc.
The television camera TVK shown in FIG. 1 is used to read this information. CHP indicates the circuit chip area to be tested.

Figure 6 is a front view of the opening of the wafer cassette WK.
A state in which, for example, 25 wafers WF of FIG. 5 are stored is shown. Here, wafers WF2 and WF3 are shown with emphasis on the case where they have been deformed into arcuate shapes due to various treatments of the wafers. Strictly speaking, it is generally considered that the wafer has lost its flatness, so it is necessary to pay close attention to the insertion position of the fingers FG when taking out the wafer. Therefore, it is sufficient to accurately detect the position of each wafer and calculate the safest insertion position of the finger FG, that is, the center position P1 to P26 between each wafer, etc., before the finger moves in and out.

FIG. 7A is a control block diagram for that purpose, and FIG. 8 is a flowchart showing the control procedure, which is stored as a microinstruction in the read-only memory ROM shown in FIGS. Explain. In step SAI, the wafer position detection means, for example, the laser sensor PS starts to descend and starts detecting the wafer position, for example, when it detects the uppermost position signal generator KS4U shown in FIG. The first distance d26 is determined when the sensor PS detects the first wafer WF25, the distance d25 is similarly determined when the second wafer WF24 is detected, and the detection is performed in the same manner up to the lowest dl.

At this time, the sensor PS may be set at a constant speed, that is, a constant speed that is lower than the response speed of the sensor PS, but this inevitably lowers the overall detection speed.

Therefore, as shown by waveform A in FIGS. 6 and 7B, the sensor PS is lowered at high speed at the position where the wafer is not present.
It is sufficient to move at a low speed at the position where the wafer is present.

Therefore, the positions D1 to D26 of each wafer in the standard wafer cassettes 1 to 1 may be stored in advance in a part of the ROM shown in FIG. 7A, and the operation may be controlled according to the stored contents. That is, first, the microprocessor MPU takes in the standard positional information ID26 from the address AD1 of the ROM, and based on this information, determines the pulse train DM of FIG. 7B, for example. Pulse width T of each pulse of this drive pulse train DM
1 to T10 are stored in a table format in an unillustrated portion of the ROM, and first, the binarized time information corresponding to T1 is stored in the down counter DK under the control of the MPU. At the same time, flip-flop FF is set and motor drive circuit MD operates. A quaternary counter is built in the motor drive circuit MD, and the first output thereof opens the AND gate A1, and the first drive pulse (T1) starts the pulse motor PM.

Down counter DK starts subtraction in response to an oscillation pulse from oscillator O8C, and when the value reaches a specific value, for example O, a specific value decoder, for example O decoder DC, detects this and resets flip-flop FF. This ends the drive pulse T1. Similarly, AND gate A2. △3
．． A4 is sequentially interrupted, and driving with the pulse width T1 is completed.

Next, time information corresponding to the pulse width T2 is stored in the down counter DK, and the same operation as above is performed. As shown in FIG. 7B below <73. The pulse width gradually becomes shorter from T4, and the processing speed of sensor PS gradually increases until the pulse width T
At the repeat drive period 4, the lowering speed is lowered at a constant high speed. Furthermore, when approaching the vicinity of the first wafer WF25 shown in FIG. 6, the path/l and H path widths shown in FIG. 7B become T5 and T6.
The sensor PS gradually becomes longer, and the sensor PS continues to descend while being decelerated.
When the sensor PS reaches the laser beam, the sensor PS descends at a constant low speed sufficient to detect the laser reflected light. At this time, the pulse width T7
The pulse motor PM is driven by the drive pulse. When the position detection of the wafer WFI is completed, the pulse widths T8 and T9
, becomes gradually shorter like T10, the descending speed of sensor PS gradually increases, continues descending at a constant high speed as before, and when descending a predetermined mountain, specifies the second address AD2 of the ROM in Fig. 7A, Extract location information [)m. This positional information Dm is standard positional information D2 when expressed in terms of wafer spacing.
5 to D2 are common, and only one address AD2 is sufficient as the data area for the position information Dm. Thereafter, the same operation as described above is repeated until D2 according to the position information Dm, and finally, the address AD3 of the ROM is specified, and the same operation as described above is performed according to the position information D1 of the lowest side, so that the sensor PS is connected to the lowest position signal generator KS4L. is detected and the descent is stopped. Since the actual position of each wafer is detected in this way, the next full-scale wafer removal and storage becomes extremely accurate. Also, if it is detected that no wafer actually exists during this wafer position detection, write that fact to the RAM and write d3 <, and write this wafer presence information (1, 0) to the RAM together with the wafer position information. This is convenient because it allows you to quickly skip over positions that do not exist. In addition, the standard position information of other wafer cassettes of different sizes is also stored in the ROM addresses AD4 to AD6.
It is more preferable to use it selectively.

Next, in step SA2, the microprocessor MPU calculates the detected center positions ff1dl/2 to d26/2 between each wafer, and stores the multi-values in the RAM of FIG. 7A as shown. Next, in order to take out the wafer WF1, the finger FG is moved from the lowest point (currently located) to step SA.
It is raised to the position d1/2 calculated in step 2 (step SA3). This allows the finger FG to be positioned at the safest point P1 in FIG. To stop at this position, the matching circuit CO of both the target value storage register OR and the up counter UK1 in FIG. 7A is used. Waveform B in FIG. 6 shows the rising speed curve of the finger FG at this time.

At this point P1, extend the arms AMI and AM2 and insert the fingers FG into the wafer cassette WK (step SA
4) and then drive the lifting mechanism HD to move the finger F.
G is increased by d1/2 (step SA5). The upper surface of the finger FG can now come into contact with the lower surface of the wafer WF1, so the vacuum suction mechanism described above is turned on and operated, and the finger FG attracts the wafer WF1 (step 5A6). Then finger FG, then d2/
Raise by 2 (step SA7) L, then r wafer WF
Step 5A8) <. Next, the arm extension/contraction drive unit AD is connected to the drive unit M.
Rotate the finger FG by 90° using O and move it to the XY stage S.
The wafer WF1 is directly faced to the centering hand CH by extending the arms AM1 and AM2 and transferring the wafer WF1 from the fingers FG to the centering hand CH (step SA9). The wafer WF1 is roughly transferred from the centering hand CI-1 to the XY stage ST (step S A10), and undergoes processing such as circuit burning and inspection (step S5A11).

When the processing is completed, the processed wafer WF1 is transferred from the XY stage ST to the finger FG via the hinting hand CH.
is collected and adsorbed again (step 5A12.13). Next, the arms AM1 and AM2 are retracted and rotated, and the finger FG is lowered from the current position (the highest point) to the position of dl +62 /2 in FIG. 6, that is, the safety point P2 (step 5A14). ). At this position P2, arms AMI and AM2 are extended and finger F
Insert G into the cassette WK (Step S A 1
5) Do.

Next, lower the finger FG to d2/2 (step S
A 16) Let it happen. As a result, the wafer WF1 rides on the protrusions KR and KL at the original position, and then the vacuum suction mechanism is turned off, so the wafer WF1 is separated from the finger FG (step S A 17), and then the finger FG is further moved by d2/2. It is lowered (step S A 18) and only the finger FG is pulled out from the cullet WK (step S A
19) To be killed. The above control allows the wafer WF1 to be taken out and stored accurately and safely. Next, the wafer W
The control for taking out and storing F2 is now d2/2.
The same procedure as above is performed using the value of 63/2, and the same procedure is repeated until n=26.

FIG. 9A schematically shows the above situation. In the figure, X indicates the standby position of the finger FG,
The position of the wafer is detected when descending from this point to point Y (end point), and then the finger FG is inserted into the cassette WK at point A in order to take out the wafer WF1 closest to that position. is started, unloading is completed at point 8, transfer to centering hand CH is performed at point 0, collection from centering hand CH is performed at point D, and finger FG is moved to □ storage of the wafer at point E.
The insertion of the wafer into the cassette WK is started, the storage is finished at point E, and the removal and storage of the second wafer WF starts from that position.

In the figure, all three wafers, WF1 to WF3, are used for ease of understanding.
Shown as a sheet. Further, the processes corresponding to each point A to F are indicated by the same reference numerals in the flowchart of FIG. Further, the standby position X of the finger FG is set at approximately the same height as the centering hand CH and near the top wafer WF3. Therefore, the total vertical movement distance is shorter than in the example shown in FIG. 9B, which is preferable. This also applies to FIG. 9D.

FIG. 9B shows an example in which the position of the centering hand CH, that is, the wafer delivery or collection position, and the position of the lowest wafer WFI are set to be almost on the same line, and the standby position of the finger is also set on the same line, In this case, as shown in the figure, the total distance of vertical movement of the finger becomes long. Move the finger standby position @X in Figure 9B to the top wafer WF3.
Even when the finger is set at a position near , that is, at position Y in FIG. 9B, the overall vertical movement of the finger is almost the same as in FIG. 9B and is longer than in the case of FIGS. 9A and 9D, as shown in FIG. 9E. It's obvious when you see it. However, it is preferable that the wafers closest to the position where all the scanning for wafer position detection has been completed are taken out and then the storage is started. Furthermore, it is preferable to process the wafers from the bottom to the top so that the upper wafers are less contaminated. The example in FIG. 9C shows a situation in which the wafer transfer or collection position @ (the position of the centering hand CH) is set almost at the eccentric position of the wafer cassette, that is, near the middle wafer WF2, and in this case, the vertical movement of the finger FG is The lowest total distance and most preferred. This is evident from the fact that all wafers must be transported to a predetermined, fixed point, i.e., 18 transfer or retrieval locations. This advantage can be similarly expected in the case of FIG. 9F. Saving a small amount of wasted time in this way contributes to improving the overall speed of wafer processing in this type of apparatus, which is extremely desirable.

Generally speaking, not all wafers are always stored in the wafer cassette, and there are cases where defective wafers are removed in advance in the preprocessing process and transferred to the main blower. If the operations related to the wafer removal and storage of the fingers FG are skipped at the position where the wafer is not present, the overall processing speed can be improved. This will be explained below. In FIG. 6, it is assumed that the wafer W F 2 is now missing. As explained above, when it is detected that each wafer exists at a predetermined position, a logic "1" is written to the RAM as shown in FIG. ．． d25/2
Write it to RAM like this. When the sensor PS detects that the missing wafer WF2 does not exist, the RAM address A is
Logic "0" is written in d2, and Dm is written as position information on the right side. As described above, Dm is the standard position information of the wafer. If processed in this way, the wafer WF
It is possible to avoid a measurement error in PS due to the absence of wafer 2, that is, the risk of mistakenly identifying wafer WF1 as wafer WF2 when wafer WF1 is detected.

Also, wafer WF2 that does not exist during wafer unloading processing
It is possible to increase the speed by omitting the wasteful operation of taking out the material. That is, in the above-mentioned wafer unloading mode, the wafer W
After removing or storing F1, address A, d of RAM
If 2 wafer no information rOJ is detected, di /2+D
By performing the calculation m+63/2, the vicinity of the 23 points in FIG. 6 can be obtained. Therefore, it is possible to skip the 12th point and increase the finger FG from the 11th point to the 23rd point at high speed, which is preferable.

In the example shown above, the reference positions X, Y, Z are the starting point information generators KS4 U, KS4 L, EB (or LE,
PH) etc. may be selectively made to correspond. For example, in the example of Figure 9A, X=KS4 U, Y=KS4 L,
In the example of Figure 9F, X=LE (or PH, EB), Y
= KS4L.

Furthermore, it is clear that the orientation of the wafer cassette is not limited to vertical, but may be set diagonally or horizontally (FIGS. 9C and 9F). Also, the wafer is 4 inches.

When the diameters of wafer cassettes 1 to 8 inches are different, the heights of wafer cassettes 1 to 1 are also different (the distance between each wafer is different), so
It is preferable to use a type of apparatus in which the lower side (common end) X or Y can be shared, as shown in the examples shown in FIGS. D and 9F, since it can handle various wafer sizes.

[Effects] As described above, the present invention can greatly contribute to high productivity.

[Brief explanation of the drawing]

Fig. 1 is an overview perspective view of an embodiment of the present invention, Fig. 2 is a partially enlarged side view thereof, Fig. 3 is a partially enlarged top view thereof, and Fig. 4A
Figures 4B and 4C are side views of the structure of the fingers, Figure 5 is a front view of the wafer, Figure 6 is a front view of the opening of the wafer cassette, Figure 7 is a control block diagram of the present invention, and Figure 7B. is a control waveform diagram, FIG. 8 is a control flowchart diagram,
9A to 9F are explanatory diagrams of the operation of the fingers. FG...Finger AM...Arm PS...Sensor WK...Wafer cassette

Claims (1)

[Claims] 1. A first storage means in which standard wafer position information is stored, and a wafer for detecting the actual wafer position in response to the standard wafer position information from the first storage means. a position detecting means, a second memory means for storing actual wafer position information from the wafer position detecting means, and a method for taking out and storing a wafer in response to the actual wafer position information from the second memory means. A wafer processing apparatus having a wafer insertion/removal means. 2. The wafer processing apparatus according to claim 1, wherein the first storage means stores position information of each wafer in a plurality of wafer cassettes.