JPH0128503B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an alignment method, for example an alignment method applied to mask alignment.

In the process of processing semiconductor wafers (hereinafter simply referred to as wafers) in the manufacture of semiconductor devices, in order to perform pattern printing on the wafer, the wafer is automatically transferred to an XY stage under a microscope, and a target provided on the wafer is placed on the wafer. It is necessary to align the mask. Generally, as shown in FIG. 1, this mechanism first transfers a wafer 1 from a cartridge 2 to a pre-alignment table 3 using air or a belt, and then rotates the wafer 1 on this pre-alignment table 3 for pre-alignment. After performing this, a two-stage mechanism is adopted in which the sample is transported to the XY stage 4 under the microscope again by a vacuum suction tool. In the above pre-alignment, since the wafer 1 is provided with an orientation flat (identification surface) 5 by cutting out a part of the wafer 1 in a straight line in advance, three rotating By rotating the wafer 1 with rollers (one rotating in the opposite direction) 6, align the identification surface 5 so that it comes into contact with the roller 6 (the one roller in the center does not contact the identification surface 5). .

By the way, although the positioning accuracy of the above pre-alignment can be adjusted to a relatively high precision (for example, ±50 μm), the positional deviation of the wafer 1 with respect to the optical field center when it is transferred onto the XY stage 4 is extremely large. If the target on the wafer 1 deviates from the optical field of view (for example, ±250 μm) and subsequent alignment becomes difficult, or if the deviation is large, the amount of movement correction of the XY stage is large and the correction time becomes long. This will cause some inconvenience.
It is believed that the above-mentioned large deviation is caused by a combination of mechanical deformation and wear of the arm shaft that supports the vacuum suction tool, changes in the relative positional relationship between the pre-alignment table and the XY stage, and positional deviation of the target with respect to the identification surface. It will be done.

By the way, the operation of moving the wafer target to the center of the optical field on the XY stage is not only to correct the relative misalignment between the XY stage and the pre-alignment table, but also the misalignment between the identification plane and the target. It won't happen.

Therefore, the present inventor memorized the amount of movement of the XY stage from the initial transfer position to the final destination position each time precision alignment of wafers was performed, and calculated the average amount of movement for each operation of ²ⁿ wafers. After the wafer is transferred to the XY stage, the XY stage is automatically moved and
The inventors realized that by using the XY table as a correction origin, the target can always be located within the optical field of view and the deviation of the target from the center of the optical field can be reduced.

Therefore, an object of the present invention is to improve work efficiency by ensuring that the target is located within the optical field of view and that the amount of positional deviation of the target is small during the target alignment operation of the sample. It's about trying.

In order to achieve such an object, the present invention provides an alignment method for sequentially positioning a sample having a position detection target within a specific range of field of view having a center, which includes: a) positioning the sample in a predetermined direction; a step of aligning, b a step of positioning the XY stage at the machine origin, c a step of placing the sample in step a above on the XY stage of step b above, d a step of moving the XY stage of step c above from the machine origin to the correction origin. a step of positioning the target for position detection within the field of view within the specific range; e a step of moving the XY stage in step d to position the target for position detection in the center of the field of view; f moving the target in step e. The corrected origin position in step d is an alignment method that includes at least the step of cumulatively adding the amount for each sample, and the corrected origin position in step d is corrected by referring to the average value of the cumulative addition. The present invention will be explained below with reference to Examples.

FIG. 2 is a schematic diagram showing an outline of a mask alignment apparatus to which an embodiment of the alignment method of the present invention is applied, and FIG. 3 is a flowchart showing the same operation.

In this apparatus, wafers 1 housed in a cartridge 2 are transported one by one to a pre-alignment table 3 by a belt conveyor (transport mechanism) 7. On this pre-alignment table 3, the wafer 1 is rotationally corrected and positioned (pre-aligned) by the three rollers 6 and the stopper 8 described above (see FIG. 4b). Further, the pre-aligned wafer 1 is held by vacuum suction by a vacuum suction tool 10 attached to the tip of the arm 9, and is transferred onto the XY stage 4 under the microscope. The arm 9 rotates left and right around its support part and connects the XY stage 4 and pre-alignment table 3.
The wafer 1 is transported by reciprocating between the two.
The XY stage 4 moves in the X direction with an X motor 11 consisting of a pulse motor, and moves in the Y direction with a Y motor 12.
Move in direction.

On the other hand, this device is provided with a stage correction circuit 13. This stage correction circuit 13 is provided with a stage movement amount detection section 14 . Information is sent to the stage movement amount detection unit 14 from a microscope (microscope camera) 15 that observes the wafer 1 on the XY stage 4. This information is converted from optical information to electrical information. Stage movement amount detection section 1
The output of 4 is sent to the adder 16 and the movement amount cumulative adder 1.
7 is input. The output of the movement amount cumulative addition section 17 is
It is input to the averaging unit 18 which calculates a moving average value by dividing the cumulative value of 2 ⁿ times by 2 ⁿ . The output of the averaging section 18 is sent to the correction amount subtracting section 19, and the origin correction section 2
Origin correction is performed at 0, and the signal is sent to the addition section 16. In accordance with this result, the adding section 16 moves the X/Y motors 11 and 12 after the wafer 1 is placed on the XY stage 4.

Next, the alignment operation will be explained with reference to the operation flow. The wafer 1 is aligned on a pre-alignment table 3 with an accuracy of within ±50 μm. Therefore, the vacuum suction tool (transport mechanism) 10
The wafer 1 is then transported onto the XY stage 4, which has been returned to its mechanical origin. Due to play in the transfer section of the vacuum suction tool 10, slight movement during wafer suction and release, etc., the alignment accuracy of the wafer 1 on the XY stage 4 is lower than the ±50μ accuracy in pre-alignment. There are many things. Therefore, while observing a target (not shown) on the wafer surface with a microscope, the target is aligned at a desired position, that is, the position of the target on the mask. This amount of movement is stored and added by the amount of movement cumulative addition section 17. When mask alignment is completed and exposure work is completed,
The wafer 1 on the XY stage 4 is automatically carried out from the XY stage 4. Then, the next new wafer is carried to the XY stage 4 again by the above procedure. Here too, mask alignment is performed by operating the XY stage 4. The corrected movement amount of the XY stage 4 due to this mask alignment is stored and cumulatively added. Also, it is asked whether or not the number of mask alignments for the wafer is the ^2nth time. Since this is the second mask alignment at this stage, the two movement amounts are averaged and a moving average value is calculated. The corrected origin of the XY stage 4 is determined from this moving average value. In other words, when the wafer 1 is placed on the XY stage 4 which has returned to the origin (mechanically) during the next third to sixth mask alignment, the corrected origin will be adjusted to the moving average value. Move by the corresponding amount and position at the correction origin. Therefore, mask alignment is performed while observing with a microscope. In this way, once and twice
Anticipate the movement correction before mask alignment by referring to the movement correction of the XY stage.
Since the XY stage 4 is automatically returned to the corrected origin, the target 21 of the wafer 1 almost certainly falls within the narrow microscope field of view (optical field of view) 22, as shown in Fig. 4d, and its precision is high. becomes. Therefore, the time required to correct the mask alignment (the time required for correction by shortening the correction movement distance of the XY stage) is shortened, and the work time can be shortened (speeded up). The correction origin remains constant until the 6th mask alignment, but when the 6th mask alignment is completed, since it corresponds to 2 ² after the correction amount is output, the moving average value from 3rd to 6th times and the previous correction origin By subtracting the correction amount from (from the first and second times), the correction origin from the next 7th time to the 15th time (7+2 ³ ) is corrected, and mask alignment can be performed. In other words, the modified origin is performed every 2 ⁿ times. Alternatively, after completing a certain number of times, it may be cleared and the process may be started again from the beginning.

According to such correction, it is thought that the error converges.

Here, let's consider the deviation with reference to FIGS. 4a to 4d. Displacement R from the center of the field of view when viewing the wafer transferred to the XY stage with a microscope
is the following formula.

R=Rα+Rβ+Rγ+R _P …(1) Rα: Misalignment between the visual pattern during primary printing and the identification surface of the wafer Rβ: Directional misalignment due to poor settings or mechanical wear in the pre-alignment mechanism Rγ: Misalignment during arm transfer R _q : Directional deviation due to poor adjustment of the X, Y stage and the center of the field of view R _p : Accidental deviation Rα + Rβ + Rγ in equation (1) is a value that slowly changes over a considerable period of time. It is possible that the value is almost constant over a short period of time. Its value is
almost
It is considered that _o-1 = _o-1 〓 ⁱ⁼¹ R _i /n-1 (average value of n-1 times), R _o − _o-1 = R _p ...(2), and the correction amount is i By subtracting =1/R _o-1 , the deviation can be reduced to only an accidental one.

Note that the present invention is not limited to the above embodiments. That is, by using two targets for alignment, the position can be corrected not only in the XY directions but also in the rotational direction (θ direction).

Furthermore, the alignment method of the present invention is applicable not only to mask alignment devices, but also to devices that require moving an object (sample) on an X, Y, θ table to a predetermined center of field of view using visual inspection or an automatic recognition device.
For example, it can be applied to a wide range of applications such as pattern printing equipment, wire bonders, wafer scribers, etc.

As described above, according to the positioning method of the present invention, the rough movement adjustment is automatically performed before the fine adjustment, so that the working time can be shortened and the working speed can be increased.

[Brief explanation of drawings]

FIG. 1 is an explanatory diagram showing an overview of a wafer alignment mechanism in a conventional mask aligner, FIG. 2 is a schematic diagram showing a mask alignment apparatus to which an embodiment of the alignment method of the present invention is applied, and FIG. 3 is a similar diagram. The flow chart and FIGS. 4A to 4D are explanatory diagrams showing the state of deviation at each position of the wafer. 1...Wafer, 3...Pre-alignment table, 4...XY stage, 5...Identification surface, 10...
...Vacuum suction tool, 13...Stage correction circuit.

Claims (1)

[Claims] 1. A positioning method for sequentially positioning a sample having a position detection target within a field of view of a specific range having a center, which includes: a) adjusting the attitude of the sample in a predetermined direction; b) an XY stage; c) placing the sample in the above step a on the XY stage in the above step b; d moving the XY stage in the above step c from the machine origin to the modified origin to detect the position a step of positioning the target within the field of view in the specific range, e a step of moving the XY stage in step d to position the target for position detection at the center of the field of view, f adjusting the amount of movement in the above step e for each sample. An alignment method comprising at least the step of cumulatively adding the corrected origin in step d, the origin position being corrected by referring to the average value of the cumulatively adding.