JPH059934B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an imaging optical device, and more particularly to an optical device capable of adjusting imaging magnification error. The present invention is particularly effective for correcting lateral magnification errors that occur in precision optical systems that project and image semiconductor integrated circuit patterns onto wafers.

In recent years, the demand for miniaturization of semiconductor integrated circuits has been rapidly increasing, and the dimensions of micropatterns in electronic circuits are approaching 1 μm. One of the processes for manufacturing such highly integrated devices is a photo process in which a mask pattern is transferred onto a wafer.

A so-called semiconductor printing device is a device used for transferring such a mask pattern, and there are several methods thereof. For example, there are contact exposure methods in which the mask and wafer are printed while being in contact with each other, close exposure methods in which the mask wafer is printed several micrometers apart, and optical projection exposure methods in which the masks are printed using a lens or a muller.

Generally, semiconductor integrated circuits do not have a single layer structure;
As the degree of integration increases, there is a tendency for the number of layers to increase accordingly. Therefore, in the photo process, it is necessary to selectively use the above-mentioned several printing methods to repeatedly print different mask patterns for each layer on a single wafer with high dimensional accuracy. On the other hand, wafer diameters are being increased in order to increase circuit productivity. Nowadays, 5" diameter ones are becoming mainstream.

However, when transferring a mask image to such a large wafer, due to a combination of various factors,
No matter how many alignment operations are performed, development sometimes occurs in which the desired positional relationship between the two cannot be followed down to a microscopic amount of 1 μm or less. In other words, the mask image and the existing pattern on the wafer deviate from the normal relationship.

There are a wide variety of ways this lateral displacement occurs, but
Most of this is what is called magnification error.
Magnification error can be expressed as a linear function of position. However, if this magnification error occurs, the image of the mask on the wafer will be enlarged or reduced from the normal size and transferred, as shown in Figures 1A and B, and as a result, it will not be able to combine well with the existing pattern, resulting in defects. It will be a good product.

In general, when we consider the causes of magnification errors, firstly, there is a difference in magnification of the mask image that depends on the difference in the projection method mentioned above, and secondly, there is a difference in magnification that is unique to each device even when using the same method. It will be done. In addition, expansion and contraction of the wafer due to temperature changes during the processing process can also cause magnification errors.

Introducing errors into the mask image is an inconvenience in any method, but it is more serious in the mirror projection exposure method. Therefore, this method will be explained.

The mirror projection exposure method referred to here is a method in which a mask pattern is printed onto a wafer using an off-axis annular region in which aberrations are corrected in an optical system mainly composed of mirror surfaces. In this method, the good image area of the optical system is annular. Therefore, in order to print a large-area mask onto a wafer, the mask M and the wafer W are scanned synchronously in a direction perpendicular to the direction of the annular zone (T in FIG. 2) projected onto the wafer.

In order to specifically illustrate the above configuration, a configuration diagram is shown in FIG. 3 as an example. In the figure, 1S is the illumination system, and AS is the mask and wafer alignment system. PS is a mirror optical system, by which the pattern of the mask M is transferred onto the wafer W.

The mirror optical system PS is an axially symmetrical system with respect to its optical axis OO', and aberrations are corrected at the image height R. Then, the fine pattern on the mask is transferred onto the wafer using a ring-shaped area including a part of the circumference of the radius R, which is the good image area. Also, regarding imaging,
The light beam diverged from one point on the mask M is a plane mirror.
After reflecting at BS1, convex mirror which is concave mirror M1 aperture
M2, then reflected again by concave mirror M1, and then reflected by plane mirror
The optical path is bent by BS2 and it converges on one point on the wafer W.

The magnification error that occurs in this device can be divided into two types: one is caused optically in the projection optical system, and the other is caused by mechanical feed errors when scanning the mask and wafer.As will be explained later, the difficulty of the latter is It has been resolved.

On the other hand, the following reasons can be considered for the former.

As mentioned earlier, with the increase in wafer diameter, 5-inch diameter wafers are currently being used.
In the mirror projection exposure method, in order to print a wafer of this size, concave mirror M1 must have a diameter of 400 mm.
It will be close in size. While responding to the increase in the diameter of mirrors, in order to faithfully transfer fine patterns of nearly 1 μm onto wafers, high precision of less than 1/10 of the wavelength of surface distortion after mirror polishing is required. must be kept to a minimum. This is because, as shown by the dotted line in FIG. 3, if there is a surface distortion on the mirror surface at the position where the light beam hits, the light beam will deviate from the direction in which it is supposed to be reflected. As a result, the image is formed on the wafer at a position that is changed from the specified position in a direction perpendicular to the optical axis, resulting in a magnification error in the image.

In a scanning mirror optical system, this type of error occurs mainly in a direction perpendicular to the scanning direction y on the wafer, and this is schematically shown by a small arrow in FIG. 4A. On the other hand, dimensional errors caused by the second mechanical cause occur as the mask wafer moves, and occur mainly along the scanning direction y, as depicted in Figure 4B. do.

Therefore, if these two magnification errors are removed independently, the accuracy of overprinting integrated circuits over the entire wafer surface is improved, and it becomes possible to use the printer in combination with other printing devices.

In particular, this mirror optical system is telecentric on both the object side and the image side, and has an optical special feature in that the imaging magnification does not change at all even if the position of the object mask or wafer in the optical axis direction changes. have. Therefore, the magnification error in the scanning direction and the vertical direction is determined by the optical system, and no means for correcting this is known.

On the other hand, the main mechanical causes of magnification errors are thought to be poor machine assembly and machining accuracy. For example, if a sliding pair K on which a mask and a wafer are placed is moved along a guide surface G while being supported by hydrostatic gas bearings b ₁ and b ₂ as shown in FIG. If it had an upwardly convex shape, the sliding body K would be inclined before and after the scanning range. Therefore, the wafer surface will also be tilted, which is shown in Figure 4B.
This will result in an error of the form shown below. Therefore, one way to prevent this is to increase the pressure supplied to bearing b ₁ at the front end of the scanning range, and increase the pressure supplied to bearing b ₂ at the rear end, so that the sliding body K always moves horizontally. It is best to control the pressure accordingly.

In addition to this, there is also a method of removing the magnification error by controlling the temperature of the wafer, and a method of correcting it by moving an optical member that contributes to the imaging characteristics within the optical system. However, in the temperature-controlled method, the wafer expands and contracts radially from its center.
It is not possible to correct magnification errors independently in the scanning direction and in the direction perpendicular to this direction. Moreover, it also has the disadvantage that it takes time to utilize heat conduction. Additionally, a method has been announced in which compensation is achieved by moving the optical members that contribute to the imaging characteristics within the optical system.
In that case, there is a risk that the imaging performance of the projection system itself will be deteriorated.

SUMMARY OF THE INVENTION It is an object of the present invention to adjust magnification errors occurring in imaging without introducing any disadvantages. In order to achieve this purpose, in the embodiments to be described later, there is no formation of a light beam that forms an image with the object, and more preferably, in the vicinity of the object, image, or intermediate image formation plane, other than the deviation of the image formation position of the projection optical system. A transparent thin film member that has little effect on image performance is inserted, and this part is curved to correct the imaging position.

Examples of the present invention will be described below. Most of the parts constituting the optical system in FIG. 3 have already been explained. That is, M is a mask and W is a wafer, which move together in the scanning direction y. Reference numeral S denotes a light-shielding plate, which is provided with an axle opening shaped like T in FIG. 2, and is fixed at a position close to the wafer. BS1 and BS2 are optical path changing mirrors, and the convex mirror _M2 of the concave mirror _M1 is arranged on the optical axis OO′ with its spherical center shifted. Then, the central ray emitted from the illuminated mask M is reflected by the mirrors, then reversed in order through the concave mirror M ₁ , convex mirror M ₂ , concave mirror M ₁ , and mirror B ₂ and enters the wafer W.

Next, the portion B fixed near the wafer W is a member characteristic of the present invention, and is made of a thin body bent into a semi-cylindrical shape with a thickness that does not affect the imaging performance. The curved thin film B is arranged so that its generating line coincides with the scanning direction y, and its appearance in the direction perpendicular to the drawing is as shown in FIGS. 6 and 7.

In general, an optical thin film such as a pellicle can be optically thought of as a very thin parallel surface. Since this optical member is very thin and has no power, it is known that the imaging performance hardly deteriorates. Therefore, even if a transparent optical member that is inclined with respect to the ray (principal ray) that is the center of the luminous flux, such as an optical thin body that is curved in the optical axis direction, is newly inserted near the object or image plane, The degree to which this deteriorates the imaging characteristics on the image plane is small. Nevertheless, the chief ray proceeds while undergoing refraction at the thin body B, as shown in FIG. 6, according to Snell's law.

For the above reasons, in semiconductor printing equipment,
A mask that is the object surface in the printing optical path, or
Near the wafer, which is the image plane, there is a moderately curved
By inserting a transparent optical thin body, the printed image on the wafer can be moved without deteriorating its imaging characteristics.

The optical system of a semiconductor printing apparatus is usually called a telecentric system, and is designed so that the principal rays of all points within the printing screen are projected perpendicularly to the wafer surface. By doing so, even when the wafer is out of focus, the center of the light beam does not shift laterally on the wafer, so there is an advantage that no magnification error occurs.

In FIG. 6, if the thickness of the optical thin film just inserted is dmm, the refractive index is n, the direction in which the printed image is desired to be laterally shifted on the wafer is taken as the x-axis, and the direction of the chief ray perpendicular to this is taken as the z-axis, then The amount of curvature zmm of the optical thin film to be inserted in the z direction is expressed as a function of x.

The amount of lateral deviation Δxmm of the printed image in the x direction obtained at this time is Δx=d(1-1/n)·dz/dx .

In particular, when the optical thin body is stretched into a spherical shape with a radius of R ^mm , the relationship z=x ² /2R exists, so the equation becomes Δ=d/R(1-1/n)... If n=1.5 and d/R is varied, Figure 8 is obtained. Using this diagram, d/R for the amount of deviation to be corrected can be found. Although good results were obtained using a pellicle with a thickness of 10 μm in the experiments in the Examples, thicker or thinner pellicles can be used depending on the performance of the engineering system. However, in the field of precision engineering, one standard is to suppress the wavefront aberration before and after the insertion of the field of view to be between λ (wavelength) and -λ.

Returning to FIG. 6, assume that the chief rays are incident on a, b, c, d, and e on the wafer W when there is no thin body. Now, when thin body B is inserted with its concave surface facing wafer W, each light ray is refracted,
Images are formed on a′, b′, d′, and e′. As a result, the imaging magnification can be reduced in the direction of the x-axis in the figure.

Conversely, if the thin body B is inserted with its convex surface facing the wafer W as shown in FIG. do. Therefore, the image can be enlarged in the x-axis direction in the figure.

That is, by determining the direction of curvature of the font B and the numerical value according to the above formula, any magnification error can be dealt with.

Note that since the curved thin body B is formed in a semi-cylindrical shape, it acts as a mere parallel plane plate in the direction perpendicular to the drawing, that is, in the direction parallel to the generatrix, and the magnification in this direction is affected. not give. Further, in the above description, it has been assumed that the magnification adjustment member is inserted on the wafer side B in FIG. 3, but the same effect can be obtained in principle even if it is inserted on the mask side B'. However, in this case, wafer side B
The relationship between expansion and contraction is different when you put it in . That is, if the wafer is inserted in a convex manner onto the wafer in B, the wafer printed image will be enlarged, whereas if the wafer is inserted in a convex manner into the mask in A, the wafer printed image will be reduced. In addition, optical path switching mirror BS ₁ or
A thin body may be moved between BS ₂ and concave mirror M ₁ ,
Furthermore, if there is an intermediate image formed in the optical path, a thin body can be placed near it.

As described above, as the first embodiment of the present invention, there is a method in which an optical thin film member is inserted into the optical system PS into a semi-cylindrical curved metal frame suitable for each correction form.
The insertion of this element makes it possible to correct the subtle magnification light during multiple occasions. Furthermore, if there is expansion or contraction due to the wafer process, the correction member of the present invention can be replaced according to the amount of expansion or contraction, and if the curvature is made freely changeable, changes in magnification can be freely controlled with one member. It is possible to do so.

Next, the correction member including the frame supporting the optical thin body will be explained in more detail. First, as shown in FIGS. 10A and 10B, there is an example in which an optical thin film several microns thick is applied to a metal frame curved into a semi-cylindrical shape, and the curvature of the circumference of the metal frame is changed. Specifically, we created an aperture in the shape of a shaft that allows sufficient light to pass through a rectangular metal frame.
AP is provided and force is applied from both sides to make it into a semi-cylindrical shape. By attaching an optical thin film to this metal frame and adjusting the force applied to the metal frame, the curvature of the semicircular circumference can be freely applied. Therefore, the magnification of the image printed on the wafer can be corrected. As a third embodiment, the thin film may be stretched in a general shape other than a mere semi-cylindrical shape. In other words, when the thin film is formed into a semi-cylindrical shape, the amount of correction of the position on the wafer surface is a linear function of the position on the wafer, as shown in FIG. However, among the printing image deviations that actually occur, there are
Some are expressed as quadratic or higher-order functions of position. In order to remove these image deviations,
It is necessary to curve the optical member into an appropriate shape that satisfies the formula. For example, it is expressed as a quadratic or cubic relationship of the image shift position (11th
If this is to be corrected in the case shown in Figure 1, the amount of curvature of the thin film must follow a 3rd or 4th order function of position, respectively, according to the equation.

Furthermore, it is also possible to perform magnification correction of an image two-dimensionally by curving the optical thin body three-dimensionally.
In Fig. 12, a hemispherical thin body indicated as B'' is inserted into a projection exposure device using a projection lens PL, aligning the optical axis with the spherical center to correct the magnification.Lens Since the printing is performed in an area centered on the optical axis in a device using the optical axis, the correction means has a three-dimensional hemispherical shape.The hemispherical member can be made of a solid material such as glass or plastic. Alternatively, it can be achieved by stretching a nitrocellulose membrane inside an airtight box and inflating it into a hemispherical shape using the difference in air pressure.

Furthermore, the technical idea of the present invention can also be used to correct minute distortions in lens optical systems. For example, in a lens system, there may be a slight difference in magnification between a specific direction and a direction perpendicular to the specific direction due to the accumulation of processing errors and assembly errors in the glass. Even in such a case, it is possible to adjust the magnification difference by inserting an optical member as described in the present invention into the optical path and curving the member into a cylindrical shape. At this time,
The generating line of the cylinder is set in a direction that substantially coincides with one of the two directions in which the greatest difference in magnification occurs. This situation is shown in Figure 13.
An optical member called CL corrects the variable magnification. Of course, this correction may be performed on the mask (or reticle) side indicated by M.

In the example explained above, a curved optical thin film with a uniform refractive index and thickness was used, but a parallel plate whose refractive index changes continuously with a predetermined distribution or a parallel plate whose thickness changes minutely and continuously is used. A flat plate may be placed near the object plane, image plane, or intermediate image forming plane. In this case, the change in the refractive index or thickness of the flat plate imparts a small prism-like refraction effect to the light beam, allowing the desired magnification to be corrected. For example, to form a parallel plate refractive index distribution using glass, it is appropriate to use the ion diffusion method, and to form a thickness variation, one method is to evaporate a substance with the same refractive index onto a thin film. It is.

According to the present invention described above, it is possible to correct the lateral magnification without modifying the conventional optical system. In the well-known semiconductor baking process, a pellicle is stretched flat and slightly away from the wafer to prevent dust from falling onto the wafer, but the present invention makes active use of this type of pellicle. can be realized.

By shaping the optical thin film into a predetermined shape according to the present invention, it has become possible to correct minute machine differences or expansion and contraction of the wafer due to the process. This fact has a very important meaning in the production of VS1, where alignment errors are on the order of 0.1 μm.
Furthermore, when applied to a mirror optical system, the magnification in the direction perpendicular to the scanning direction can be freely changed, which is an advantage not seen in the past.

By implementing the present invention, alignment errors in manufacturing LS1 have been dramatically reduced, making it possible to produce with high yield. Furthermore, the technical idea of the present invention is not limited to these fields, but can be applied to any field that requires minute alignment.

[Brief explanation of the drawing]

FIGS. 1A and 1B are plan views schematically showing magnification errors. FIG. 2 is a plan view showing an image on a wafer. Third
The figure is an optical sectional view showing an embodiment of the present invention. FIGS. 4A and 4B are plan views schematically showing magnification errors. FIG. 5 is an auxiliary explanatory diagram of the scanning mechanism. FIG. 6 is an enlarged view of the main parts of the embodiment. FIG. 7 is a lateral optical cross-sectional view of the embodiment.
Figure 8 is a diagram with radius and thickness as variables. Figure 9 is a modified view of the main part. FIG. 10A is a plan view of the constituent members, and FIG. 10B is a perspective view. FIG. 11 is a diagram showing the position and amount of deviation. 12 and 13 are perspective views of the optical system. In the figure, M is a mask, W is a wafer, M ₁ is a concave mirror, M ₂ is a convex mirror, B and B' are thin bodies, C is a frame with an opening, and y is the scanning direction.

Claims (1)

[Scope of Claims] 1. An optical means that is equipped with an imaging optical system and has an optical thickness and a curved shape that does not substantially affect the imaging performance other than the deviation of the imaging position is set in the optical path. An image-adjusted optical device characterized by being disposed inside. 2. The image-adjusted optical device according to claim 1, wherein the optical means is arranged near an object plane, an image plane, or an intermediate image plane of the imaging optical system. 3. The image-adjusted optical device according to claim 1, wherein the degree of curvature of the optical means is changed according to the information density of the object. 4. The optical device is a device that exposes a band-shaped image of an object by the imaging optical system and a photosensitive surface by relatively scanning it, and the optical means is curved into a cylindrical shape whose generatrix direction substantially coincides with the scanning direction. An image-adjusted optical device according to claim 1. 5. The image-adjusted optical device according to claim 1, wherein the optical means is a pellicle held in a frame. 6. An optical system that is equipped with an imaging optical system and has an optical thickness that does not substantially affect imaging performance other than the deviation of the imaging position, and that has a refractive index distribution or thickness distribution that changes continuously. 1. An image-adjusted optical device, characterized in that means is provided within the light.