TWI657324B - Optimization method for finding system transfer parameters of auto-alignment equipments - Google Patents

Abstract

In order to solve the unit conversion between the visual system and the motion system of the automatic alignment device, the two-system unit conversion is not one-to-one, so it is necessary to find an appropriate optimization method for the system conversion parameters. The optimization method includes: defining the system conversion parameter as an experimental factor, setting the number of alignment iterations as an output value; selecting an appropriate orthogonal matrix and setting a stepwise adjusted proportional value; arranging the system to convert the parameter values into the orthogonal matrix; and sequentially converting the system The corresponding value of the matrix of the parameter is used as the input parameter to perform the online experiment; the visual system and the conversion parameter are used to generate the alignment motion value and perform the alignment, thereby obtaining the optimal iteration number and the corresponding system conversion parameter.

Description

System conversion parameter optimization method for automatic alignment device

The invention relates to a system conversion parameter optimization method for an automatic alignment device.

Over the years, many products have become lighter, thinner, shorter, and smaller. The precision and speed of these products have surpassed the limits that can be handled by humans. Most of them rely on the combination of visual modules and motion control modules. With the help of the "automatic alignment system", we can meet the strict requirements of fast and accurate processes. In order to improve the manufacturing precision, the automatic alignment technology has been paid more and more attention, but because it is combined with computer vision, alignment platform, alignment method, motion control, multi-coordinate system, coordinate transformation and system integration, we can know its technology. The threshold is very high, especially for device development in lithography processes.

The most important work initiated by the developer of an automatic alignment machine is to understand the specifications and characteristics of the vision system and motion control system (including screw, motor, drive, control), and then match the developed alignment method. Find the optimal system conversion parameters for a set of vision and motion control systems belonging to the machine. After obtaining the appropriate system conversion parameters, the actual production process can be entered to achieve the high performance of the automatic exposure machine.

In the case where an appropriate system conversion parameter cannot be obtained, the situation as shown in FIGS. 1A to 1C is often caused. 1A-1C are schematic diagrams showing target position-alignment time of various different alignment actions in the prior art. In FIG. 1A, although the position of the target object can gradually close to the target position, it takes a long time to complete. In Fig. 1B, the position of the target object converges in an oscillating manner, and can gradually move closer to the target position, and it takes a long time to complete the alignment. In Fig. 1C, the position of the target object is adjusted in a oscillating manner, but it is divergent and the alignment cannot be completed. Therefore, quickly finding quality conversion parameters has become an important issue in the field.

The invention provides a system conversion parameter optimization method, which can effectively and quickly find the optimal system conversion parameters.

The system conversion parameter optimization method of the invention comprises: defining a system conversion parameter as an experimental factor, setting a pair of bit iteration times as an output value; selecting a consistent matrix and setting a proportional value of the stepwise adjustment; and arranging the system conversion parameter to the orthogonal matrix And obtain the matrix of the system conversion parameters; sequentially use the corresponding values in the system conversion parameter matrix as the majority of the input parameters to perform the experiment on the straight line matrix; use the image system and each input parameter to generate the alignment motion The required value and the alignment action; check the alignment result of the alignment action through the image system, perform the orthogonal matrix alignment experiment at this stage, and determine whether the number of registration iterations is not greater than the demand value; and, if The number of bit iterations is not greater than the demand value, and the optimal number of iterations and the corresponding system conversion parameters are obtained.

In an embodiment of the invention, the alignment method further comprises: if the number of alignment iterations is greater than the demand value, using the Taguchi inference and the scale value, determining the system conversion parameter value of the next phase orthogonal matrix experiment, and re-straightening Matrix alignment experiment.

In an embodiment of the invention, the step of using the image system and each input parameter to generate a desired value of the aligning motion and performing the aligning action includes: providing a visual system to image the aligned object, according to spatial relative The coordinates and the input parameters are calculated to generate a first aligning displacement distance value to cause the motion system to generate motion and perform a aligning action.

In an embodiment of the invention, the alignment method further comprises: providing a vision system to determine whether the alignment error of the alignment action is less than a set value; and when the alignment error is not less than the set value, causing the vision system to be aligned The object image is taken according to the space relative coordinate and the output system conversion parameter group to generate the next alignment motion displacement distance value to cause the motion system to generate motion and perform the next alignment action.

In an embodiment of the present invention, if the number of alignment iterations is greater than the demand value, the Taguchi inference and the scale value are used to determine the system conversion parameter value of the next phase orthogonal matrix experiment, and the alignment of the orthogonal matrix is performed again. The experimental steps include: determining whether the average value of the number of registration iterations generated in the multiple alignment actions is less than the demand value; and, when the average value is greater than the demand value, determining the next stage according to the Taguchi inference and the proportional value The system of the orthogonal matrix experiment converts the parameter values for the orthogonal matrix alignment experiment.

In an embodiment of the invention, the system conversion parameter set includes a first axis conversion parameter, a second axis conversion parameter, and a third axis conversion parameter.

Based on the above, the present invention generates a matrix of system conversion parameters through the orthogonal matrix and the set adjustment ratio, and performs a straight-line matrix experiment by using a plurality of corresponding values in the matrix of the system conversion parameters to respectively serve as a plurality of input parameters. By checking the number of alignment iterations generated in the alignment action, the optimal number of iterations and the corresponding system conversion parameters can be obtained, and the automatic alignment action of the automatic alignment device is optimized.

The above described features and advantages of the invention will be apparent from the following description.

Please refer to FIG. 2. FIG. 2 is a schematic diagram of an automatic alignment device according to an embodiment of the present invention. The automatic registration device 200 includes vision systems 221, 222, motion systems 211, 212, and 213. The automatic registration device 200 adjusts the object OBJ through the motion systems 211, 212, and 213 to perform an alignment operation. In this embodiment, the vision system 221, 222 can image the alignment marks AK1, AK2 on the object OBJ through the mask MSK, and drive the motion systems 211, 212, and 213 to perform the object OBJ according to the result of the image capturing. Position adjustment. Wherein, the motion systems 211 and 212 can respectively adjust the vertical axis and the parallel axis of the object OBJ. The motion system 213 can rotate the object OBJ. There are a plurality of system conversion parameter groups between the vision systems 221 and 222 of the automatic alignment device and the motion systems 211, 212 and 213. In the case of a three-axis motion system, the system conversion parameter set has three system conversion parameters and corresponds to the parallel axis, the vertical axis, and the rotation axis on the automatic alignment device 200, respectively.

Incidentally, the object OBJ in the embodiment of the present invention may be a printed circuit board or an integrated circuit substrate, a wafer, a glass substrate, and related flat objects.

Referring to FIG. 3, FIG. 3 is a flowchart of a system conversion parameter optimization method according to an embodiment of the present invention. First, the system conversion parameter is defined as an experimental factor in step S310, and the system conversion parameter may have three conversion parameters fx, fy, and fq, and the conversion parameters fx, fy, and fz may respectively correspond to three different axes, and set the alignment iteration. The number of times is the output value. Step S310 is used to set a plurality of system conversion parameters between the visual system of the automatic alignment device and the motion system, and perform an alignment experiment according to the system conversion parameters.

It is worth mentioning that, in the embodiment of the present invention, the four measurement factors are directly related to the system conversion parameters, which are position errors (e1) and angles of the target position and the first alignment arrival point in the alignment process, respectively. Error (θ1), position error (e2) and angle error (θ2) of the first target position and the last arrival point. In order to reflect the nonlinear effects and collect sufficient data to represent the positioning performance of the automatic alignment device, a five-level full factor experiment can be used, and the system conversion parameters are set according to the experimental results.

Thereby, under the condition that the position error and the angle error generated during the alignment test experiment are satisfied, step S310 can set an appropriate system conversion parameter as an experimental factor.

Next, step S320 selects an appropriate orthogonal matrix, sets the adjustment scale value, and arranges the system conversion parameters into the orthogonal matrix. Wherein, the system conversion parameter group is selected according to the orthogonal matrix and the adjustment ratio value, thereby generating a matrix of system conversion parameters. And the matrix of system conversion parameters has a plurality of corresponding values. As for the details of the action, for example, the orthogonal matrix can be as shown in Table 1:

Table 1 1 1 1 1 2 2 2 1 3 3 3 2 1 2 3 : : : : : : : : : : : : : : : : : : : :

Of course, the orthogonal matrix is not limited as shown in Table 1. Those skilled in the art can select any suitable orthogonal matrix without any fixed restrictions.

In addition, the adjustment ratio value can be set, for example, to 0.02, and, for example, when the system conversion parameter (fx, fy, fq) = (0.06, 0.06, 0.14), corresponds to the orthogonal matrix of Table 1 and arranges the system conversion parameters to the orthogonal matrix. In the middle, a matrix of system conversion parameters with a plurality of corresponding values can be obtained, as shown in Table 2 below:

Table 2 : fx fy fq 0.06 0.06 0.14 0.06 0.08 0.16 0.06 0.10 0.18 0.08 0.06 0.16 : : : : : : : : : : : : : : :

Of course, the conversion parameter matrix shown in Table 2 is only an illustrative example, and is not intended to limit the invention.

Wherein, corresponding to the three rows of the orthogonal matrix, each column in Table 2 can be obtained by adjusting the system conversion parameters (fx, fy, fq) according to the adjusted scale value according to the values in the rows corresponding to the orthogonal matrix. For example, when the value in the corresponding orthogonal matrix is equal to 1, the corresponding value in Table 2 is equal to the corresponding value in the system conversion parameter; when the value in the corresponding orthogonal matrix is equal to 2, the corresponding value in Table 2 is equal to the system. The corresponding value in the conversion parameter plus the adjustment ratio value (0.02); when the value in the corresponding orthogonal matrix is equal to 3, the corresponding value in Table 2 is equal to the corresponding value in the system conversion parameter plus twice the adjustment ratio value ( 0.04).

Step S330, the system conversion parameter groups in the conversion parameter matrix are sequentially input to the multi-layer perceptual registration model to perform the test action of the online experiment, and the alignment error ratio is generated. Taking the matrix of the system conversion parameters of Table 2 as an example, step S330 may first make the corresponding value of the first column in the matrix of the system conversion parameters ((fx, fy, fq)=(0.06, 0.06, 0.14)) as an input value. Perform a straight line matrix experiment, then make the corresponding value of the second column ((fx, fy, fq)=(0.06, 0.08, 0.16)) as the input value to perform the experiment on the orthogonal matrix, and so on, so that in Table 2 A plurality of corresponding values are sequentially used as input parameters for performing orthogonal matrix experiments.

Please continue to refer to FIG. 3, in step S340, the vision system is imaged on the object to be aligned, the spatial relative coordinates are calculated, and the corresponding system conversion parameters are calculated, and the displacement displacement distance is generated and the alignment action is performed. Step S340 selects one of the plurality of corresponding values (system conversion parameters) according to the alignment error ratio generated by the experiment on the orthogonal matrix line executed in step S330, and makes the motion according to the selected system conversion parameter. The system produces motion and performs the alignment action. According to the embodiment of the present invention, the output system conversion parameter group can be generated according to the alignment error ratio value, and the aligned object is imaged by the visual system, and the space relative coordinate and the system conversion parameter group are selected to calculate the displacement displacement distance value. Make the movement system move and perform the alignment action.

It should be noted that if the number of registration iterations corresponding to the output of the alignment action in step S350 can be no more than a predetermined demand value, it can be determined that the selected system conversion parameter can meet the requirement. In contrast, if the number of registration iterations corresponding to the output of the alignment action in step S350 is greater than the predetermined demand value, it indicates that the selected system conversion parameter cannot meet the requirement.

To further optimize the alignment operation of the embodiment of the present invention, please refer to FIG. 4. FIG. 4 is a flowchart of a further implementation manner of a system conversion parameter optimization method according to an embodiment of the present invention. In step S410, the step S350 is continued, and the visual system can determine whether the registration error generated after the alignment operation is less than a set value. If the determination result is yes, it can be determined that the registration action is completed, and step S420 is performed; if the determination result is negative, step S350 is performed again, and then the vision system takes the image of the object to be aligned, and according to the spatial relative coordinates And the system conversion parameters are calculated to generate a new value of the displacement motion displacement distance to cause the motion system to generate motion and perform the alignment action.

In addition, step S420 determines whether the orthogonal matrix experiment has been completed. If the result of the determination is yes, step S430 may be performed; if the result of the determination is no, step S340 needs to be performed again.

In step S430, it is determined whether the number of iterations required by the selected system conversion parameter satisfies the requirement. In other words, whether the average value of the plurality of iterations in the plurality of alignment actions is less than a preset target value, and if so, the selected system conversion parameter is the optimal system conversion parameter, and is applied. The alignment operation is performed; if the result of the determination is negative, step S440 is performed.

In order to accelerate the generation of the optimal system conversion parameter, step S440 sets a new system conversion parameter according to the Taguchi theory and the scale value to perform the system conversion parameter value of the next-stage orthogonal matrix experiment, and re-executes step S330. The above-mentioned Taguchi theory was developed by Dr. Taguchi Genichi in the 1950s. Using simple orthogonal test design and simple analysis of variance numbers, analysis with a small amount of experimental data can effectively improve product quality. The field is well known to those skilled in the art and will not be described here.

In summary, the present invention integrates a full factor experimental design and Taguchi theory to find the optimal system conversion parameters of an automatic alignment device (eg, a wafer exposure machine), which is an effective, low cost, and time saving method. The purpose of the full-factor experiment design is to collect experimental data representing the positioning model of the exposure machine, while Taguchi theory is used to obtain better system conversion parameters more efficiently. Example  To illustrate the system conversion parameters found in the embodiments of the present invention.  For the requirement of positioning accuracy of ±5μm, the alignment efficiency can be improved by up to 55% compared with the existing method.  

Although the present invention has been disclosed in the above embodiments, it is not intended to limit the present invention, and any one of ordinary skill in the art can make some changes and refinements without departing from the spirit and scope of the present invention. The scope of the invention is defined by the scope of the appended claims.

200: automatic alignment device 221, 222: vision system 211, 212, 213: motion system S310~S350, S410~S450: steps of system conversion parameter optimization method OBJ: object MSK: mask AK1, AK2: alignment flag fx , fy, fq: system conversion parameters

1A-1C are schematic diagrams showing target position-alignment time of various different positioning actions in the prior art. 2 is a schematic diagram of an automatic alignment device according to an embodiment of the invention. FIG. 3 is a flow chart of a system conversion parameter optimization method according to an embodiment of the invention. 4 is a flow chart of a further implementation manner of a system conversion parameter optimization method according to an embodiment of the present invention.

Claims (6)

A system conversion parameter optimization method is applicable to an automatic alignment device, comprising: defining a system conversion parameter as an experimental factor, setting a pair of bit iteration times as an output value; selecting a constant matrix and setting a proportional value for stepwise adjustment; Arranging the system conversion parameters into the orthogonal matrix, and obtaining a matrix of system conversion parameters; sequentially, the corresponding values in the system conversion parameter matrix are respectively used as a plurality of input parameters to perform the experiment on the straight line matrix; An image system and each of the input parameters generate a desired value of the alignment motion and perform a one-bit action; check the alignment result of the alignment action via the image system, and perform the orthogonal matrix alignment experiment at this stage, And determining whether the number of times of the number of iterations is not greater than a demand value; and if the number of times of the number of iterations is not greater than the demand value, obtaining an optimal number of iterations and corresponding system conversion parameters. The system conversion parameter optimization method according to claim 1, further comprising: if the number of the iterations of the alignment is greater than the demand value, using the Taguchi inference and the ratio value, determining the next phase of the orthogonal matrix experiment. The system converts the parameter values and re-executes the orthogonal matrix alignment experiment. The system conversion parameter optimization method according to claim 1, wherein the step of using the image system and each of the input parameters to generate the required value of the alignment motion and performing the alignment operation comprises: providing a vision system Taking an image of the object to be aligned, calculating according to a spatial relative coordinate and each of the input parameters to generate a first alignment motion displacement distance value causes a motion system to generate motion and perform the alignment action. The method for optimizing a system conversion parameter according to claim 3, further comprising: providing the vision system to determine whether a pair of bit errors of the alignment action is less than a set value; and when the alignment error is not less than the setting When the value is obtained, the visual system is further imaged by the vision system, and the space relative coordinate and the output system conversion parameter group are calculated to generate the next alignment motion displacement distance value to cause the motion system to generate motion and perform The next alignment action. The system conversion parameter optimization method according to claim 2, wherein if the number of the iterations is greater than the demand value, the Taguchi inference and the ratio are used to determine the system conversion of the orthogonal experiment of the next stage. The parameter value, and the step of re-executing the orthogonal matrix alignment experiment includes: determining whether an average value of the number of registration iterations generated in the plurality of alignment operations is less than the required value; When the average value is greater than the demand value, the system conversion parameter value of the orthogonal experiment of the next stage is determined according to the Taguchi inference and the ratio value to perform the orthogonal matrix alignment experiment. The system conversion parameter optimization method according to claim 1, wherein the system conversion parameter set includes a first axis conversion parameter, a second axis conversion parameter, and a third axis conversion parameter.