JP2005116852A - Method for correcting distortion aberration and aligner using same - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To overcome the problem that a working indicated value including a measuring error is presented when an inclination of a non-spherical working surface includes the error caused by disorder of an examination pattern at each point when working the non-spherical surface. <P>SOLUTION: A table indicating how a distortion changes is used when working a shape that can be represented with a Zernike function. On the basis of the table, the shape of the non-spherical surface is optimized. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a semiconductor manufacturing apparatus that exposes and transfers an electronic circuit pattern onto a wafer in a projection optical system, and a method for correcting distortion occurring therein.

  The progress of semiconductor technology has been increasing in recent years, and the progress of microfabrication technology has been remarkable. In particular, optical processing technology using a projection exposure apparatus in semiconductor manufacturing, which is at the center of the process, has stepped into a region near half the exposure wavelength due to advances in super-resolution technology (RET) and lens manufacturing technology.

Synchronously with the improvement of the resolution, the allowable amount of overlay has also decreased rapidly, and 10 nm or less is required in the screen. Therefore, for example, as proposed in Japanese Patent Laid-Open No. 11-031652 by the present applicant, the distortion is corrected by arranging a plate near the reticle and performing aspherical processing. (See Figure 1)
A conventional correction method using aspherical processing is as follows. Since a plate for correcting distortion (hereinafter referred to as a correction plate) is disposed relatively close to the reticle, the light flux on the surface subjected to aspherical processing for correcting distortion (hereinafter referred to as aspherical processing surface). Is relatively thin. For this reason, it can be handled almost the same as the principal ray. Therefore, the distortion is controlled by changing the direction of the chief ray so that the distortion does not occur on the wafer or has a predetermined shape. For example, as shown in FIG. 7, when the distortion at a certain point on the wafer is Δ, the inclination of the portion corresponding to that point on the aspherical surface for correcting it is

Θ satisfying At this time, m is the magnification of the projection lens, D is the distance between the mask pattern surface and the aspherical surface, and n is the refractive index of the correction plate. The machining shape is calculated by calculating the inclination at each inspection point of distortion and connecting them using a function such as a two-dimensional spline so as to become a plane.
Japanese Patent Laid-Open No. 11-031652

However, when calculated from the measured values at each point as described above,
(1) A planar shape that always satisfies the inclination at each point is not always obtained.

(2) If there is a measurement error due to the inspection pattern displacement at each point in the inclination of the aspherical processing surface, a processing instruction value is generated including it.
There is a problem.

As a method to create a processing instruction value with a surface shape that can be reliably processed and average the measurement error of each inspection point,
(1) A plurality of specific surface shapes that can actually be processed on an aspherical processed surface, in this case, a function that can be expressed by a function (for example, one in which each term is orthogonal such as a Zernike polynomial is preferable, ), The amount of distortion at each inspection point is calculated or actually created and incorporated into the device, and the result is measured to obtain the distortion per unit surface shape. Make a table.

(2) Calculate the aspherical surface shape by adding each of the surface shapes tabulated above multiplied by a coefficient to reduce distortion (optimal solution such as minimizing maximum value, minimizing 3σ, etc.) .
I propose a method. As a result, because the surface shapes that can actually be processed are added together, a realistic surface shape can be obtained, and because optimization is performed with reference to all points, high accuracy is achieved regardless of measurement errors at certain locations. Distortion can be corrected.

  By these methods, it is possible to create a correction plate that reduces the measurement variation of each inspection point and the influence of abnormal points. In addition, since the surface shape can be instructed reliably, stable distortion correction accuracy can be obtained.

  First, a method for creating an aspherical processing instruction value in this embodiment will be described.

  The aspherical surface is set near the reticle as shown in FIG. For example, when 4 terms (quadratic function), 9 terms (quadratic function), and 16 terms (sixth function), which are rotational symmetry terms of the Zernike coefficient, are created on this aspherical surface, a 22 mm square is formed on the wafer. Calculate what the distortion shape will be when the image is baked in the exposure area. (A) to (e) of FIG. 2 represent the shapes when the inspection points are 5 × 5 in total 25 points in a vector diagram. FIG. 3 is a table in which the table is converted into the X direction and the Y direction.

  Since the distortion shape is differential with respect to the surface shape, fifth-order distortion can be generated with a surface shape of a sixth order function.

The distortion aberration is first inspected in the state of a plane before performing aspherical processing. At this time, it is assumed that the distortion aberration R as shown in FIG. 4 remains. R is an array of (x1, x2, x3,..., X25, y1, y2, y3,..., R25). x1 to x25, and y1 to y25 represent the X component and Y component of the distortion aberration on the 25 points at which the distortion aberration is inspected. In order to correct this, the Zernike 4 term, 9 term and 16 term arrays arranged in the table as shown in FIG.
Z4 = (x4_01, x4_02, x4_03,… x4_25, y4_01, y4_02, y4_03,…, y4_25) (2)
Z9 = (x9_01, x9_02, x9_03, ... x9_25, y9_01, y9_02, y9_03, ..., y9_25) (3)
Z16 = (x16_01, x16_02, x16_03,… x16_25, y16_01, y16_02, y16_03,…, y16_25) (4)
The vector
R- (a × Z4 + b × Z9 + c × Z16) = Rc (5)
Then, a, b, and c that reduce Rc are obtained. There are various ways to reduce Rc, such as minimizing the maximum value of each element of Rc, or reducing 3σ of all elements, but it is possible to use different methods in each case. In this embodiment, the maximum value minimization of each element is calculated, and the result is shown in FIG.

  In the optimization calculation, since (5) is linear, a good optimal solution can be obtained by performing optimization by linear programming.

  After the values of a, b, and c determined in this manner are applied to each Zernike term, it is possible to calculate an indication shape as to whether the shape of addition is changed from the current shape of the aspherical surface. In this method, since the sizes of the already-defined shapes such as the zernicke term and the 9th term are changed and added together, the shape actually exists. Even if a random measurement error is placed on each inspection point, the fitting is made to the Zernike 4-term shape or the 9-term shape, so that the influence can be avoided.

  On the other hand, if this is a method of connecting the inclination of each point with a spline function or the like, there are cases where the shape cannot be connected with a desired inclination, and in severe cases, the points cannot be connected and processing cannot be performed.

After the aspherical surface processing, the transmitted wavefront of the processed workpiece is measured with an interferometer. The aspherical shape can be checked by interference due to surface reflection, but the reason for making it a transmitted wavefront is
(1) The transmitted wavefront can easily remove the system offset of the interferometer. (2) Interference due to surface reflection is because noise tends to ride on the interference fringes.

  After the transmitted wavefront measurement, it is determined whether or not the wavefront shape is a desired one. Here, a simulation is performed to determine how the distortion changes from the wavefront shape, and it is determined whether or not to proceed to the next step depending on how much the distortion becomes.

  After it is determined that the aspherical surface processing is completed, it is cut into a predetermined shape and coated. Thereafter, the mechanical part is bonded or incorporated into the mechanical part and mounted on the exposure apparatus. This makes it possible to correct distortion.

  In the above embodiment, only the Zernike terms 4, 9, and 16 are used for explanation, but the actual distortion does not remain only in rotationally symmetric components. Therefore, for example, when creating a shape using a Zernike polynomial, the array of Z4, Z9, etc. described above is obtained over all necessary terms, and incorporated into the equation as shown in equation (5). Apply. As a result, an asymmetric aspherical shape can be easily formed, and highly accurate distortion correction can be performed.

  Furthermore, this method not only corrects distortion due to manufacturing errors, but also corrects distortion remaining as a lens design value.

  In addition, distortion aberration caused by deflection of the reticle and drawing errors of the reticle can be taken.

  FIG. 6 is a diagram in which the aspherical surface of the present invention is mounted on a scanner, not a stepper. In the case of a scanner, distortion is affected not only by the lens but also by the habit of the stage. Therefore, if the aspherical correction plate is moved with respect to the lens together with the reticle scanning stage 1, the distortion caused by the stage running, the deflection of the reticle, etc. can also be corrected by the above-described distortion correction aspherical surface creation procedure. It can be corrected by obtaining and processing a spherical shape.

  When the distortion changes due to some reason after the correction plate according to the present invention is mounted on the exposure apparatus, the change can be additionally processed if the correction plate is removable.

  In addition, when aspherical processing is performed, the correction plate when the data for creating the processing instruction value is collected. The aspherical surface is formed by another member based on the data, and the aspherical surface is formed on the device after completion. It is also possible to replace it with the correction plate used when collecting data. In this way, the trouble of removing the correction plate and the processing process before the aspherical surface processing for the correction plate can be omitted (because a separate member can be processed in parallel with the inspection), the process can be shortened. However, at that time, it is necessary to consider the difference in the transmitted wavefront before processing of the correction plate that was originally included and the correction plate that has been subjected to new aspherical processing for replacement. If not taken into account, the difference becomes a distortion residual correction.

  Moreover, it is necessary to consider the thickness of both. If the thickness is different, the optical path length is changed accordingly, and the magnification component and the distortion component may change in proportion to the optical path length difference. Therefore, when the magnification or distortion changes due to the optical path length difference, it is necessary to add the amount generated by the optical path length difference to the aspherical processing instruction value or cancel it with the correction function of the exposure apparatus.

  Next, as another embodiment, a method for considering the characteristics of the processing machine when calculating the aspheric processing instruction value will be described.

This embodiment will also be described based on the previous embodiment. The distortion aberration is first inspected in the state of a plane before performing aspherical processing. At this time, it is assumed that the distortion aberration R as shown in FIG. 4 remains. R is an array of (x1, x2, x3,..., X25, y1, y2, y3,..., Y25). x1 to x25, and y1 to y25 represent the X component and Y component of the distortion aberration on the 25 points at which the distortion aberration is inspected. In order to correct this, the Zernike 4 term, 9 term and 16 term arrays arranged in the table as shown in FIG. 3 are made into vectors of (2), (3) and (4), respectively. At this time, as the characteristics of the aspherical surface processing machine, the processing difficulty level for each Zernike term may be different. Therefore, in order to take this characteristic into consideration, there is a limit on the value that each term can take. For example, in a, b, and c of equation (5),
−W4 ≦ a ≦ + W4 (6)
−W9 ≦ b ≦ + W9 (7)
−W16 ≦ c ≦ W16 (8)
A, b, and c that reduce Rc in (5) are obtained. W4, W9, and W16 represent constraint values for each term. There are various ways to reduce Rc, such as minimizing the maximum value of each element of Rc, or reducing 3σ of all elements, but it is possible to use different methods in each case.

  In the optimization calculation, since (5) is linear, an optimal solution can be obtained by performing optimization by linear programming.

  In the embodiments described so far, it has been described that the magnification component such as the Z4 term is also aspherically processed. However, the component that can be corrected by the aberration correction mechanism mounted on the exposure apparatus is corrected there, The amount that cannot be removed may be corrected by aspherical processing.

  Further, the present invention is not limited to the above-described embodiment, and can be applied to a device manufacturing method using the exposure apparatus described in this embodiment. Specifically, using the above-described exposure apparatus, a step of exposing an object to be exposed such as a wafer or a glass substrate, a step of developing the exposed object to be exposed, and a step of further developing the exposed object to be developed A method for manufacturing a device.

Correction plate of the present invention and exposure apparatus equipped with the same Distortion processing sensitivity on aspherical surfaces Distortion processing sensitivity table on aspherical surface Before distortion correction After distortion correction Scanner with correction plate Principle of distortion correction

Claims (9)

  A residual aberration correction plate for correcting a residual aberration of the photographic optical system by providing a pattern on the first object plane in an optical path of a photographic exposure apparatus that shoots and exposes a second object plane by a photographic optical system; In the correction method for correcting the residual aberration by performing aspheric processing on the correction plate, the aspheric shape processed on the correction plate is divided into a plurality of shapes that can be fitted with a specific function in advance, and each component The amount of change in residual aberration is obtained, and the difference between the linear sum of each component and the amount of residual aberration, or the ratio of each component is calculated so that the sum is minimized with respect to the residual aberration amount to be corrected. A residual aberration correction method in which the linear sum of the unit shapes is an aspheric surface processing value.   The correction method according to claim 1, wherein the function for fitting the aspherical shape is a Zernike polynomial.   The correction method according to claim 1, wherein the difference between the linear sum of each component and the residual aberration amount, or a method for obtaining a minimum value of the sum is linear programming.   The correction method according to claim 1, wherein the difference between the linear sum of the respective components and the residual aberration amount, or the minimum value of the sum is a value obtained by minimizing the maximum absolute value of the measurement point of the residual aberration after correction.   The correction method according to claim 1, wherein the difference between the linear sum of the respective components and the residual aberration amount, or the minimum sum is a value obtained by minimizing 3σ of the measurement point of the residual aberration after correction.   The correction method according to claim 1, wherein a restriction is set on a ratio of each component.   A residual aberration correction plate for correcting a residual aberration of the photographic optical system by providing a pattern on the first object plane in an optical path of a photographic exposure apparatus that shoots and exposes a second object plane by a photographic optical system; The correction plate is subjected to aspheric processing for correcting residual aberration, and the corrected aspheric shape is divided into a plurality of shapes that can be fitted with a specific function in advance, and the residual aberration for each component. The amount of change is calculated, and the difference between the linear sum of each component and the amount of residual aberration, or the ratio of each component is calculated so that the sum is minimized, and the ratio and the unit shape of each component. An exposure apparatus equipped with a correction plate having a shape obtained by linear summation.   The exposure apparatus according to claim 7, wherein the correction plate is detachable. 9. A device manufacturing method, comprising: a step of exposing an object to be exposed using the exposure apparatus according to claim 7; and a step of developing the exposed object to be exposed.

Images