JP2007024945A - Alignment error detection method, and exposure processing method using the same - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an alignment error detection method detecting an error with high accuracy, and also to provide an exposure processing method using the detection method. <P>SOLUTION: A plurality of measurement points are measured to obtain a measurement deviation from the measurement result. Then a characteristic parameter representing characteristics of a substrate is obtained by using a least square method from the obtained measurement deviation. Then a calculation deviation which is a deviation by calculation at each measurement point is obtained from the obtained measurement deviation and the obtained characteristic parameter. An alignment error at each measurement point is detected, using the error between the obtained measurement deviation and the calculation deviation. Specifically, if the error is out of a predetermined range based on EGA (Enhanced Global Arrayment) superposition accuracy, it is judged that a measurement error is present. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an alignment error detection method and an exposure processing method using the alignment error detection method, and more particularly to a technique for increasing error detection accuracy.

  An alignment method in a conventional projection exposure apparatus is disclosed in, for example, Patent Document 1. As a conventional exposure apparatus, a step-and-repeat exposure apparatus (hereinafter referred to as a stepper) is manufactured and sold by Nikon Corporation, for example.

  The stepper accurately measures a plurality of coordinate positions from an arbitrary number of alignment marks formed in advance on a substrate such as a glass substrate, and calculates a deviation from the stage coordinates, that is, design coordinates (XDi, YDi). . By statistically processing the measurement results, scaling (linear expansion / contraction) Mx, My in each of the X and Y directions, rotation θx representing a deviation along the Y axis, rotation θy with respect to the X axis, and each of the X and Y directions Substrate characteristic parameters such as offsets .DELTA.X and .DELTA.Y are obtained.

  That is, in the equation (1) represented by the following linear equation (determinant), the characteristic is such that the sum of squares of the error between the calculated deviation (Xi, Yi) and the measured value (Xri, Yri) is minimized. Parameters (Mx, My, θx, θy, ΔX, ΔY) are obtained (least square method). The calculated characteristic parameter is used as a correction value (reverse sign) at the time of alignment, and exposure is performed.

  Of the alignment marks formed on the substrate, the coordinate position must be accurately measured to calculate the amount of deviation from the design coordinates (stage coordinates). This measurement method uses a laser system and a CCD camera. Image processing method.

  Regardless of which method is used, misrecognition of the alignment mark may occur in this measurement (the misrecognition mentioned here excludes the case where the alignment mark is broken and cannot be recognized at all). When such misrecognition occurs, accurate measurement cannot be performed and an erroneous shift amount (measurement error) is calculated. As a result, since exposure is performed with an incorrect correction value, it is necessary to perform re-exposure or cause a product defect.

  In a general stepper, the user can set an allowable value for each measured value (Xri, Yri) and each obtained characteristic parameter (Mx, My, θx, θy, ΔX, ΔY). is there. In other words, when a preset allowable value is exceeded, a warning is given as an error. Hereinafter, the allowable value set for the measured value is referred to as a first allowable value, and the allowable value set for the parameter is referred to as a second allowable value.

  Examples of conventional alignment error detection methods are disclosed in Patent Documents 2 to 5, for example.

Japanese Patent Application Laid-Open No. 11-195591 JP 2000-164478 A JP 2001-203247 A JP 2002-110518 A JP 2000-114150 A

  In Patent Documents 2 to 4, as described above, an alignment error is detected using a preset first tolerance or second tolerance. However, for example, when the scalings Mx and My are large, the measured values are different values. Therefore, it is difficult to provide an appropriate first allowable value that can detect an arbitrary measurement error in the measurement value.

  Further, in the parameters such as scaling Mx, My, etc., it is possible to provide a second allowable value, but the scaling Mx, My is greatly influenced by the expansion / contraction and warpage of the substrate and varies depending on the process and process condition difference. It is not always constant. Therefore, it is difficult to set the second tolerance value appropriately. Therefore, there is a problem that the error detection accuracy is lowered.

  In Patent Document 5, an alignment error is detected by setting an allowable value for an error between each measured value and a calculated coordinate. However, in the method disclosed in Patent Document 5, appropriate detection is difficult when using a process that exposes a glass substrate to a high temperature. Therefore, in such a case, there is a problem that the error detection accuracy is lowered.

  The present invention has been made to solve the above problems, and an object of the present invention is to provide an alignment error detection method capable of highly accurate error detection and an exposure processing method using the same.

  An alignment error detection method according to the present invention includes a measurement deviation amount obtaining step for obtaining respective measurement deviation amounts measured at a plurality of measurement points on a substrate, and a minimum from each measurement deviation amount obtained in the measurement deviation amount obtaining step. A calculation parameter that is a calculation error amount at each measurement point using a characteristic parameter calculation process that obtains a characteristic parameter that represents the characteristics of the substrate using the square method, and a linear expression based on the characteristic parameter obtained in the characteristic parameter calculation process. An error that detects an alignment error at each measurement point based on the error between the calculation deviation amount calculation process for obtaining the amount, and the error between the measurement deviation amount obtained in the measurement deviation amount calculation process and the calculation deviation amount obtained in the calculation deviation amount calculation process. A detection step.

  An alignment error detection method according to the present invention includes a measurement deviation amount obtaining step for obtaining respective measurement deviation amounts measured at a plurality of measurement points on a substrate, and a minimum from each measurement deviation amount obtained in the measurement deviation amount obtaining step. A calculation parameter that is a calculation error amount at each measurement point using a characteristic parameter calculation process that obtains a characteristic parameter that represents the characteristic of the substrate using the square method and a linear expression based on the characteristic parameter obtained in the characteristic parameter calculation process. An error that detects an alignment error at each measurement point based on the error between the calculation deviation amount calculation process for obtaining the amount, and the error between the measurement deviation amount obtained in the measurement deviation amount calculation process and the calculation deviation amount obtained in the calculation deviation amount calculation process. A detection step. Therefore, highly accurate error detection can be performed.

<Embodiment 1>
Below, the process of forming patterns, such as TFT, on a glass substrate using a projection exposure apparatus in the manufacturing process of a liquid crystal panel is demonstrated. The present invention is particularly effective when using a process in which a glass substrate is exposed to a high temperature of about 400 ° C. to 600 ° C. Since the present invention has no characteristics in the TFT formation process itself, a detailed description of the process will be omitted below.

  First, on the glass substrate (or on the insulating film formed on the glass substrate), for example, silicon to be a first layer pattern of TFT is formed by CVD or the like.

  Next, in the photolithography process, a desired pattern is formed by mask exposure (hereinafter simply referred to as exposure) using a stepper so as to have a designed arrangement. In the first exposure, alignment is not performed, and apart from the desired pattern, an alignment mark (hereinafter simply referred to as a mark) to be a mark to be aligned is formed simultaneously in the subsequent steps.

  In most cases of TFT formation, contacts must be formed by exposure on the gate electrode, the wiring electrode, and the interlayer film that insulates the gate electrode, the wiring electrode, and it is necessary to accurately overlay at this time. . For example, when the gate electrode is exposed while accurately overlaying the pattern on the first layer, multiple marks formed on the first layer are measured by the EGA (Enhanced Global Arrayment) method and statistical After calculating the amount of deviation from the processing, exposure is performed using a correction value based on the amount of deviation. The above series of operations is an existing calculation process performed in an existing stepper and is not particularly new.

  The exposure is performed in the same manner when forming contacts and wiring electrodes. When it is particularly important to superimpose with another pattern, a new mark may be formed simultaneously with the pattern and alignment may be performed using this mark.

  As described above, alignment is performed at the time of exposure for TFT formation. Further, the TFT has a metal film processed on a gate electrode, a wiring electrode, and the like, and an interlayer insulating film for insulating these is also necessary.

  For example, a substrate that has undergone many steps is contaminated or scratched during the process. As a result, when a part of the mark is lost, the mark may not be correctly recognized by the laser or the CCD camera during alignment. In other words, if the mark cannot be recognized at all, it will be recognized as an alignment error by the stepper, so no problem will occur.However, if a part of the mark is missing, the stepper will not recognize it as an alignment error. Since erroneous recognition (measurement error) occurs, it becomes a problem. This measurement error can occur only in the X direction, only in the Y direction, or in both the X and Y directions.

  As described above, in the conventional alignment error detection method, for such a measurement error, a tolerance value (first tolerance value) is set, and those exceeding the first tolerance value are excluded from statistical processing. Is adopted.

  However, for example, an interlayer insulating film such as an oxide film or a nitride film has different and large film stresses due to thermal history such as film formation, so that the substrate expands and contracts or warps.

  FIG. 1 shows measurement values at measurement points 1 to 13 on the substrate, calculation deviations based on characteristic parameters, and errors between measurement values and calculation deviations. This is only under one condition, and varies depending on the film material, thermal conditions, and the process surrounding them.

  For example, with respect to the result of FIG. 1, the final overlap amount (hereinafter referred to as the residual amount) when a correction value is calculated and exposed is expressed by an average value (Ave) + variation (3σ). In many cases, x = 0.32 μm and y = 0.14 μm.

  This amount of residual deviation falls within the accuracy inherent to the stepper device. The apparatus-specific accuracy is called EGA overlay accuracy, and is determined from the horizontal and vertical movement accuracy, stop position accuracy, repeat accuracy, and alignment sensor measurement reproducibility of the stage holding the substrate. Usually guaranteed by the equipment manufacturer.

  On the other hand, in FIG. 2, the measurement point 5 in FIG. 1 is a measurement error. However, it is impossible to determine which measurement point is a measurement error only by the measurement values (Xri ′, Yri ′).

  For the above reasons, it is difficult to detect one or a plurality of measurement errors from a plurality of measurement values with respect to the alignment of the substrate having the characteristics as described above. The present invention provides an alignment error detection method for solving such problems.

  In the present invention, as for the alignment measurement, a plurality of measurement points (marks) are measured as in the prior art, and the amount of measurement deviation is obtained from all these measurement results. Then, the characteristic parameters (Mx, My, θx, θy, ΔX, ΔY) representing the characteristics of the substrate in Expression (1) are obtained from the obtained measurement deviation amount using the least square method. Then, from the obtained measurement deviation amount and the obtained characteristic parameter, a calculation deviation amount that is a calculation deviation amount at each measurement point is obtained using Equation (1).

  In the case of FIG. 1 with no measurement error, the error between the measured value (Xri, Yri) and the calculated deviation (Xi, Yi) obtained by the least square method is minimized.

  However, the error at the measurement point 5 (measurement error) in FIG. 2 is clearly large. Naturally, the characteristic parameters (Mx ′, My ′, θx ′, θy ′, ΔX ′, ΔY ′) obtained by the least square method are different from those when no measurement error is included. The error also increases at other measurement points inside.

  In the alignment error detection method according to the present invention, an alignment error at each measurement point is detected based on the error between the obtained measurement deviation amount and the calculated deviation amount. Specifically, if this error is not within a predetermined range based on EGA overlay accuracy (stage alignment accuracy of the exposure apparatus), it is determined that a measurement error has occurred.

  In the present invention, this error range is determined as shown in the following formulas (2) to (3) using the coefficient k, thereby detecting an error.

  As shown in the equation (3), the coefficient k needs to be set to a value of 1 or more so that the error range does not fall below the EGA overlay accuracy. That is, if the coefficient k is made smaller than 1 to narrow the error range, the positioning accuracy of the apparatus cannot catch up. In addition, the measurement error detection accuracy decreases as the coefficient k increases, and as a result, the amount of residual deviation increases. Therefore, first find a setting that matches the product standard and management level when the coefficient k is 1.8 or less. is required. In the case of our product, k = 1.6 corresponds to the upper limit specification of the management level of the residual deviation amount. Therefore, the error range is set to k = 1.4 in consideration of the margin.

  FIG. 3 shows a trend of scaling Mx and My among the characteristic parameters subjected to the alignment statistical processing for one process (FIG. 3A) and another process (FIG. 3B) in our production line. Is. The unit of the vertical axis is ppm, and the horizontal axis represents the processing history for each substrate.

  In FIG. 3, there are obvious abnormal points (there are measurement error locations) in the portion surrounded by the dotted circle, but these are exposed with incorrect correction values, so that redoing is necessary. Become. This time, by setting k = 1.4 and applying the present invention, it is possible to detect the measurement error location that is the cause of all these abnormal points.

  In order to perform accurate statistical processing in the exposure processing method using the alignment error detection method according to the present invention, if there are a sufficient number of measurement points, an algorithm that recalculates statistical processing that excludes only measurement errors is used.

  The residual deviation when the correction value is calculated and the exposure is performed without removing the measurement error portion is x = 0.70 μm and y = 0.14 μm, which may cause a product trouble. . When the correction value is calculated for the remaining portion by eliminating the measurement error portion, the residual deviation amounts are x = 0.32 μm and y = 0.15 μm, which is almost the same as the result without the measurement error. .

  In addition, when the number of measurement points is small or the number of measurement errors increases, the error in the statistical processing itself increases as the data used in the calculation decreases.Therefore, the algorithm is re-measured and the statistical processing calculation is performed again. To do. Alternatively, it can be recognized as an alignment error between the apparatus side and the operator side. In an exposure apparatus equipped with such an algorithm, a measurement error can be extracted from a plurality of measurement points during alignment, and statistical processing can be accurately performed. Thereby, accurate overlay exposure with an accurate correction value can be performed.

  As described above, in the alignment error detection method and the exposure processing method using the alignment error detection method according to the present embodiment, the alignment error at each measurement point is detected based on the error between the measurement deviation amount and the calculation deviation amount. Therefore, highly accurate error detection can be performed.

It is a figure which shows the difference | error of calculation based on a measured value and a characteristic parameter in each measurement point which concerns on Embodiment 1, and the difference | error of a measured value and calculation deviation. 6 is a diagram including measurement errors at each measurement point according to Embodiment 1. FIG. 6 is a graph in which scalings Mx and My subjected to alignment statistical processing according to the first embodiment are trended.

Claims (4)

A measurement deviation amount obtaining step for obtaining measurement deviation amounts respectively measured at a plurality of measurement points on the substrate;
A characteristic parameter obtaining step for obtaining a characteristic parameter representing the characteristic of the substrate using a least square method from each of the measured deviation amounts obtained in the measurement deviation amount obtaining step;
A calculation deviation amount obtaining step for obtaining a calculation deviation amount which is a calculation deviation amount at each measurement point using a linear equation based on the characteristic parameter obtained in the characteristic parameter obtaining step;
Alignment comprising: an error detection step for detecting an alignment error at each measurement point based on an error between the measurement deviation amount obtained in the measurement deviation amount obtaining step and the calculation deviation amount obtained in the calculation deviation amount obtaining step Error detection method. The alignment error detection method according to claim 1,
An alignment error detection method in which, in the error detection step, the error is detected as an alignment error when the error is not within a predetermined range based on EGA (Enhanced Global Arrayment) overlay accuracy. The alignment error detection method according to claim 2,
The substrate is a glass substrate;
Further comprising a step of treating the glass substrate at 400 to 600 ° C .;
An alignment error detection method in which, in the error detection step, the error is detected as an alignment error when the error is larger than 1.8 times the EGA overlay accuracy. An exposure processing method using the alignment error detection method according to any one of claims 1 to 3,
An exposure processing method comprising a step of performing statistical processing by removing a measurement point where an alignment error is detected in the error detection step from the plurality of measurement points.

Images