JP2015040985A - Method for producing substrate for mask blank, method for producing mask blank, method for producing transfer mask, and method for producing semiconductor device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for producing a substrate for mask blanks, in which the substrate copes with deformation, when a transfer mask is fastened to a stage of an exposure device, while suppressing a decrease in the production throughput thereof and has effective and extremely high flatness on the main surface thereof and to provide a method for producing a mask blank and a method for producing a transfer mask.SOLUTION: The method for producing the substrate for mask blanks comprises the steps of: acquiring surface shape data of the main surface of a translucent substrate; acquiring the main surface shape of the translucent substrate, which is fastened to the mask stage of the exposure device, by simulation; setting the virtual reference plane, which is defined by the Zernike polynominal expressed in a polar coordinate system and becomes an optically-effective and flat reference plane; acquiring the data (PV values) on a difference between the maximum value and the minimum value of another difference between the main surface shape of the translucent substrate, which shape is acquired by the simulation, and the set virtual reference plane; and selecting the substrate for mask blanks, the value (PV value) of which substrate becomes a predetermined value or smaller.

Description

  The present invention relates to a manufacturing method of a mask blank substrate used in photolithography, a manufacturing method of a mask blank, a manufacturing method of a transfer mask, and a manufacturing method of a semiconductor device. In particular, the present invention has a focus margin when performing photolithography. The present invention relates to a mask blank substrate manufacturing method, a mask blank manufacturing method, a transfer mask manufacturing method, and a semiconductor device manufacturing method, which are suitable for securing the mask blank.

  In order to cope with the miniaturization of semiconductor devices, the exposure apparatus using ArF exposure light having a wavelength of 193 nm has been increased in NA (high numerical aperture), and further increased in NA by introducing immersion exposure technology. Now, NA 1.35 has come into practical use.

  In order to cope with such a demand for miniaturization and high NA, it is required to increase the flatness of the transfer mask. When the flatness of the mask surface, which is an object point, decreases, the in-focus position of the image point on the wafer transferred through the projection lens shifts. For this reason, when the flatness of the mask surface decreases, the allowable focus tolerance decreases. On the other hand, due to the optical principle, increasing the NA of the projection lens decreases the depth of focus. Therefore, since the latitude in focus in the lithography process decreases as the NA increases, high flatness on the mask surface is required. For this reason, high flatness is required for the main surface on the side where a thin film for forming a pattern is provided, even for a light-transmitting substrate used as a mask blank that is an original for producing a transfer mask. It has been. In order to meet this flatness requirement, for example, as disclosed in Patent Document 1, double-side polishing that polishes both the front and back surfaces of a mask blank substrate using a polishing liquid such as a polishing cloth and a polishing liquid containing polishing grains. Has been used frequently. However, in the polishing of a translucent substrate by a conventional double-side polishing apparatus, there has been a limit to increasing the flatness of the main surface. For this reason, as shown in Patent Document 2, a technique for measuring the shape of the main surface of the substrate and performing plasma etching on a relatively convex portion has been developed.

  On the other hand, the transfer mask is fixed by vacuum chucking to the mask stage of the exposure apparatus. However, depending on the exposure apparatus, a chuck system that has a strong force for vacuum chucking the transfer mask may be employed. is there. A transfer mask used in such a chuck type exposure apparatus may be greatly deformed during chucking due to compatibility with a mask stage and a vacuum chuck. Conventionally, product management is performed based on the flatness of the transfer mask before chucking, so even if the main surface shape before chucking is a good product with high flatness, depending on the compatibility with the mask stage and vacuum chuck, In some cases, the flatness of the transfer mask is greatly deteriorated due to deformation when chucked on the mask stage of the exposure apparatus. In particular, the tendency is remarkable in a substrate having a low symmetry of the shape of the main surface and a twisted shape. For this reason, it is necessary to consider the flatness when the transfer mask is chucked on the vacuum chuck. Patent Document 3 discloses a method for obtaining a flatness of a mask substrate after being chucked on a mask stage of an exposure apparatus by simulation and selecting a good mask substrate based on whether the result satisfies the flatness specification. Proposed.

  Further, as the NA of the exposure apparatus is increased, the focus tolerance in the lithography process is decreasing, and thus the influence of lens aberration of the projection optical system on the transfer accuracy is increasing. Patent Document 4 discloses two correction optical elements having a surface shape that can be defined by a Zernike polynomial for correcting aberration caused by the lens heating effect of the projection optical system.

Japanese Unexamined Patent Publication No. 1-404067 JP 2002-318450 A JP 2004-046259 A JP 2008-028388 A

  As described above, the mask blank substrate is required to have high flatness, and in response to this, technological development for locally processing the main surface of the translucent substrate has been advanced. For example, as described above, Patent Document 2 discloses a technique for manufacturing a mask blank substrate having high flatness by locally performing plasma etching on a relatively convex portion. However, in this method, in order to obtain a desired high flatness, it is necessary to measure the shape of the main surface for each substrate and to perform local processing such as plasma etching. There was a problem of lowering.

  On the other hand, when a translucent substrate is manufactured by the method disclosed in Patent Document 3, the main surface before chucking such that the main surface shape after chucking of the translucent substrate calculated by simulation has high flatness. Only the shape is selected as the mask blank substrate. There is a problem that it is difficult to manufacture a translucent substrate having such a characteristic and having a main surface shape before chucking at a high yield. In addition, the exposure apparatus disclosed in Patent Document 4 aims to control lens aberration correction and change of the wavefront of exposure light due to the history during exposure processing, and has a flatness for mask blanks and masks. A high one was necessary.

  The problem to be solved by the present invention is to reduce the throughput when processing the mask blank substrate, suppress the burden on the manufacturing equipment, and in the exposure state chucked on the stage of the exposure apparatus, An object of the present invention is to provide a mask blank substrate manufacturing method, a mask blank manufacturing method, and a transfer mask manufacturing method having an extremely high effective surface flatness. Another object of the present invention is to manufacture a semiconductor device having a high circuit accuracy and a stable circuit operation by using the transfer mask.

  The present inventor has studied the following points in order to solve the above-mentioned problems found by the present inventor. The main surface of the mask blank substrate is required to have extremely high flatness as described in the background art. However, it is very difficult to actually manufacture such a completely flat surface. In addition, there has been a contrivance to bring the surface close to flat using a technique such as local plasma etching, but this method has low side effects such as low throughput, high device cost, and easy occurrence of foreign matter defects. Therefore, the present inventor changed the idea not to excessively pursue a mechanical flat surface but to pursue an optical flat surface, in other words, an equiwavefront flat surface as a virtual reference surface. The present inventor has found that what is essentially required for image transfer through a projection lens is not necessarily a mechanical plane, but an optical flat surface with a uniform wavefront. This is the first characteristic point of the present invention.

  The inventor has paid attention to the fact that the projection lens of the exposure apparatus is provided with an aberration correction function for correcting lens aberration. This function is originally provided to improve the performance of the exposure apparatus, such as assembling the exposure apparatus lens, adjusting the installation, and dealing with changes over time. However, if this function is used, the optical flat surface becomes a mechanical flat surface. It came to the idea that it can be moved closer to the polished surface. This is the second characteristic point of the present invention.

  The inventor has also conceived of using a Zernike polynomial approximation surface, which is a polar coordinate system, as a description of the optical flat surface. This is the third characteristic point of the present invention.

  Furthermore, the present inventor has noticed that the shape of the main surface of the mask blank substrate measured and measured is different from the shape of the main surface in the exposure state. Then, the shape of the mask blank substrate main surface (transfer main surface) in the exposure state is obtained by deformation simulation when chucked on the mask stage of the exposure apparatus, and the optical flat surface and the main surface obtained by the simulation are obtained. It was found that it is effective to fit the surface shape and sort based on the amount of deviation. This is the fourth characteristic point of the present invention.

As described above, the present invention has been made based on the above-described idea, and has the following configuration.
(Configuration 1)
A mask blank substrate manufacturing method used for manufacturing a mask blank in which a thin film for forming a transfer pattern is provided on one main surface of a translucent substrate having a pair of opposing main surfaces,
A shape measuring step for obtaining a surface shape of the main surface on the side where the thin film is provided in the light-transmitting substrate;
A simulation step for obtaining a chucked main surface shape by simulation when the translucent substrate is chucked on a mask stage of an exposure apparatus;
The main surface shape after chucking is a calculation area inside a circle having a predetermined diameter with respect to the center of the translucent substrate, and a virtual reference plane defined by a Zernike polynomial expressed in a polar coordinate system Performing shape fitting, obtaining difference data between the post chuck main surface shape and the virtual reference plane;
A mask blank substrate comprising a step of selecting the translucent substrate having a surface shape in which a difference between a maximum height and a minimum height in the calculation area of the difference data is equal to or less than a predetermined value. Manufacturing method.

(Configuration 2)
2. The method of manufacturing a mask blank substrate according to Configuration 1, wherein the predetermined value is ８ of a wavelength λ of exposure light used for exposure transfer.

(Configuration 3)
3. The method of manufacturing a mask blank substrate according to Configuration 1 or 2, wherein the predetermined diameter is 104 mm.

(Configuration 4)
The virtual reference plane has a shape defined by a Zernike polynomial in which the order of the variable related to the radius is composed only of terms of the second order or less, and the order of the variable related to the radius includes one or more of the second order terms. 4. A method for manufacturing a mask blank substrate according to any one of configurations 1 to 3.

(Configuration 5)
Method for producing a mask blank substrate according to any one of configurations 1, wherein 4 of the coefficient of determination R ² calculated from the difference shape is 0.9 or more.

(Configuration 6)
Any one of configurations 1 to 5, wherein the flatness calculated in a rectangular inner region having a side of 132 mm on the basis of the center of the translucent substrate in the post-chuck main surface shape is 0.2 μm or less. A method for producing a mask blank substrate according to claim 1.

(Configuration 7)
A step of associating the selected light-transmitting substrate with information related to the Zernike polynomial in the virtual reference plane obtained by performing shape fitting on the light-transmitting substrate and recording the information on a recording apparatus. A method for manufacturing a mask blank substrate according to any one of configurations 1 to 6.

(Configuration 8)
Manufacturing of a mask blank comprising a step of providing a thin film for forming a transfer pattern on one main surface of a mask blank manufactured by the method for manufacturing a mask blank substrate according to any one of configurations 1 to 7 Method.

(Configuration 9)
A mask blank manufacturing method in which a thin film for forming a transfer pattern is provided on one main surface of a translucent substrate having a pair of main surfaces facing each other,
A shape measuring step for obtaining a surface shape of the thin film in the mask blank;
A simulation process for obtaining a chucked surface shape by simulation when the mask blank is chucked on a mask stage of an exposure apparatus;
The post-chuck surface shape is calculated with respect to a virtual reference plane defined by a Zernike polynomial expressed in a polar coordinate system in a calculation region inside a circle having a predetermined diameter with respect to the center of the translucent substrate. Performing a fitting and obtaining difference data between the post chuck surface shape and the virtual reference plane;
And a step of selecting the mask blank having a surface shape in which a difference between a maximum height and a minimum height in the calculation area of the difference data is equal to or less than a predetermined value.

(Configuration 10)
10. The method of manufacturing a mask blank according to Configuration 9, wherein the predetermined value is 1/8 of a wavelength [lambda] of exposure light used for exposure transfer.

(Configuration 11)
The method for manufacturing a mask blank according to Configuration 9 or 10, wherein the predetermined diameter is 104 mm.

(Configuration 12)
The virtual reference plane has a shape defined by a Zernike polynomial in which the order of the variable related to the radius is composed only of terms of the second order or less, and the order of the variable related to the radius includes one or more of the second order terms. 12. A method for manufacturing a mask blank according to any one of configurations 9 to 11,

(Configuration 13)
Mask blank manufacturing method according to any one of configurations 9, wherein 12 of the coefficient of determination R ² calculated from the difference shape is 0.9 or more.

(Configuration 14)
Any one of configurations 9 to 13, wherein the flatness calculated in a rectangular inner region having a side of 132 mm on the basis of the center of the translucent substrate in the main surface shape after chucking is 0.2 μm or less. The manufacturing method of the mask blank of crab.

(Configuration 15)
And a step of associating the selected mask blank with information related to the Zernike polynomial on the virtual reference plane obtained by performing shape fitting on the mask blank and recording the information on a recording apparatus. To 14. The method for producing a mask blank according to any one of 14 to 14.

(Configuration 16)
A method for producing a transfer mask, comprising a step of forming a transfer pattern on the thin film of the mask blank produced by the method for producing a mask blank according to any one of Structures 8 to 15.

(Configuration 17)
A method of manufacturing a semiconductor device, comprising: an exposure step of chucking the transfer mask according to Configuration 16 on a mask stage of an exposure apparatus and transferring the transfer pattern of the transfer mask onto a semiconductor substrate by lithography. .

(Configuration 18)
The exposure apparatus has a function of performing wavefront correction of a shape defined by Zernike polynomials on a wavefront of transmitted light transmitted from a transfer pattern of the transfer mask, and a mask stage used in the simulation process 18. A method of manufacturing a semiconductor device according to Configuration 17, wherein the semiconductor device has the same shape as that of the semiconductor device.

(Configuration 19)
A step of producing a mask blank by providing the transfer pattern forming thin film on one main surface of the mask blank substrate produced by the method for producing a mask blank substrate according to Configuration 7,
Forming a transfer pattern on the transfer pattern forming thin film of the manufactured mask blank and manufacturing a transfer mask;
An exposure step of chucking the produced transfer mask on a mask stage of an exposure apparatus, and transferring the transfer pattern of the transfer mask onto a semiconductor substrate by a lithography method,
The exposure apparatus has a function of performing wavefront correction of a shape defined by Zernike polynomials on a wavefront of transmitted light transmitted from a transfer pattern of the transfer mask, and a mask stage used in the simulation process Has the same shape as
The exposure step uses the information related to the Zernike polynomial in the virtual reference plane associated with the mask blank substrate, and corrects the wavefront of transmitted light transmitted from the transfer pattern of the transfer mask. A method for manufacturing a semiconductor device.

(Configuration 20)
A step of producing a transfer mask by forming a transfer pattern on the transfer pattern forming thin film of the mask blank produced by the method for producing a mask blank according to Configuration 15,
An exposure step of chucking the produced transfer mask on a mask stage of an exposure apparatus, and transferring the transfer pattern of the transfer mask onto a semiconductor substrate by a lithography method,
The exposure apparatus has a function of performing wavefront correction of a shape defined by Zernike polynomials on a wavefront of transmitted light transmitted from a transfer pattern of the transfer mask, and a mask stage used in the simulation process Has the same shape as
The exposure step uses the information related to the Zernike polynomial in the virtual reference plane associated with the mask blank to correct the wavefront of the transmitted light transmitted from the transfer pattern of the transfer mask. A method for manufacturing a semiconductor device.

  According to the present invention, considering the influence when the transfer mask is chucked on the mask stage of the exposure apparatus, the main surface shape (mask blank substrate) of the transfer mask in the exposure state (mask blank substrate) is determined by simulation. A mask blank substrate that satisfies the requirements for extremely high flatness can be obtained by obtaining the shape of the main transfer surface and selecting the reference value based on the replacement of mechanical flatness with optical flatness. In addition, it is possible to manufacture a highly flat mask blank substrate without reducing the throughput during processing of the mask blank substrate and with less burden on manufacturing equipment. The effective flatness of the mask blank and the transfer mask manufactured using the mask blank substrate is as high as that of the mask blank substrate. When exposure is performed using the transfer mask, focus tolerance and high positional accuracy are ensured, and extremely high transfer accuracy is obtained. As a result, the circuit operating characteristics of the manufactured semiconductor device are stabilized.

It is sectional drawing of the substrate for mask blanks for demonstrating the concept of this invention. It is a process flow figure showing a manufacturing process of a substrate for mask blanks by the present invention. It is a cross-sectional structure diagram when a transmission type transfer mask is placed on a mask stage. It is a process flow figure showing a manufacturing process of a substrate for mask blanks by the present invention. It is a surface height contour map of the main surface of the mask blank substrate, where (a) shows the case of sample A, (b) shows sample B, and (c) shows the case of sample C. It is a contour distribution diagram showing the surface shape distribution of the main surface within a diameter of 104 mm in sample A, (a) is a surface shape simulation diagram obtained by simulation of the surface shape when chucked on the mask stage, (b) corresponds to it. FIG. 4C is a contour distribution diagram showing the virtual reference plane, and FIG. 4C is a differential contour map showing the difference between the surface shape acquired by the simulation and the virtual reference plane. It is a contour distribution diagram showing the surface shape distribution of the main surface within a diameter of 104 mm in sample B, (a) is a surface shape simulation diagram obtained by simulation of the surface shape when chucked on the mask stage, (b) corresponds to it. FIG. 4C is a contour distribution diagram showing the virtual reference plane, and FIG. 4C is a differential contour map showing the difference between the surface shape acquired by the simulation and the virtual reference plane. It is a contour line distribution diagram showing the surface shape distribution of the main surface within a diameter of 104 mm in sample C, (a) is a surface shape simulation diagram obtained by simulation of the surface shape when chucked on the mask stage, (b) corresponds to it. FIG. 4C is a contour distribution diagram showing the virtual reference plane, and FIG. 4C is a differential contour map showing the difference between the surface shape acquired by the simulation and the virtual reference plane. It is a principal part cross-section block diagram of the exposure apparatus optical part which shows the outline | summary of the structure of the illumination of an exposure apparatus, and a projection optical system.

  The best mode for carrying out the present invention will be specifically described below including the concept thereof with reference to the drawings. In the drawings, the same or corresponding parts are denoted by the same reference numerals, and the description thereof may be simplified or omitted.

[Manufacturing method of mask blank substrate]
Here, a mask blank substrate and a manufacturing method thereof will be described. First, the configuration concept of the present invention will be described, and then, examples implemented based on the concept will be shown together with comparative examples.

  First, the present invention is based on the fact that what is essentially required for image transfer through a projection lens is not a mechanical plane, but an optical flat surface with a uniform wavefront. . This point will be described with reference to FIG. The main surface of the mask blank substrate (one main surface on the side where the thin film for forming the transfer pattern is provided) is required to have extremely high flatness as described in the background art. FIG. 1 shows a cross section of a mask blank. In general, the main surface of a mask blank substrate 1 can be formed to be a completely flat surface as indicated by reference numeral 2 in FIG. It is considered ideal. However, it is very difficult to actually manufacture such a completely flat surface. In addition, a method of increasing the flatness by using a technique such as local plasma etching has been performed. However, this method has a low throughput, an apparatus cost, and a lot of side effects such as the occurrence of a foreign substance defect. Therefore, in the present invention, the idea is changed not to pursue excessively a mechanical flat surface but to pursue an optical flat surface, in other words, an equiwavefront flat surface as a virtual reference surface.

  In the present invention, the optical flat surface is shifted from the mechanical flat surface and brought close to the polished surface by using an aberration correction function for correcting lens aberration provided in the projection lens of the exposure apparatus. . This point will be described with reference to FIG. Reference numeral 3 in FIG. 1 indicates a cross-sectional shape of the main surface formed by polishing. For example, the deviation from the ideal mechanical flat surface at point A in the figure is d1, but the deviation from the optical flat surface (cross section) 4 using the aberration correction function of the exposure apparatus is extremely small as d2. it can. The optically flat surface here is a surface that shows ideal imaging by light rays that are aligned from the wavefront and intentionally give aberration to the projection lens using the lens aberration correction function. That is. In other words, the optical flat surface referred to here is an object point surface that is on the conjugate of the projection lens when the wafer surface is the image point surface, and mechanically gives aberration to the projection lens. An object surface that is deformed from a flat surface to a desired shape. In the present application, the flatness of the object surface is referred to as optical flatness. The optical flat surface is not a single fixed surface that is uniquely determined as a mechanically flat surface, but provides a certain degree of freedom by giving aberration to the projection lens according to the surface shape of the mask blank substrate. You can decide. Therefore, it is easy to increase the optical flatness as compared with the mechanical flatness.

  In the present invention, a Zernike polynomial approximate surface which is a polar coordinate system is used to describe the optical flat surface. The origin of the polar coordinates is the center of the mask blank substrate. A mask blank substrate used in an ArF exposure apparatus or the like is chamfered at some corners, but has a rectangular shape with a width of about 152 mm in both vertical and horizontal directions, and the mask pattern graphic layout is also displayed in XY coordinates. For this reason, the XY coordinate system is generally used as the coordinate system display. A characteristic feature of the present invention is that a square mask blank substrate is described in a polar coordinate system. Zernike polynomials are orthogonal polynomials, and each variable is independent and easy to handle. Since the aberration characteristics of the projection lens are associated with the terms of the Zernike polynomial expansion of the wavefront at the pupil plane of the projection lens, which is the Fourier transform plane, the projection lens aberration correction is linked to the optical flat surface of the blank substrate surface. It is very suitable for this. As an ArF exposure apparatus, the current mainstream is an ArF scanner, but the ArF exposure apparatus is not limited to a scanner, and may be a stepper. Further, the light source is not limited to the ArF excimer laser (wavelength 193 nm), and may be a KrF excimer laser (wavelength 248 nm), for example.

  In this way, with the optical flat surface as the virtual reference surface and the surface shape represented by the Zernike polynomial approximation, fitting with the main surface shape of the polished translucent substrate, the difference is the reference It is the gist of the present invention to select a substrate having a value less than the value as a mask blank substrate. Here, in this fitting, the transfer mask substrate portion (translucent substrate or mask blank substrate) caused by stress generated when the transfer mask manufactured from the mask blank substrate is chucked on the mask stage of the exposure apparatus. (Hereinafter, referred to as a light-transmitting substrate or a mask blank substrate) and a self-weight bending deformation of the mask blank substrate. That is, when the main surface shape of the polished translucent substrate is measured in a free standing state, the translucency associated with the stress generated when chucking the mask stage of the exposure apparatus using the measurement data of the main surface shape The deformation of the substrate and the self-weight bending deformation of the translucent substrate are obtained by simulation, and the main surface shape (transfer main surface shape) of the translucent substrate when exposed by the exposure apparatus is fitted with the virtual reference plane. Do. A typical method of this simulation is the finite element method, but it is not limited to this method. In this simulation, it is preferable to consider the chuck shape of the mask stage, the main surface shape of the translucent substrate including the area where the chuck contacts, the elastic coefficient of the translucent substrate, the specific gravity of the translucent substrate, and the like. Here, it is preferable to incorporate the main surface shape of the translucent substrate in the simulation up to the region outside the region to be chucked.

  As an index for selecting the mask blank substrate by the fitting, the virtual reference plane calculated in the calculation area inside the circle having a predetermined diameter on the main surface of the translucent substrate and the main surface shape of the translucent substrate are used. As a result of examination, it was found that a difference between the maximum height and the minimum height of the difference, that is, a so-called PV value is suitable. If the PV value deviates even at one point, the wavefront is shifted by that amount, and the transfer characteristics at that location are adversely affected.

  The calculation area should be set in consideration of the exposure slit length during the scan exposure of the exposure apparatus. The calculation area (equivalent to the size of the virtual reference plane) is at least desirably an inner area of a circle having a diameter of 84 mm or more. The calculation area is preferably an inner area of a circle having a diameter of 94 mm or more, and more preferably an inner area of a circle having a diameter of 104 mm, which is the maximum value of the exposure slit length.

  The reference value for selection of the mask blank substrate may be determined according to the required accuracy for the transfer mask produced from the selected mask blank substrate, but if sufficient wavefront control is required, the phase difference It is preferable to set the exposure wavelength λ to 1/8, that is, λ / 8, which is one of the measurement accuracy standards of the wavefront measuring and measuring apparatus, ie, λ / 8. As a result of the transfer evaluation, usually, sufficient transfer accuracy was obtained with this standard. When higher accuracy is required, such as when adjusting the exposure apparatus or when performing quality control (QC), it is more preferable to set the selection reference value to λ / 10.

Second index to select a substrate for a mask blank by fitting a coefficient of determination R ² calculated from the above-described difference shape. The coefficient of determination R ² is the square of the multiple correlation coefficient, which is also known as the contribution rate is a well-index which is used as a measure of the goodness of true regression equation obtained from the sample value. The defining equation is that the actual measurement value is y and the estimated value by the regression equation is f.

であり、この値が１に近いほど相対的な残差が少ないことを表す。ＰＶ値がポイントでの異常を選別する指標に対し、決定係数Ｒ^２は形状全体の残差の大きさを表す指標となる。マスクを作成して転写との相関をとって種々調べた結果、決定係数Ｒ^２が０．９以上で十分な転写精度が得られた。

The closer this value is to 1, the smaller the relative residual. To index the PV value is selected for malfunction at a point, the coefficient of determination R ² is an indicator of the magnitude of the residuals overall shape. By creating a mask result of examining various taking a correlation between the transfer coefficient of determination R ² was obtained sufficient transfer accuracy 0.9 or more.

  In addition, in order to manufacture a mask blank substrate that gives more preferable transfer performance, the following points were examined in detail.

  The aberration correction function of the projection lens can also be a higher-order term in which the degree of the variable related to the radius in the Zernike polynomial is third or higher. For this reason, the order of the variable related to the radius in the Zernike polynomial defining the virtual reference plane is not limited except that it is the order of the shape that can be corrected by the aberration correction function of the projection lens of the exposure apparatus. For example, the Zernike polynomial defining the virtual reference plane may include one or more terms in which the degree of the variable relating to the radius is the fourth order, and the term in which the order of the variable relating to the radius is the third order is 1 One or more may be included.

  On the other hand, when fitting was performed using even higher-order terms, it was good at a certain point in time, but it was also found that depending on the exposure situation, higher-order aberrations of the projection lens may fluctuate, resulting in inconvenience. Details of this point will be described later in [Exposure Method and Device Manufacturing Method Using the Same]. It was also found that if the order is only a first-order term, it is a one-dimensional tilt correction, and sufficient optical flatness cannot be obtained with this. As a result of detailed examination, it has been found that it is preferable that the order of the variable related to the radius is composed only of terms of the second order or less, and that the order of the variable related to the radius includes one or more terms of the second order. The necessary representative second-order term is the defocus term, which corresponds to the fourth term in the University of Arizona notation and the fifth term in the standard form. There are various types of Zernike polynomial approximation, such as the standard form, the University of Arizona method, and the Fringe Zernike method. However, any type of Zernike polynomial approximation may be used in applying the present invention.

  In the above-described method, sufficient optical flatness is ensured in an inner region having a predetermined diameter (up to 104 mm) with respect to the center of the mask blank, but chip exposure is performed in a maximum 104 mm × 132 mm region. Therefore, in addition to the above-described optical flatness selection, the main surface shape after chucking (transfer main surface) when chucked on the mask stage of the exposure apparatus in a rectangular inner region having a side of 132 mm with respect to the center of the mask blank substrate. When a standard for flatness after simulation of the shape) of 0.2 μm or less was used in combination, a better transfer result was obtained over the entire surface. Note that the maximum chip exposure is 104 mm × 132 mm, which is smaller than 132 mm × 132 mm in the main measurement reference region, because this does not limit the orientation of the mask blank. Further, the main surface of the mask blank substrate needs to be mirror-polished to a surface roughness of a predetermined level or more. The main surface preferably has a square root mean roughness Rq calculated in a square inner region having a side of 5 μm of 0.2 nm or less, and more preferably 0.15 nm or less. The surface roughness can be measured by, for example, an atomic force microscope (AFM).

  Next, a process for manufacturing a mask blank substrate having a high flatness according to the concept of the present invention will be described with reference to a mask blank substrate manufacturing process flowchart shown in FIG.

  First, as shown in step S1 of FIG. 2, the light-transmitting substrate is cut out from the synthetic quartz ingot into the shape of the mask blank substrate, and then, as shown in step S2 of FIG. A grinding process for polishing the main surface, the end face and the chamfered surface, followed by a process for precisely polishing the surface of the main surface as shown in step S3 in FIG. This polishing is usually performed in multiple stages. There are various polishing methods, and there is no particular limitation here, but CMP (Chemical Mechanical Polishing) using a polishing agent such as cerium oxide or polishing using a polishing agent such as colloidal silica is suitably performed. . Thereafter, as shown in step S4 of FIG. Up to the above step S4 may be a normal method. Here, the case of a general synthetic quartz glass is shown as a mask blank material, but it is not necessarily limited to this as long as it is transparent to exposure light and can be used as a substrate for a transfer mask. Is not to be done. For example, depending on the exposure wavelength, soda lime glass, aluminosilicate glass, borosilicate glass, alkali-free glass, calcium fluoride glass, and the like are also applicable.

  The feature of the present invention is after step S5 in FIG. First, in step S5, the deformation of the main surface (transfer main surface) when the translucent substrate is chucked on the mask stage of the exposure apparatus is calculated and produced based on the mask blank substrate selected in the subsequent step. The surface shape of the main surface of the substrate portion in the exposure state of the transfer mask to be obtained is obtained by simulation.

  FIG. 3 is a cross-sectional view of the transmissive transfer mask chucked on the mask stage of the exposure apparatus. In this case, a transmission type transfer mask 51 is placed on the mask stage 50 and vacuum suction or vacuum suction is performed by the suction portion 52. This is because, particularly when the exposure apparatus is a scanner, it is necessary to scan the mask at high speed in synchronization with the wafer, and the transfer mask 51 is fixed so as not to be displaced from the mask stage 50. A portion 55 where the mask stage 50 and the transfer mask 51 are in contact is the peripheral portion of the transfer mask and is on the main surface (transfer main surface) 54 side where the mask pattern 53 is formed. Because of this usage, in simulating the main surface shape of the translucent substrate, the main surface shape of the mask blank substrate including the outer peripheral part that may come into contact with the mask stage, vacuum suction or It is desirable to take into account the shape of the chuck portion for suctioning under reduced pressure and the surface shape of the mask blank substrate where the chuck portion contacts. Since the transfer mask 51 is placed horizontally in a state where a part of the outer peripheral portion is supported, bending due to its own weight also occurs. Therefore, the simulation is performed in consideration of the stress generated by the contact portion between the vacuum suction or vacuum suction and the mask stage and the self-weight deflection. Here, although this simulation is performed based on the main surface shape acquired at process S4, it is necessary to pay attention to the measuring method of this shape. In the case of measurement in the free-standing state where the shape of the main surface of the translucent substrate is less likely to be deformed by the weight of the translucent substrate or external stress, actual measurement values can be applied. When measuring by placing the translucent substrate horizontally, it is deformed from the free state due to its own weight deflection or the like, so it is desirable to perform the simulation in consideration of the deformation. Note that the free standing measurement is, for example, a method of measuring in a state where the translucent substrate stands vertically with the side surface of the translucent substrate as the bottom.

Next, calculation calculation of a virtual reference plane is performed. As described above, this virtual reference surface is an optical flat surface determined by the Zernike polynomial expressed in the polar coordinate system. As described above, in view of efficiency, this virtual reference plane is composed of Zernike that is composed only of terms in which the degree of the variable relating to the radius is not more than second order, and the degree of the variable relating to the radius is one or more of the second order terms. Although it is practical to be defined by a polynomial, the accuracy may be increased by including higher-order terms. The range is, for example, within a circle with a diameter of 104 mm centered on the center of the translucent substrate. The virtual reference plane is calculated with reference to the surface shape (transfer main surface) of the main surface of the translucent substrate simulated in step S5. Next, as shown in step S7 of FIG. 2, the difference between the virtual reference surface shape calculated in step S5 and the main surface shape of the light-transmitting substrate simulated in step S5 is calculated, and the difference shape data is calculated. (Difference data) is acquired. After that, as shown in step S8 of FIG. 2, a difference between the maximum height and the minimum height, so-called PV value, is calculated from the difference data obtained in step S7. Thereafter, as shown in step S9 of FIG. 2, it is determined whether the PV value obtained in step S8 is a predetermined value X or less, or a value exceeding the predetermined value X. Selection is completed as a highly flat mask blank substrate (step S10 in FIG. 2). Translucent substrates exceeding this range are considered to be used as mask blank substrates for middle layers and rough layers as low / medium flat products, or returned to the polishing step of step S3, or through a local processing step to step S4. Thereafter, the same process is performed again or the translucent substrate is discarded (process S11 in FIG. 2). Here, the predetermined value X is an application layer of a transmission type transfer mask manufactured using a mask blank substrate selected based on this criterion, for example, a poly layer (gate layer) or a wiring layer. Since the required accuracy differs depending on whether it is a diffusion layer or the like, it is practical to determine the value according to the required required accuracy. From an optical standpoint, the predetermined value X is, for example, 1/8 of the exposure wavelength λ. This is because, as is well known in optical measurement, the value of 1/8 of the exposure wavelength λ is a typical reference value for wavefront control, and there is a great difference in imaging performance and the like with that value as a boundary.
By the above mask blank substrate manufacturing method, a mask blank substrate having extremely high flatness can be manufactured with high throughput. In the case of ArF excimer exposure, λ / 8 is 25 nm (rounded up after the decimal point), and optical flatness much higher than that of the conventional method can be obtained without reducing the throughput when processing the mask blank substrate. It is possible to reduce the burden on the manufacturing equipment.

Next, a process for producing another mask blank substrate with a coefficient of determination R ^2, will be described with reference to the manufacturing process flow chart of a substrate for a mask blank of FIG. The process up to step S9 is the same as the method for manufacturing the mask blank substrate of FIG. The difference is that after step S9, if the PV value is a translucent substrate having a predetermined value X or less, a determination coefficient R calculated from the difference shape (difference data) as shown in step S12 of FIG. ² is calculated. Then, as shown in step S13 in FIG. 4, the translucent substrate of the coefficient of determination R ² is 0.9 or more is selected as a substrate for a high flat mask blank ends. A translucent substrate of less than 0.9 is considered to be used as a mask blank substrate for middle layer or rough layer as a low / medium flat product, or it is returned to the polishing step of step S3, or through a local processing step After step S4, the same steps are performed again, or the light transmitting substrate is discarded. In this method, the transfer accuracy of the transfer mask manufactured using the mask blank substrate is selected because the selection is made not only by determining one abnormal point but also by using the fitting degree with the optical flat surface of the entire shape. Is expensive.

The steps S12 and S13 may be performed at any stage as long as the step S7 for calculating the difference shape is performed. For example, step S12 may be performed between step S8 and step S9, and step S12 and step S13 may be performed before step S9. The calculation and predetermined value of the PV value and the comparison determination (reference value) (step S8 and S9), also change the order of the comparison determination (step S12 and S13) between the calculated and reference values of the coefficient of determination R ² Is possible. Note that it is preferable to retain the virtual reference plane information determined here and reflect the information on the lens aberration correction function of the exposure apparatus because exposure on the conjugate plane of the projection lens can be easily performed.

[Manufacturing method of mask blank]
The mask blank manufacturing method of the present invention includes a step of providing a transfer pattern forming thin film on one main surface of the mask blank substrate manufactured by the above-described mask blank substrate manufacturing method.

  What is important here is the control of the stress of the thin film. When the mask blank substrate is distorted by the stress of the thin film, the flatness of the substrate surface changes. The deformation of the main surface of the substrate due to the film stress is a relatively simple deformation of a concentric quadratic curved surface and can be dealt with by correcting the aberration of the exposure machine. On the other hand, if the stress of the thin film is too large, A problem arises in that the thin film pattern is displaced during patterning of the thin film performed when the transfer mask is manufactured. When the relationship between the amount of change in mechanical flatness and the film stress in the inner region of a square having a side of 132 mm on the basis of the center of the mask blank was examined, the amount of change in flatness was 10 nm, 20 nm, 25 nm, 30 nm, 40 nm, and 50 nm. The film stresses corresponding to were 55 MPa, 110 MPa, 137 MPa, 165 MPa, 220 MPa, and 275 MPa, respectively. From this result, it is understood that the stress of the thin film is desirably 275 MPa or less, more desirably 165 MPa or less, and even more desirably 110 MPa or less.

  Therefore, it is necessary to adjust the film stress of the thin film. Examples of the method include a method of performing a heat treatment (annealing) and a method of performing a light irradiation process of irradiating the thin film with high energy light such as a flash lamp. and so on. If thin film formation is performed while paying attention to this film stress adjustment, an optically high flat mask blank of 1/8 of the exposure wavelength λ can be manufactured, and a transfer mask manufactured using the mask blank can be manufactured. When the exposure is performed, the depth of focus, the positional deviation, and the resolution are excellent, and the circuit characteristics of a semiconductor device manufactured using the same are also stabilized.

The mask blank manufactured by the mask blank of the present invention and the mask blank manufacturing method of the present invention can be applied with the following configurations (1) to (3).
(1) Binary mask blank provided with a light-shielding film made of a material containing a transition metal The binary mask blank has a light-shielding film (a thin film for forming a transfer pattern) on a light-transmitting substrate. The film is made of a material containing a transition metal alone or a compound thereof such as chromium, tantalum, ruthenium, tungsten, titanium, hafnium, molybdenum, nickel, vanadium, zirconium, niobium, palladium, rhodium. For example, a light-shielding film composed of chromium or a chromium compound in which one or more elements selected from elements such as oxygen, nitrogen, and carbon are added to chromium. Further, for example, a light shielding film composed of a tantalum compound in which one or more elements selected from elements such as oxygen, nitrogen, and boron are added to tantalum. Such binary mask blanks include a light shielding film having a two-layer structure of a light shielding layer and a front surface antireflection layer, or a three-layer structure in which a back surface antireflection layer is added between the light shielding layer and the substrate. Moreover, it is good also as a composition gradient film | membrane from which the composition in the film thickness direction of a light shielding film differs continuously or in steps.

(2) Phase shift mask blank provided with a light semi-transmissive film made of a material containing a compound of transition metal and silicon (including transition metal silicide, particularly molybdenum silicide) As such a phase shift mask blank, a light-transmitting substrate (glass A half-tone type phase shift mask having a light semi-transmissive film (thin film for forming a transfer pattern) on a substrate) and having a shifter portion by patterning the light semi-transmissive film is produced. The In such a phase shift mask, in order to prevent a pattern defect of the transferred substrate due to the light semi-transmissive film pattern formed in the transfer region based on the light transmitted through the light semi-transmissive film, the light semi-transmissive is formed on the light-transmissive substrate. The thing which has a form which has a film | membrane and the light shielding film (light shielding zone) on it is mentioned. In addition to halftone phase shift mask blanks, mask blanks for Levenson type phase shift masks and enhancer type phase shift masks, which are substrate digging types in which a translucent substrate is dug by etching or the like to provide a shifter portion. Is mentioned.

  The light-semitransmissive film of the halftone phase shift mask blank transmits light having an intensity that does not substantially contribute to exposure (for example, 1% to 30% with respect to the exposure wavelength). It has a phase difference (for example, 180 degrees). The light semi-transmissive portion is transmitted through the light semi-transmissive portion by a light semi-transmissive portion patterned with the light semi-transmissive film and a light transmissive portion that does not have the light semi-transmissive film and transmits light having an intensity substantially contributing to exposure. By making the phase of the light a substantially inverted relationship with respect to the phase of the light transmitted through the light transmission part, the light passes through the vicinity of the boundary between the light semi-transmission part and the light transmission part, and is caused by diffraction phenomenon The light that has entered the area of the other party cancels each other, and the light intensity at the boundary is made almost zero, thereby improving the contrast or resolution of the boundary.

  This light semi-transmissive film is made of a material containing a compound of, for example, a transition metal and silicon (including a transition metal silicide), and includes a material mainly composed of these transition metal and silicon, and oxygen and / or nitrogen. . As the transition metal, molybdenum, tantalum, tungsten, titanium, hafnium, nickel, vanadium, zirconium, niobium, palladium, ruthenium, rhodium, chromium, or the like is applicable. In the case of having a light-shielding film on the light semi-transmissive film, the material of the light semi-transmissive film contains a transition metal and silicon. It is preferable to have chromium (having etching resistance), particularly chromium, or a chromium compound in which elements such as oxygen, nitrogen, and carbon are added to chromium.

  Since the Levenson-type phase shift mask is manufactured from a mask blank having the same configuration as the binary mask blank, the configuration of the thin film for forming the transfer pattern is the same as that of the light shielding film of the binary mask blank. The light semi-transmissive film of the mask blank for the enhancer-type phase shift mask transmits light having an intensity that does not substantially contribute to exposure (for example, 1% to 30% with respect to the exposure wavelength). This is a film having a small phase difference generated in the exposure light (for example, a phase difference of 30 degrees or less, preferably 0 degrees), and this is different from the light semi-transmissive film of the halftone phase shift mask blank. The material of this light semi-transmissive film includes the same elements as the light semi-transmissive film of the halftone type phase shift mask blank, but the composition ratio and film thickness of each element have a predetermined transmittance and predetermined ratio to the exposure light. The phase difference is adjusted to be small.

(3) Binary mask blank provided with a light shielding film made of a material containing a compound of transition metal, transition metal and silicon (including transition metal silicide, particularly molybdenum silicide). This light shielding film (thin film for forming a transfer pattern) A material containing a compound of a metal and silicon, and a material mainly composed of at least one of these transition metals and silicon and oxygen or nitrogen can be given. Examples of the light shielding film include a material mainly composed of a transition metal and at least one of oxygen, nitrogen, and boron. As the transition metal, molybdenum, tantalum, tungsten, titanium, hafnium, nickel, vanadium, zirconium, niobium, palladium, ruthenium, rhodium, chromium, or the like is applicable. In particular, when the light shielding film is formed of a molybdenum silicide compound, it has a two-layer structure of a light shielding layer (MoSi, etc.) and a surface antireflection layer (MoSiON, etc.), and the back surface antireflection between the light shielding layer and the substrate. There is a three-layer structure to which layers (MoSiON, etc.) are added. Moreover, it is good also as a composition gradient film | membrane from which the composition in the film thickness direction of a light shielding film differs continuously or in steps.

  In order to form a fine pattern by reducing the thickness of the resist film, an etching mask film may be provided over the light shielding film. This etching mask film has etching selectivity (etching resistance) with respect to etching of the light-shielding film containing transition metal silicide, and in particular, chromium, or a chromium compound in which elements such as oxygen, nitrogen, and carbon are added to chromium. It is preferable to use a material. At this time, by providing the etching mask film with an antireflection function, the transfer mask may be manufactured with the etching mask film remaining on the light shielding film.

In the above (1) to (3), the light-shielding film or the light semi-transmissive film is interposed between the light-transmitting substrate (glass substrate) and the light-shielding film or between the light semi-transmissive film and the light-shielding film. An etching stopper film having etching resistance may be provided. The etching stopper film may be a material that can peel off the etching mask film at the same time when the etching stopper film is etched.

[Transfer Mask Manufacturing Method]
The method for manufacturing a transfer mask according to the present invention includes a step of forming a transfer pattern on a thin film of the mask blank manufactured by the above-described mask blank manufacturing method. Hereinafter, the process of manufacturing the transfer mask from the mask blank will be described. The mask blank used here is the phase shift mask blank of (2) described above, and has a structure in which a light semi-transmissive film (a thin film for forming a transfer pattern) and a light shielding film are sequentially laminated on a translucent substrate. Prepare. In addition, the method of manufacturing the transfer mask (phase shift mask) is an example, and the transfer mask (phase shift mask) can be manufactured even if some procedures are changed.

  First, a resist film is formed on the light shielding film of the phase shift mask blank by a spin coating method. For this resist film, a chemically amplified resist for electron beam exposure drawing is preferably used. Next, a transfer pattern to be formed on the light translucent film is exposed and drawn with an electron beam on the resist film, and a predetermined process such as development is performed to form a resist pattern having the transfer pattern. Subsequently, dry etching using the resist pattern as a mask is performed on the light shielding film to form a transfer pattern to be formed on the light semi-transmissive film on the light shielding film. After dry etching, the resist pattern is removed. Next, the light semi-transmissive film is dry-etched using a light-shielding film having a transfer pattern as a mask to form a transfer pattern on the light semi-transmissive film. Subsequently, a resist film is formed again by a spin coating method, and a pattern (pattern such as a light shielding band) to be formed on the light shielding film is exposed and drawn with an electron beam, and a predetermined process such as development is performed to form a resist pattern. . The light shielding film is subjected to dry etching using a resist pattern having a pattern such as a light shielding band as a mask to form a pattern such as a light shielding band on the light shielding film. Then, a predetermined cleaning process or the like is performed to complete a transfer mask (phase shift mask).

  The optical flatness of the substrate exposed surface of the transfer mask manufactured by this method (the substrate main surface of the opening where the thin film for pattern formation is not left) is 1/8 of the desired predetermined value X or the exposure wavelength λ. It was possible to manufacture a transfer mask that was extremely high as follows and had sufficient wavefront control. Since the wavefront control is sufficient, when exposure is performed using this transfer mask, the focal depth, the positional deviation, and the resolution are excellent, and the circuit characteristics of a semiconductor device manufactured using the same are also stable.

  The present invention is effective regardless of the type of transfer mask, and is effective for both binary masks, halftone phase shift masks, enhancer masks, and Levenson type phase shift masks.

  Among them, the binary mask is most widely used, and it is not necessary to make a light shielding band by a special method, so that the mass production effect is great. In addition, with regard to the halftone phase shift mask, since exposure light is transmitted from not only the pattern opening but also from the light semi-transmission part, the influence of the wavefront control on the transfer performance is great, so the transfer mask manufactured by this method Is particularly effective.

[Exposure Method and Device Manufacturing Method Using the Same]
Here, an exposure method using the mask manufactured by the above-described method and a device manufacturing method using the same will be described.
First, the outline of the optical system portion of the exposure apparatus will be described with reference to FIG. 9 which is a schematic view of the apparatus configuration showing the outline of the apparatus configuration as a sectional view. The optical system portion of the exposure apparatus has the following configuration. Exposure light 32 is emitted from the light source 31, and the exposure light is irradiated onto the transfer mask 34 via the illumination optical system 33. The exposure light transmitted through the transfer mask 34 is irradiated onto the wafer 40 placed on the wafer stage 39 via the projection lenses 35 and 38 to perform exposure. In general, a movable diaphragm is installed at the pupil 36 between the projection lenses 35 and 38 so that the numerical aperture (NA) of the projection lens can be adjusted. Although each of the projection lenses 35 and 38 is depicted as a single lens in this figure, it actually comprises a large number of lens groups, and their mutual positions can be partially moved so that the lower order is the center. It incorporates a mechanism that can correct the lens aberration. In addition, a phase filter 37 is incorporated in the vicinity of the pupil 36. By adjusting the phase filter 37, high-order lens aberrations, particularly high-order aberrations by lens partial heating can be corrected in real time. ing.

  This low-order lens aberration correction includes up to six Zernike polynomials such as tilt and astigmatism. That is, there is a function of correcting a term defined by a Zernike polynomial in which the order of the variable related to the radius is composed only of terms of the second order or less and the order of the variable related to the radius includes one or more of the second order terms. When the transfer mask of the present invention is used, the main surface of the transfer mask can be brought to a position conjugate with the projection lens 35, 38 on the wafer 40 by this low-order lens aberration correction. The main surface is an optically flat surface. For this reason, when exposure is performed using this transfer mask, the depth of focus, the positional deviation, and the resolution are excellent, and the circuit characteristics of a semiconductor device manufactured using the same are also stabilized.

  The illumination optical system 33 incorporates a zooming mechanism, a movable multi-mirror optical system, and the like so that illumination of a desired shape can be set. Normal illumination consists of an illumination part and a light-shielding part (a part where light is blocked). The illumination unit is a circle centered at the center, and the illumination condition is defined by the size of the circle. (This is called coherency). On the other hand, dipole illumination has recently been frequently used mainly in memory devices. A small circular illumination part is arranged away from the center on the X axis, and a light shielding part around it is called an X dipole. Deformed dipole illumination in which the illumination unit is transformed from a circular shape into a fan shape may also be used. This X dipole illumination has a high resolution in the X direction and is suitable for forming a dense and fine pattern in the X direction. In the X-dipole illumination, a mechanism for increasing the irradiation efficiency by concentrating the illumination light to the illumination unit, usually the light that travels to the light shielding unit (field unit), is incorporated in the illumination unit 33. Therefore, in the projection lenses 35 and 38, intense exposure light intensively passes through a part of the lens, and partial lens heating occurs. The lens is distorted by this heat, and complicated high-order lens aberration is generated. In device manufacturing, not only X dipole illumination but also Y dipole illumination is frequently used. This case is suitable for forming a dense and fine pattern in the Y direction. In a memory, a word line and a bit line are particularly required to form a fine pattern, but generally both of them are arranged in an orthogonal relationship, that is, a dense wiring in the X direction and a dense wiring in the Y direction. . For this reason, both X dipole illumination and Y dipole illumination are frequently used. Also, illumination is often used for pattern formation of various shapes such as logic patterns. Since various illuminations are used in this way, there are various places where lens heating occurs, and there are various occurrences of lens high-order aberrations. Because of the heating, the high-order lens aberration correction needs to follow changes with time, even when a large amount of processing is performed continuously after the start of exposure. This higher-order aberration correction is a third-order term in the radial direction when expressed by a Zernike polynomial, and the term is sequentially corrected. Therefore, even if the optical flat surface of the transfer mask is corrected up to a third or higher order Zernike polynomial term in the radial direction, it is only optically flat in a certain illumination state at a certain point in time. Not enough wavefront control. Therefore, as described above, the virtual reference plane of the transfer mask is formed by a Zernike polynomial in which the order of the variable related to the radius is composed only of the terms of the second order or less and the order of the variable of the radius is one or more of the second order terms. It is efficient and effective to set the optical flat surface. However, this is the case when the exposure apparatus is applied to a general-purpose layer or device, and when higher accuracy is required even if it is specifically used for a specific layer or requires sequential correction, it is up to higher order terms. It is effective to capture.

Hereinafter, three application examples of exposure application will be described below.
<Example of sub-peak transfer avoidance of halftone phase shift mask>
The following shows an example in which sub-peak transfer failure, which often becomes a problem when a halftone phase shift mask is used, is improved. When a positive resist is used as a wafer resist, a resist dent may be generated at a position to be a resist portion due to a sub-peak phenomenon. This indentation penetrates when the film to be processed is etched, and becomes a nest of defects in the device circuit, thereby reducing the manufacturing yield of the device or causing the circuit operation to become unstable. Although this problem can be solved if the resist film thickness can be increased, it is difficult to increase the resist thickness due to problems such as resist resolution and pattern collapse. One solution to this problem is to give low-order aberrations to the lens and make it difficult to produce sub-peaks. On the other hand, this method reduces the exposure latitude, particularly the focus tolerance. Therefore, a stricter flatness is required for the mask blank substrate and the transfer mask. Therefore, the mask blank substrate and the transfer mask of the present embodiment are used, and the projection lens is defined by a Zernike polynomial in polar coordinates so as to give an optical flat surface to the mask blank substrate and the transfer mask. Aberration correction was added, and further, low-order correction for preventing sub-peak transfer was added and exposure was performed. As a result, it was possible to avoid the problem of sub-peak transfer when the halftone phase shift mask was used while securing the required focus latitude. This is because an optical flatness of λ / 8 or less is achieved in the mask blank substrate and transfer mask of the present embodiment.

<Application example of exposure apparatus QC>
Here, an example applied to quality control (QC: Quality Control) of an exposure apparatus is shown. As described above, the higher-order aberration correction of the projection lens of the exposure apparatus is sequentially adjusted according to the exposure situation. However, the order of the variable related to the radius is composed only of terms of the second order or less, and the radius A low-order term such that the order of such a variable is described by a Zernike polynomial including one or more second-order terms varies depending on the application layer of the semiconductor device as described above, but is a reference value in terms of exposure apparatus management. Should be operated semi-fixed. Normally, these low-order lens aberration corrections do not change with time, but they change when a power failure, a change in the temperature environment of the exposure apparatus due to an abnormal stop of the temperature adjustment chamber, an earthquake, or the like occurs. Therefore, a low-order lens aberration correction management QC of the exposure apparatus is required, but this QC requires a reference mask that is extremely flat and has a uniform wavefront. Since this is an evaluation of lens aberration, the reference mask used there is required to have a flatness with an optical flatness of λ / 8 or less similar to that required for an advanced optical measuring instrument. Since the mask blank substrate and the transfer mask according to the present embodiment satisfy this requirement, they are optimal for adjusting the lens aberration correction function of the exposure apparatus.

<Example of exposure apparatus lens aberration correction function adjustment application>
Here, an example applied to the lens aberration correction function adjustment of the exposure apparatus is shown. As described above, the lens aberration correction function is incorporated in the exposure apparatus. In adjusting and evaluating this function, a reference mask that is extremely flat and does not become a source of wavefront aberration is required. Since this is an evaluation of lens aberration, the reference mask used there is required to have a flatness with an optical flatness of λ / 8 or less similar to that required for an advanced optical measuring instrument. Since the mask blank substrate and the transfer mask according to the present embodiment satisfy this requirement, they are optimal for adjusting the lens aberration correction function of the exposure apparatus.

[Manufacture of mask blank substrates]
According to the mask blank substrate manufacturing method of the present embodiment, three mask blank substrate samples were prepared and evaluated. Sample A is the example, and samples B and C are the reference examples. Until the mask blank substrate selection, all three sheets were manufactured in the same process shown below.

First, a synthetic quartz glass substrate (size: 152.4 mm × 152.4 mm, thickness: 6.75 mm) is cut out, the end surface of the synthetic quartz glass substrate is chamfered and ground, and further contains cerium oxide abrasive grains. Rough polishing treatment and precision polishing were performed with a polishing liquid. Thereafter, this glass substrate was set on a carrier of a double-side polishing apparatus, and ultraprecision polishing was performed under the following conditions.
Polishing pad: Soft polisher (suede type)
Polishing liquid: colloidal silica abrasive grains (average particle diameter 100 nm) and water processing pressure: 50 to 100 g / cm ²
Processing time: 60 minutes After the ultraprecision polishing was completed, the glass substrate was immersed in a dilute hydrofluoric acid solution to perform cleaning to remove the colloidal silica abrasive grains. Thereafter, scrub cleaning was performed on the main surface and the end surface of the glass substrate, followed by spin cleaning with pure water and spin drying to prepare three glass substrates whose surfaces were polished. The surface shape (flatness) of the glass substrate was measured in a free-standing state using a flatness measuring device (UltraFlat 200M manufactured by Corning Tropel).

The actual measurement data is shown in FIG. (A), (b), and (c) show the contour distribution map of the main surface shape measured in a circular region with a diameter of 104 mm with reference to the center of the glass substrate in the case of Samples A, B, and C, respectively. Here, the vertical axis and the horizontal axis in the figure are displayed with the radius 52 mm normalized to 1. The Z value shown on the right side represents the height, and its unit is μm.
The measured values of the mechanical flatness of the samples A, B, and C, that is, the mechanical flatness in the free state before chucking to the mask stage, is the sum of the absolute value at the highest point and the absolute value at the lowest point. In the case of a 132 mm × 132 mm area where the transfer exposure area (shot area) fits, expressed in (Total Indicator Reading), the order is 358 nm, 195 nm, and 266 nm, and in the case of a 104 mm diameter circular area, 190 nm, 87 nm and 147 nm.

  For these three samples, the shape of the main surface (transfer main surface) on the side where the thin film is provided when the sample was chucked to the mask stage of the exposure apparatus was simulated by the method described in the embodiment. Further, a virtual reference plane defined by a Zernike polynomial in which the order of the variable related to the radius is composed only of terms of the second order or less and the order of the variable related to the radius includes one or more of the second order terms is obtained, and the simulation The difference shape (difference data) between the measured main surface shape and the virtual reference plane was obtained. The results expressed in contour map are shown in FIGS. 6, 7 and 8 in the order of samples A, B and C. (A) of each figure is the shape of the main surface (transfer main surface) obtained by simulation of the shape after chucking, (b) is the shape of the virtual reference surface defined by the Zernike polynomial, and (c) is the difference between the two. This is the shape (difference data). Here, the vertical axis and the horizontal axis in the figure are displayed with the radius 52 mm normalized to 1. The Z value shown on the right side represents the height, and its unit is μm. By the way, as the Zernike polynomial here, the one expressed by the University of Arizona was used, and the virtual reference plane was generated by fitting from the first to sixth terms so as to approximate the simulation shape after the mask stage chuck. However, as described above, this is an example, and the Zernike polynomial is not limited to the one represented by the University of Arizona. Even when other notation methods such as standard Zernike notation and fringe Zernike notation are applied, the same virtual reference plane can be obtained.

  Table 1 shows each term of the Zernike polynomial in the University of Arizona used in this example. Each term is expressed in a polar coordinate system in which the radius is ρ and the phase (azimuth angle) is θ. In Table 1, j is the number of a term (jth term), and Zj (ρ, θ) is the content of the term of that number. In Table 1, up to the 10th term is shown for reference, but up to the 6th term is used in this example.

In the case of Sample A, the mechanical flatness obtained by simulation assuming that the mask blank substrate was chucked on the mask stage was 97 nm in a 104 mm diameter circular region expressed by a PV value which is one of the flatness indexes. It was. On the other hand, in terms of the optical flatness according to the present invention calculated from the difference shape (difference data) between the main surface shape after mask stage chucking by simulation and the virtual reference surface shape described by the Zernike polynomial, the PV value Was 24 nm, which was a value equal to or less than 25 nm (rounded up after the decimal point), which is λ / 8 of the exposure wavelength λ (193 nm) of ArF exposure. That is, by applying the present invention, it was possible to obtain extremely high flatness of λ / 8 or less in a state of being chucked on a mask stage which is an environment at the time of exposure. The coefficient of determination R ² calculated from the difference shape between the virtual reference plane shape (difference data) written in the main surface shape and Zernike polynomials after the mask stage chuck simulation is 0.974, 0.9 or higher It was a sufficient value. In addition, the mechanical flatness after the mask stage chuck obtained by simulation of a 132 mm × 132 mm region where the transfer exposure region (shot region) fits was 172 nm. In Sample A, the flatness of the 104 mm diameter circular region and the 132 mm × 132 mm region, and the determination coefficient R ² had extremely sufficient values.

In the case of sample B, the mechanical flatness obtained by simulation assuming that the mask blank substrate was chucked on the mask stage was 54 nm in a circular region having a diameter of 104 mm, expressed as a PV value. On the other hand, in terms of the optical flatness according to the present invention calculated from the difference shape (difference data) between the main surface shape after mask stage chucking by simulation and the virtual reference surface shape described by the Zernike polynomial, the PV value Was 24 nm, which was a value equal to or less than 25 nm (rounded up after the decimal point), which is λ / 8 of the exposure wavelength λ (193 nm) of ArF exposure. That is, by applying the present invention, it was possible to obtain extremely high flatness of λ / 8 or less in a state of being chucked on a mask stage which is an environment at the time of exposure. On the other hand, the coefficient of determination R ² calculated from the difference shape between the virtual reference plane shape (difference data) written in the main surface shape and Zernike polynomials after the mask stage chuck simulation is 0.745, 0.9 The fitting divergence from the optical flat surface (virtual reference surface) in the entire circular region having a diameter of 104 mm was conspicuous. The mechanical flatness after the mask stage chuck obtained by simulation of a 132 mm × 132 mm region where the transfer exposure region (shot region) fits was 152 nm.

In the case of sample C, the mechanical flatness obtained by simulation assuming that the mask blank substrate was chucked on the mask stage was 99 nm in a circular region having a diameter of 104 mm, expressed as a PV value. On the other hand, in terms of the optical flatness according to the present invention calculated from the difference shape (difference data) between the main surface shape after mask stage chucking by simulation and the virtual reference surface shape described by the Zernike polynomial, the PV value Was 48 nm, which was about λ / 4 of the exposure wavelength λ (193 nm) of ArF exposure. The coefficient of determination R ² calculated from the difference shape between the virtual reference plane shape (difference data) written in the main surface shape and Zernike polynomials after the mask stage chuck simulation is 0.820, 0.9 Below. The mechanical flatness after the mask stage chuck obtained by simulation of a 132 mm × 132 mm region where the transfer exposure region (shot region) fits was 245 nm. Sample C is a sample that is less flat than Samples A and B even when the method of the present invention is used.

[Manufacture of mask blanks]
Here, the example which manufactured the mask blank for halftones is shown. First, a mask blank substrate (Example Samples A and B) which was manufactured by the above-described method and passed the selection standard was prepared, and a light semi-transmissive film made of nitrided molybdenum and silicon was formed thereon. Specifically, using a mixed target of molybdenum (Mo) and silicon (Si) (Mo: Si = 10 mol%: 90 mol%), mixing argon (Ar), nitrogen (N ₂ ), and helium (He). In a gas atmosphere (gas flow ratio Ar: N ₂ : He = 5: 49: 46), a gas pressure of 0.3 Pa, a DC power source power of 3.0 kW, and reactive sputtering (DC sputtering), molybdenum, silicon, and A MoSiN film made of nitrogen was formed to a thickness of 69 nm. Next, the substrate on which the MoSiN film was formed was subjected to heat treatment using a heating furnace in the atmosphere at a heating temperature of 450 ° C. and a heating time of 1 hour. This MoSiN film had an transmittance of 6.16% and a phase difference of 184.4 degrees in an ArF excimer laser.

Next, a light shielding film was formed on the light semi-transmissive film. Specifically, a chromium (Cr) target is used as a sputtering target, and a mixed gas atmosphere of argon (Ar), carbon dioxide (CO ₂ ), nitrogen (N ₂ ), and helium (He) (gas pressure 0.2 Pa, Gas flow ratio Ar: CO ₂ : N ₂ : He = 20: 35: 10: 30), the power of the DC power source is 1.7 kW, and a 30 nm thick CrOCN layer is formed by reactive sputtering (DC sputtering). Filmed. Subsequently, a mixed gas atmosphere of argon (Ar) and nitrogen (N ₂ ) (gas pressure 0.1 Pa, gas flow ratio Ar: N ₂ = 25: 5) was set, and the power of the DC power source was set to 1.7 kW. A 4 nm thick CrN layer was formed by reactive sputtering (DC sputtering). Finally, a mixed gas atmosphere of argon (Ar), carbon dioxide (CO ₂ ), nitrogen (N ₂ ), and helium (He) (gas pressure 0.2 Pa, gas flow ratio Ar: CO ₂ : N ₂ : He = 20: 35: 5: 30), the power of the DC power source is 1.7 kW, and a CrOCN layer having a film thickness of 14 nm is formed by reactive sputtering (DC sputtering). A chromium-based light shielding film was formed. Thereafter, a heat treatment at 280 ° C. for 15 minutes was applied to reduce the film stress to near zero.

  The optical flatness of the surface of the mask blank manufactured by this method is 25 nm or less which is λ / 8 of the exposure wavelength λ (193 nm) of ArF exposure in the exposure state chucked on the mask stage. A mask blank with sufficient wavefront control.

  In the above mask blank manufacturing method, the flatness of the mask blank substrate is measured, the main surface shape when it is chucked on the mask stage is simulated, and the virtual reference plane and differential shape data are taken to obtain the optical After performing the flatness screening, a thin film was formed to produce a mask blank, but the order of thin film formation and optical flatness screening may be reversed. That is, after forming a thin film on the mask blank substrate, measure the flatness of the mask blank, simulate the main surface shape when chucked on the mask stage, take the difference shape data and the virtual reference plane, Optical flatness sorting may be performed.

[Manufacture of transfer masks and semiconductor devices]
Here, the transfer mask was manufactured by patterning the thin film on the mask blank manufactured by the method described above. The manufacturing process of the transfer mask is the same as the method described in the above [Transfer mask and manufacturing method thereof], and the description thereof will be omitted. In the exposure using this transfer mask, the applied Zernike polynomial parameters are reflected in the lens aberration correction function of the exposure apparatus so that the transfer main surface 54 of the transfer mask becomes an optical flat surface. .

  The optical flatness of the transfer main surface 54 of the transfer mask manufactured by this method is extremely high at 1/8 or less of the exposure wavelength λ in the exposure state chucked on the mask stage, and sufficient wavefront control is possible. It has become possible to produce a transfer mask that has been made. Since the wavefront control is sufficient, when exposure is performed using this transfer mask, the focal depth, the positional deviation, and the resolution are excellent, and the circuit characteristics of a semiconductor device manufactured using the same are also stable.

  In the mask blank substrate and mask blank of the present invention, in consideration of the load on the aberration correction function in the exposure apparatus, the surface shape when the aberration correction function with the second order of the Zernike polynomial is used is λ. / 8 or less. However, even if more load is allocated to the correction of wavefront aberration related to the surface shape of the substrate or mask blank due to the performance improvement of the aberration correction function of the exposure apparatus and the quality improvement of the projection lens, the influence on the exposure transfer is not affected In a small case, the range of the virtual reference plane is defined by a Zernike polynomial in which the order of the variable related to the radius is composed only of the third order or less and the order of the variable related to the radius is one or more of the third order terms. It may be extended to the surface shape. By using the surface shape defined by such a Zernike polynomial as a virtual reference plane, the yield in manufacturing the mask blank substrate and the mask blank of the present invention can be greatly improved.

[Comparative example]
Here, the case where it evaluated by the prior art by a mechanical flat surface using the three samples A, B, and C produced in the Example is shown. Accordingly, the object itself is the same as that of the embodiment from the stage of the mask blank substrate to the transfer mask. The simulation of the shape of the main surface (transfer main surface) of the mask blank substrate at the time of chucking to the mask stage is performed in the same manner as in the example. The difference from the embodiment is whether a mechanical flat surface or an optical flat surface is used. Calculates a virtual reference plane described by the Zernike polynomial in the embodiment, and obtains difference shape data from the main surface shape (transfer main surface shape) of the mask blank substrate when chucked on the mask stage obtained by simulation However, the process of selecting from the PV calculated value is not performed in the comparative example. In the comparative example, a mechanical flat surface obtained by chuck simulation is used. Therefore, even in exposure using this transfer mask, lens aberration correction specific to this transfer mask, that is, reflection of Zernike polynomial parameters for making an optical flat surface is not performed on the exposure apparatus.

  As described above, the mechanical surface shape of the 104 mm diameter region of the main surface (transfer main surface) when the mask blank substrates of Samples A, B, and C are chucked on the mask stage is shown in FIGS. (A). When the mechanical flatness was expressed using PV values, it was 97 nm for sample A, 54 nm for sample B, and 99 nm for sample C in the case of a 104 mm diameter circular region. Even in the case of the best sample B, it was 54 nm, which was more than double the value of 24 nm of the example using the optical flat surface. The value of 54 nm was about λ / 4 of the exposure wavelength λ (193 nm) of ArF exposure, and sufficient wavefront control was not possible. Samples A and C are nearly double the value, which corresponds to about λ / 2 and is insufficient. In the case of a 132 mm × 132 mm region in which the transfer exposure region (shot region) fits, it was 172 nm for sample A, 152 nm for sample B, and 245 nm for sample C. Because of this insufficient wavefront control, when exposure is performed using this transfer mask, in-plane distribution occurs in the depth of focus, misalignment, and resolution, and the circuit characteristics of the semiconductor device manufactured using the in-plane distribution Became unstable.

  DESCRIPTION OF SYMBOLS 1 ... Mask blank substrate, 2 ... Mechanical flat surface, 3 ... Substrate main surface, 4 ... Optical flat surface, 31 ... Light source, 32 ... Exposure light, 33 ... Illumination optical system, 34 ... Mask, 35 ... Projection lens 36 ... pupil, 37 ... phase filter, 38 ... projection lens, 39 ... wafer stage, 40 ... wafer, 50 ... mask stage, 51 ... transfer mask, 52 ... suction part, 53 ... mask pattern, 54 ... main surface, 55: Contact portion.

Claims (20)

A mask blank substrate manufacturing method used for manufacturing a mask blank in which a thin film for forming a transfer pattern is provided on one main surface of a translucent substrate having a pair of opposing main surfaces,
A shape measuring step for obtaining a surface shape of the main surface on the side where the thin film is provided in the light-transmitting substrate;
A simulation step for obtaining a chucked main surface shape by simulation when the translucent substrate is chucked on a mask stage of an exposure apparatus;
The main surface shape after chucking is a calculation area inside a circle having a predetermined diameter with respect to the center of the translucent substrate, and a virtual reference plane defined by a Zernike polynomial expressed in a polar coordinate system Performing shape fitting, obtaining difference data between the post chuck main surface shape and the virtual reference plane;
A mask blank substrate comprising a step of selecting the translucent substrate having a surface shape in which a difference between a maximum height and a minimum height in the calculation area of the difference data is equal to or less than a predetermined value. Manufacturing method.   2. The method of manufacturing a mask blank substrate according to claim 1, wherein the predetermined value is 1/8 of a wavelength [lambda] of exposure light used for exposure transfer.   3. The method of manufacturing a mask blank substrate according to claim 1, wherein the predetermined diameter is 104 mm.   The virtual reference plane has a shape defined by a Zernike polynomial in which the order of the variable related to the radius is composed only of terms of the second order or less, and the order of the variable related to the radius includes one or more of the second order terms. The method for manufacturing a mask blank substrate according to any one of claims 1 to 3. Method for producing a substrate for a mask blank according to any one of claims 1 to 4, characterized in that the coefficient of determination R ² calculated from the difference shape is 0.9 or more.   6. The flatness calculated in a rectangular inner region having a side of 132 mm with respect to the center of the translucent substrate in the main surface shape after chucking is 0.2 μm or less. The manufacturing method of the board | substrate for mask blanks in any one.   A step of associating the selected light-transmitting substrate with information related to the Zernike polynomial in the virtual reference plane obtained by performing shape fitting on the light-transmitting substrate and recording the information on a recording apparatus. A method for manufacturing a mask blank substrate according to any one of claims 1 to 6.   A mask blank comprising: a step of providing a thin film for forming the transfer pattern on one main surface of a mask blank manufactured by the method for manufacturing a mask blank substrate according to claim 1. Production method. A mask blank manufacturing method in which a thin film for forming a transfer pattern is provided on one main surface of a translucent substrate having a pair of main surfaces facing each other,
A shape measuring step for obtaining a surface shape of the thin film in the mask blank;
A simulation process for obtaining a chucked surface shape by simulation when the mask blank is chucked on a mask stage of an exposure apparatus;
The post-chuck surface shape is calculated with respect to a virtual reference plane defined by a Zernike polynomial expressed in a polar coordinate system in a calculation region inside a circle having a predetermined diameter with respect to the center of the translucent substrate. Performing a fitting and obtaining difference data between the post chuck surface shape and the virtual reference plane;
And a step of selecting the mask blank having a surface shape in which a difference between a maximum height and a minimum height in the calculation area of the difference data is equal to or less than a predetermined value.   10. The method of manufacturing a mask blank according to claim 9, wherein the predetermined value is 1/8 of a wavelength [lambda] of exposure light used for exposure transfer.   11. The method for manufacturing a mask blank according to claim 9, wherein the predetermined diameter is 104 mm.   The virtual reference plane has a shape defined by a Zernike polynomial in which the order of the variable related to the radius is composed only of terms of the second order or less, and the order of the variable related to the radius includes one or more of the second order terms. The method for producing a mask blank according to claim 9, wherein the mask blank is produced. The method for manufacturing a mask blank according to claim 9, wherein a determination coefficient R ² calculated from the difference shape is 0.9 or more.   14. The flatness calculated in a rectangular inner region having a side of 132 mm on the basis of the center of the translucent substrate in the main surface shape after chucking is 0.2 μm or less. The manufacturing method of the mask blank in any one.   The step of associating the selected mask blank with information related to the Zernike polynomial on the virtual reference plane that has been shape-fitted on the mask blank, and recording the information on a recording apparatus. The manufacturing method of the mask blank in any one of 9 to 14.   A method for producing a transfer mask, comprising a step of forming a transfer pattern on the thin film of the mask blank produced by the method for producing a mask blank according to claim 8.   17. A semiconductor device manufacturing method comprising: an exposure step of chucking the transfer mask according to claim 16 on a mask stage of an exposure apparatus and transferring the transfer pattern of the transfer mask onto a semiconductor substrate by lithography. Method.   The exposure apparatus has a function of performing wavefront correction of a shape defined by Zernike polynomials on a wavefront of transmitted light transmitted from a transfer pattern of the transfer mask, and a mask stage used in the simulation process 18. The method of manufacturing a semiconductor device according to claim 17, wherein the semiconductor device has the same shape as that of the semiconductor device. Providing a mask blank by providing the thin film for forming the transfer pattern on one main surface of the mask blank substrate manufactured by the method for manufacturing a mask blank substrate according to claim 7;
Forming a transfer pattern on the transfer pattern forming thin film of the manufactured mask blank and manufacturing a transfer mask;
An exposure step of chucking the produced transfer mask on a mask stage of an exposure apparatus, and transferring the transfer pattern of the transfer mask onto a semiconductor substrate by a lithography method,
The exposure apparatus has a function of performing wavefront correction of a shape defined by Zernike polynomials on a wavefront of transmitted light transmitted from a transfer pattern of the transfer mask, and a mask stage used in the simulation process Has the same shape as
The exposure step uses the information related to the Zernike polynomial in the virtual reference plane associated with the mask blank substrate, and corrects the wavefront of transmitted light transmitted from the transfer pattern of the transfer mask. A method for manufacturing a semiconductor device. Forming a transfer pattern on the transfer pattern forming thin film of the mask blank manufactured by the mask blank manufacturing method according to claim 15, and manufacturing a transfer mask;
An exposure step of chucking the produced transfer mask on a mask stage of an exposure apparatus, and transferring the transfer pattern of the transfer mask onto a semiconductor substrate by a lithography method,
The exposure apparatus has a function of performing wavefront correction of a shape defined by Zernike polynomials on a wavefront of transmitted light transmitted from a transfer pattern of the transfer mask, and a mask stage used in the simulation process Has the same shape as
The exposure step uses the information related to the Zernike polynomial in the virtual reference plane associated with the mask blank to correct the wavefront of the transmitted light transmitted from the transfer pattern of the transfer mask. A method for manufacturing a semiconductor device.

Images