JP2007328908A - Nanoimprint method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a nanoimprint method, by which a pattern of nanometer size is printed onto a doughnut-shaped substrate inexpensively at a time and further pattern errors are prevented. <P>SOLUTION: The method includes steps of depositing a processing layer on a circular substrate that has an hole in central portion; depositing a mask layer on the processing layer; and forming concave portions by pressing a master board against, having protrusions formed in a concentric-circular shape to the circular substrate, the mask layer, wherein the concave portions on the mask layer are formed so that distance d from the bottom of the concave portion formed in the mask layer to the processing layer surface may increase or decrease from the inner periphery of the circular substrate toward the outer periphery thereof, or so that the distance d may decrease from the outer periphery and middle of the circular substrate toward the center-periphery part; separating the master board from the mask layer; and performing etching to transfer the pattern including the concave portions in the mask layer to the processing layer. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a nanoimprint method, and in particular, is applied to the manufacture of a so-called patterned medium in which the magnetic recording density is improved by dividing and arranging a part of the medium.

  In recent years, thermal fluctuation has become a big problem for increasing the density of magnetic recording media. In currently used magnetic recording media, information is recorded on a magnetic thin film having magnetic particles separated magnetically as structural units. In order to perform high-density magnetic recording, it is necessary to reduce the magnetic domain, which is an information recording unit, but at the same time, it is necessary to reduce the magnetic particles in order to clarify the boundary line between the magnetic domain and the magnetic domain. Then, the magnetic anisotropy energy (represented by the product of the magnetic anisotropy energy density Ku and the volume V of the magnetic particles) necessary to keep the magnetization direction of the magnetic particles in one direction is about the thermal fluctuation energy at room temperature. As a result, the recorded information disappears because the magnetization fluctuates with time. This is called the thermal fluctuation problem.

  Patterned media have been proposed to solve this thermal fluctuation problem. The patterned medium is obtained by dividing a magnetic thin film into the size of a recording magnetic domain. By doing so, V can be increased, so that the thermal fluctuation problem can be avoided.

  However, the size of the recording magnetic domain is still several hundred nm × several tens of nm, and the biggest problem is that it is difficult to process the magnetic film to a size smaller than this. This level of processing is not impossible using the latest semiconductor processing technology. For example, there is a method in which a mask is formed directly on a magnetic film coated with a resist by drawing by electron beam (EB) exposure, and the magnetic film is processed based on the mask. However, when EB direct drawing is performed, nanopatterns are drawn one by one on the entire surface of the hard disk substrate, which takes enormous time. In addition, since the drawing apparatus is very expensive, it is impossible to manufacture a patterned medium in large quantities at a low cost.

  A nanoimprint method is known as a method for forming a nanopattern at low cost (Patent Documents 1 and 2). In this technique, a master on which a nano pattern is formed is pressed from above a processing layer coated with a resist, the nano pattern is mechanically transferred to the resist, and etching is performed based on the nano pattern. If this method is used, EB drawing is performed only when the master is produced, and only the master is pressed (imprint) at the time of transfer.Therefore, a single master is imprinted several times (several hundreds to tens of thousands). If possible, a nanopattern can be formed at low cost.

  However, the above technique is intended for forming a semiconductor pattern. In other words, since the pattern to be formed is a repetitive pattern of a minute size (for example, 1/100 of a wafer), the master on which the pattern to be imprinted is formed is smaller than a single wafer and imprinted on the wafer. Is repeated a plurality of times to form a pattern on the entire wafer. For this reason, it is not a technique that can be used as it is for a patterned medium that can be mass-produced only by imprinting on a donut-shaped glass substrate having a diameter of about 2.5 inches at a time.

  In addition, in order to imprint nanometer-sized patterns on a large area that is not assumed by the above technology, not only the pressing pressure uniformity and the flatness of the master disk are required, but also the behavior of the resist flowing out by pressing. It will be necessary to control as well. In the conventional semiconductor technology, a region not used as an element can be arbitrarily set on the wafer, so that a resist outflow portion can be provided outside the imprint portion using a small master. Further, in a semiconductor, it is sufficient not to use an imprint defective portion as a defective element. However, since the entire surface functions as a device in a hard disk application, a special device that does not cause an imprint defect is necessary.

  Furthermore, since the conventional technique assumes semiconductor processing, only the pattern based on a rectangle has been examined. Since the symmetry of this pattern is, for example, twice axisymmetric, the flow of pressed resist can be considered along that axis. Actually, since the pattern takes a complicated shape, the symmetry is lost, and it is impossible to control the resist flow and the like with the pattern. However, in the case of a magnetic recording medium, since it is circularly symmetric, the extension of the conventional technique cannot be used as it is.

As described above, nanoimprint technology is known as a method for producing patterned media at low cost, but all are intended for semiconductor applications, and the entire surface is imprinted at once using a donut-shaped substrate. In some cases, no consideration has been given to the application to hard disks with a large area up to about 2.5 inches in diameter.
US Pat. No. 5,772,905 US Pat. No. 5,956,216

  An object of the present invention is to provide a nanoimprint method capable of printing a nanometer-sized pattern on a donut-shaped substrate at a low cost at a low cost and further suppressing pattern errors.

  The nanoimprint method according to one aspect of the present invention includes a step of depositing a processing layer on a circular substrate having a hole in a central portion, a step of depositing a mask layer on the processing layer, and a concentric shape with the circular substrate. A step of pressing a formed master having a convex portion against the mask layer to form a concave portion, wherein a distance d from the bottom of the concave portion formed in the mask layer to the surface of the processed layer is within the circular substrate; Forming a recess in the mask layer so as to increase or decrease from the peripheral side to the outer peripheral side, or decrease from the outer peripheral side and the inner peripheral side to the middle peripheral portion of the circular substrate; and A step of separating from the layer, and a step of performing etching to transfer a pattern including a concave portion of the mask layer to the processed layer.

  The nanoimprint method according to another aspect of the present invention includes a step of depositing a processing layer on a circular substrate having a hole in the center, a step of depositing a mask layer on the processing layer, and a concentric shape with the circular substrate. A step of pressing a master having a convex portion against the mask layer to form a concave portion, wherein the height d ′ of the convex portion increases or decreases from the inner peripheral side toward the outer peripheral side, or the outer peripheral side and the inner peripheral side. A pattern including a step of forming a recess in the mask layer using a master that increases from the side toward the middle periphery, a step of separating the master from the mask layer, and a pattern including a recess in the mask layer And transferring the material to the processed layer.

The nanoimprint method according to still another aspect of the present invention includes a step of depositing a processing layer on a circular substrate having a hole in a central portion, a step of depositing a mask layer on the processing layer, and a concentric circle with the circular substrate. A step of pressing a master having a convex portion formed in a shape against the mask layer to form a recess, a step of separating the master from the mask layer, etching, and processing the pattern including the recess of the mask layer And imprint area of the master disk is defined as S = α / di / do with respect to the outer diameter do and the inner diameter di of the circular substrate, S / α <15000 mm ² , And it satisfies the requirement of 0.2 <α <0.4.

  The nanoimprint method according to still another aspect of the present invention includes a step of depositing a processing layer on a circular substrate having a hole in a central portion, a step of depositing a mask layer on the processing layer, and a concentric circle with the circular substrate. A step of pressing a master having a convex portion formed in a shape against the mask layer to form a recess, a step of separating the master from the mask layer, etching, and processing the pattern including the recess of the mask layer And γ = a / (a + b) with respect to the radial length a of the concave portion of the mask layer and the radial length b of the convex portion of the mask layer. And a and γ are substantially constant over the entire master and satisfy the requirement of a γ <300 nm.

  According to the nanoimprint method of the present invention, a nanometer-size pattern can be printed on a donut-shaped substrate at a low cost, and pattern transfer errors can be suppressed.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. The present invention is not limited to the following embodiment.

  FIG. 1 schematically shows a state when the mask layer is pressed by the master. A processed layer 12 is formed on the substrate 11, and a mask layer 13 is formed on the processed layer 12. On the other hand, a pattern forming a convex portion is formed on the master 14. d ′ is the height of the convex portion of the pattern formed on the master 14. d is the distance from the bottom of the recess formed in the mask layer 13 by pressing the master 14 to the surface of the processed layer.

  The substrate 11 to be imprinted is a circular substrate (disk-shaped substrate) and has a concentric hole at the center. What is generally used for a hard disk or an optical disk can be used. The material is hard and is preferably glass, Si wafer, hardened aluminum, ceramic composite material, hard plastic or the like. The processed layer 12 is a magnetic film used for a magnetic recording medium such as CoCrPt. A plurality of underlayers may be provided under the processed layer for the purpose of improving adhesion and controlling the crystallinity of the processed layer. Further, the processed layer may be a laminate of a plurality of layers.

The mask layer 13 becomes an etching barrier when the processed layer is processed by etching. The mask layer may be made of a material that does not etch everything during the etching time for etching the processed layer to a predetermined depth. For example, a material having an etching rate slower than that of the processing layer may be provided by the same thickness as the processing layer, or a material having an etching rate twice as fast as that of the processing layer may be provided by a thickness of 3 times. . A material generally used as a resist in a semiconductor manufacturing process is preferable as a mask layer because it is easy to control. Further, a metal such as Ta or W or a non-metal such as C, SiO ₂ or SiN can be used for the mask layer. Further, a stack of these materials may be used as a mask layer. For example, a PMMA resist is applied on W, an uneven pattern is formed on the resist by imprinting, the W layer is etched using this pattern as a mask, and the processed W layer is used as a mask for etching the processed layer. can do. At this time, the W / PMMA laminated film becomes a mask layer.

  The master 14 has a pattern with convex portions formed on the surface thereof, and is preferably the same as or larger than the substrate. The pattern formed on the master is concentric with the substrate and is circularly symmetric. Since the pattern is transferred to the resist by pressing, the material of the master is required to be harder than at least the mask layer or part of the mask layer.

  In the imprint method according to the present invention, the step of depositing the processed layer 12 on the substrate may be a step used for manufacturing a normal hard disk medium such as a sputtering method. Also, a dry method such as a vapor deposition method, a laser ablation method, or an MBE (molecular beam epitaxy) method that is generally known as a thin film manufacturing method, or a wet method such as spin coating, coating, or plating can be used. There is no particular limitation as long as it is a method for forming a thin film having a thickness of about 1000 nm or less.

  The step of depositing the mask layer 13 on the processed layer 12 can use, for example, spin coating.

  The step of pressing the master 14 against the mask layer 13 is a step of pressing the master uniformly against the mask layer / processed layer. As a device for pressing the master, for example, a hydraulic press machine may be used. Since it is essential to press uniformly, the pressing means is not particularly limited.

  The step of separating the master 14 from the mask layer 13 is a step of peeling the master again after pressing the master against the mask layer / processed layer.

  The etching process is a process of transferring the pattern of the mask layer to the processed layer, and an etching process generally used in a semiconductor manufacturing process can be used. For example, a dry method such as ion milling or RIE (reactive ion etching), or a wet method in which the material is immersed in a liquid (for example, an acid or an organic solvent) that is soluble in the processed layer material can be used.

  The method according to the present invention is sufficient as long as it includes at least the above-described steps, and a step for improving the efficiency and accuracy of processing may be added in addition to the above.

In the conventional nanoimprint method, it has been generally considered that the distance d from the bottom of the recess formed in the mask layer to the surface of the processed layer is preferably uniform over the entire surface of the medium. This is because the progress of the etching process is taken into consideration. FIGS. 2A to 2C schematically show an etching process after nanoimprinting. FIG. 2A corresponds to the state of FIG. 1 and shows a state in which a concave portion is formed in the mask layer 13 after the master 14 is pressed against the mask layer 13 and pulled away. FIG. 2B shows a state after “bottoming out” the bottom of the mask layer 13 by, for example, O ₂ -RIE. FIG. 2C shows a state where the pattern of the mask layer 13 is transferred to the processed layer 12 by etching.

  When the distance d from the bottom of the recess formed in the mask layer to the surface of the processed layer is not uniform, a portion where the processed layer 12 is exposed and an unexposed portion are formed at the stage of FIG. There arises a problem that it cannot be processed. To solve this problem, it can be considered that the processed layer is surely etched by extending the etching time and over-etching. However, variations occur in the processing depth in the substrate, and the etching proceeds in the lateral direction, resulting in variations in pattern shape. For this reason, as described above, it has been a conventional common sense that d is preferably uniform over the entire surface of the medium.

  The inventors made a prototype of a patterned medium by a method of keeping d constant according to the above general common sense. However, when d was kept constant, the problem was that the uniformity of the shape of the processed layer after processing did not improve. As a result of the examination, it was found that one of the causes was that the pattern was not transferred well and the pattern was missing in some places.

  Therefore, the present inventors have tried a method of imprinting so that d is not uniform, contrary to common sense. Specifically, imprinting was performed by adjusting the pressure for pressing the master so as to increase at the inner, outer, or intermediate (concentric) shape of the disk substrate. The pattern includes a mixture of a pseudo servo signal area and a track, and a pattern in which dots having the same size as the pseudo servo signal area continue. The in-plane distance between patterns (distance from one convex part to the nearest convex part) was less than 300 nm at the maximum.

  When d after imprinting was measured, there was a tendency as shown in FIG. 31 indicates a case where the outer peripheral side is strongly pressed, and d decreases from the inner peripheral side to the outer peripheral side of the substrate. 32 is a case where the middle peripheral portion is strongly pressed, and d decreases from the outer peripheral side and the inner peripheral side of the substrate toward the middle peripheral portion. Reference numeral 33 denotes a case where the inner peripheral side is strongly pressed, and d decreases from the inner peripheral side to the outer peripheral side of the substrate.

  As shown in FIG. 3, when the distribution was given to d, the pattern destruction when peeling the master after imprinting was clearly reduced. The cause is not well understood, but it is presumed that the frictional force between the mask layer and the master becomes non-uniform due to the distribution of d, resulting in a decrease in adhesion.

  Next, the processed layer was etched. As described above, the etching conditions at this time were over-etching such that the substrate was slightly etched at the site where d was the largest. As described above, the pattern shape is disturbed by over-etching. However, since the chipping of the pattern due to the transfer error was reduced, good etching was possible as a result. The rate of pattern chipping was measured at the servo signal portion, track portion, and continuous dot portion. Observed with a scanning electron microscope (SEM), the missing pattern was extracted by image processing. As a result, when d was made uniform, the pattern loss rate was about 12%, but in all cases 31, 32, and 33 of FIG. 3, it decreased to about 5%.

  The present inventors further examined to what extent the recording magnetic domain scale can be applied with the above-mentioned effect by providing the d distribution. First, a similar experiment was performed by fabricating a pattern having an in-plane distance between patterns of 3 μm. As a result, there was almost no transfer error, and the rate of missing patterns after processing the magnetic film was almost 1% of the detection limit. The cause of this is not well understood, but when the pattern is small, there are relatively many barriers (protrusions on the master) that hinder the movement of the mask layer material that is pushed in, so it is impossible to move the pattern enough to transfer, If the in-plane distance between the patterns is large, the number of barriers is relatively small, and therefore it is estimated that the mask layer material can move freely so that the convex portions of the master can be transferred satisfactorily. From this result, it was found that the effect of making d non-uniform appears significantly in a small pattern. Therefore, in order to estimate the pattern size that can obtain the effect of making d non-uniform, masters having patterns with in-plane distances of 100, 200, 300, 400, 600, 800, and 1000 nm on the same surface were fabricated. Each pattern has a concentric shape, and the shape is rectangular in the master disk radial direction and circumferential direction at equal intervals. As a result, it was found that the pattern defect rate becomes remarkable when the in-plane distance is 300 nm or less.

  The above effect can also be obtained by changing the height of the convex portion of the original. Specifically, the height d ′ of the pattern protrusion on the master is changed with a distribution similar to the characteristic curve in FIG. 3 (however, the relationship between the height and the height is reversed), and the pressure during imprinting is changed over the entire surface of the substrate. This is a uniform method. By doing so, there is an advantage that the apparatus configuration is simplified and the reproducibility is improved since it is sufficient to apply pressure uniformly during imprinting. On the other hand, since the applied pressure (stress) varies depending on the radius for the master, there is also a drawback that a portion where stress is strong deteriorates quickly.

  The essence of the present invention is that there is a distribution in d or d ', and the structure facilitates the flow of the mask layer material. Therefore, even if there is a local variation in d or d ', it is sufficient if there is a tendency as shown in FIG. 3 as a whole. For example, the distribution obtained by taking a moving average of d or d 'with a length of 1/5 of the disk radius may be as shown in FIG.

  Further, the present inventors have found that there is a preferable relationship among the inner diameter di, the outer diameter do, and the imprint area S of the substrate to be imprinted. As described above, a nanoimprint developed on the assumption of a realistic semiconductor manufacturing process differs greatly in that a nanoimprint intended for a disk substrate has a circular hole in the center. When the mask layer material is pressed by the master, the mask layer material moves about the size of the pattern to a portion having no pattern. At this time, stress acts on the mask layer material. This stress is considered to act as a hydrostatic pressure, and when there is no hole in the center of the substrate, the mask layer material cannot move in the center, so the stress acts as a force that greatly impedes the movement of the mask layer material toward the inner periphery. To do. On the other hand, when there is a hole in the central portion of the substrate, the mask layer material can move from the peripheral portion of the inner diameter toward the center, and thus the inhibition force is reduced. For this reason, the presence or absence of a hole in the central part is considered to greatly affect the distortion of the mask layer during imprinting, the adhesion / pressing pressure to the pattern convex part of the master and the flow distance. However, there is no report on the entire surface imprint on the substrate having a hole in the center, and the above-mentioned influence is not known.

  Therefore, the present inventors defined a ratio α = di / do between the inner diameter and the outer diameter as a parameter directly corresponding to this stress, and investigated the change in imprint performance due to the difference in α. As a result, there was a tendency that the larger α was, the smaller the transfer error was, and the results supporting the above-mentioned idea that the stress in the inner circumferential direction was relieved. At this time, the force for imprinting the master was assumed to be a 30 to 50 ton hydraulic press, assuming a mass production device for storage devices such as hard disks or optical discs in a clean room.

  However, when the same experiment was performed with the substrate size changed to 1.0, 1.5, 2.0, 2.5, 3.0, 3.5 inches, which are commonly used, the transfer error was not determined only by the value of α, but the imprinted area. S was found to have an important effect. There was a tendency that transfer errors increased as S increased. The reason for this is not well understood, but the larger the area, the more the flow barrier is relatively increased. Therefore, it is assumed that the stress is not uniform enough to cause an error.

We found that transfer errors are less likely to occur when α is larger and S is smaller. Therefore, we defined S / α as a parameter indicating the small number of transfer errors and summarized the data. Regardless, it was found that the error was reduced when S / α <15000 (mm ² ) was satisfied. In practice, if α is extremely small, an error occurs even if S is reduced. If α is too large, the error is small but the area efficiency is too bad for use in a storage medium. Therefore, it is preferable that substantially 0.2 <α <0.4.

  The magnitude of the hydrostatic pressure and the flow of the mask layer material can be controlled by increasing the imprinting force. For example, if imprinting is performed with a press machine such as a 100-ton press or a 300-ton press, it is theoretically considered that an imprint with a small transfer error or a large S can be made with a small transfer error. However, such a press is not preferable because the apparatus becomes too large to be installed in a clean room, and the manufacturing cost increases. Moreover, there is a possibility that the master and the substrate are destroyed by being pressed with such a large force. Further, in order to withstand a large hydrostatic pressure, it is necessary to make the pattern of the master disk stronger, which is also not preferable because it increases the cost.

  Furthermore, in the process of developing the imprint technique for a more practical disk medium, the present inventors have formed a radial length a of the recess formed in the mask layer and the mask layer. The inventors have found a condition that the transfer error is reduced when the length b in the radial direction of the convex portion is substantially constant in the pattern.

  Here, substantially substantially constant means that the pattern obtained by imprinting is constant to the extent that it can be used as a disk medium currently used. Currently, a hard disk or an optical disk, which is a commercially available disk medium, has a concentric track with a predetermined width, and information is recorded / erased / reproduced on the track. In order to perform efficient information processing, the track width is constant over the entire surface of the medium. The present inventors also created an imprint pattern in accordance with this form, and found the above-described preferable conditions. Development of such a viewpoint has not been made by an imprint technique assuming a conventional semiconductor manufacturing process. There are papers and patents that indicate that nanoimprint technology can also be applied to hard disks, but they only suggest that they can be applied in principle, and studies on transfer errors when forming track patterns are not possible. Not done.

  We have a track-based pattern of uniform width (e.g. an optical disk or a discrete track hard disk medium such as a hard disk medium or a patterned medium in which similar dots are arranged along the track). ) Was made and imprinting was repeated, but under certain conditions, it was discovered that transcription errors frequently occur.

  In the case of imprinting a pattern based on concentric grooves, it is considered that the movement of the mask layer material and the generated distortion are dominant in the disk radial direction. Even in the case of a dot pattern such as a patterned medium, the movement of the mask layer material that occurs between patterns partially has a component in the circumferential direction of the medium, but the amount of movement in that direction is hindered by the adjacent pattern. Small to make. In the case of a complete groove pattern, the mask layer material is overwhelmingly moved in the disk radial direction. As a result of investigations, it was found that pattern transfer errors strongly depend on the shape of the pattern. That is, the amount of the mask layer material that moves by being pushed in the disk radial direction has a positive correlation with a. On the other hand, the strain generated with the movement of the mask layer material has a positive correlation with γ when considered in the disk radial direction. Therefore, it was considered that by reducing a and γ, distortion in the mask layer during imprinting can be reduced, and by reducing the pressure on the master, pattern transfer errors can be reduced.

  Therefore, different patterns of a and b were prepared and an imprint transfer experiment was conducted. Thereafter, the transfer error occurrence rate was examined by the same method as described above. To summarize the results, it was found that transfer errors can be reduced if the condition of aγ <300 nm is satisfied. At this time, the force for imprinting the master was assumed to be a 30 to 50 ton hydraulic press machine assuming a mass production device of a storage device such as a hard disk or an optical disk in a clean room. The reason is the same as described above.

  Furthermore, the present inventors can vary the depth of the mask layer after transfer by filling the concave portion of the mask layer with a material having an etching rate lower than that of the processing layer instead of performing over-etching when etching the processing layer. Attempted to correct the shortcoming that occurred. Here, the filling material is not particularly limited as long as it has an etching rate lower than that of the processed layer and can be filled to a nanometer size. For example, hard ceramic materials such as Si-O, Ti-O, Ta-O, Zr-O, C, Si, etc. may be deposited by sputtering, vapor deposition, or CVD, or in liquid form (eg with etching resistance) Alternatively, a resist material made of a substance different from the mask layer may be spin-coated.

Liquid materials have the advantage that they can be filled with a simple device configuration. In particular, a liquid material known as SOG (Spin-On-Glass) is preferable. This material contains SiO ₂ and is generally used as a material that changes to SiO ₂ when baked at a temperature of several 100 ° C. or higher. The concave portion of the mask layer can be replaced by a baking process after filling to form a new mask, or baking is not necessary if the etching resistance is higher than that of the processed layer. Baking treatment increases the cost, but there is an advantage that processing can be done firmly.

It is also preferable to use tetraethoxysilane (TEOS) as a similar liquid filling material. This material can be filled into a pattern by spin coating or liquid phase CVD. This material is also generally used as a material that changes to SiO ₂ when baked at several hundred degrees Celsius or higher. The concave portion of the mask layer can be replaced by a baking process after filling to form a new mask, or baking is not necessary if the etching resistance is higher than that of the processed layer. Baking treatment increases the cost, but there is an advantage that processing can be done firmly.

(Example)
Examples of the present invention will be described below. The present invention is not limited to the following examples.

(Example 1)
[Preparation of master]
A concentric pattern was drawn on a resist spin-coated on a Si wafer using an R-θ type electron beam drawing apparatus. The pattern imitates a signal currently used for a hard disk and includes a preamble, an address, a burst signal, and a data part. These signals are dummy signals and are not configured to operate as a hard disk immediately, but are sufficient as a test when the nanoimprint is applied to a hard disk. The track width of the pattern was 50, 100, 200, 300, 400, 600, 800 nm. The recording signal of the dummy data is a fine signal “101010. The signal interval is 1/3 to 1/10 of the track width.

  The pattern thus produced was subjected to the same post treatment as that for producing an optical disc master, and a Ni master was produced by Ni electroforming. The Ni master has an outer diameter of +20 mm so that the entire surface of the 1.0, 2.5, and 3.5 inch disk substrate (doughnut shape) can be pushed, and there is no hole corresponding to the hole in the center of the disk substrate. The height d ′ of the convex portion of the pattern is 100 nm.

[Manufacture of magnetic recording media]
A 1.0, 2.5, and 3.5 inch disk substrate (doughnut-shaped) was prepared, and a Ti underlayer (50 nm), a CoCrPt magnetic layer (15 nm), and a C protective layer (5 nm) were sequentially deposited thereon by a sputtering method. On this film, a resist was applied as a mask layer by spin coating. The set coating thickness was 100 nm. From above, the Ni master was pressed with a 30-ton hydraulic press for imprinting (FIG. 2 (a)). After imprinting, the bottom of the resist was “bottomed out” by O ₂ -RIE (FIG. 2B), and then the magnetic layer portion was etched by Ar ion milling (FIG. 2C).

  The etching conditions at this time were over-etching such that the substrate was slightly etched at the site where d was the largest. As described above, although the pattern shape is disturbed by overetching, pattern defects due to transfer errors are reduced, and as a result, good etching can be performed.

[Evaluation of pattern irregularities]
The shape after etching was measured by SEM (scanning electron microscope). The measurement sample was divided into five radii of each disk substrate, and then divided into 18 parts each 20 degrees on the circumference at each radial position, and cut into 5 mm squares at these dividing points. The 90 samples cut out in this way were observed at a magnification such that at least 100 patterns were included in the field of view, fitted by image processing, and evaluated for the rate at which the shape was disturbed. The result was obtained by averaging the ratio of the shape disturbance of one sample with 90 pieces.

  (Sample A) When imprinting, place a doughnut-shaped Teflon (registered trademark) buffer sheet (0.3 mm thick) on the outer periphery of the master so that only the outer periphery of the master can be pressed relatively strongly. Imprint was performed. The resist shape after imprinting thus prepared was examined by AFM. Sampling places are 90 places as described above. As a result, the distance d from the bottom of the recess shown in FIG. 1 to the surface of the processed layer (evaluated from the coating thickness and the depth of the recess) is 10 nm on the inner peripheral side and 3 nm on the outer peripheral side, and changes with respect to the position on the substrate Was schematically shown at 31 in FIG.

  (Sample B) A buffer sheet was placed on the inner periphery of the master, and imprinting was performed so that only the inner periphery of the master could be pressed relatively strongly. As a result of the above observation, the value of d was 3 nm on the inner peripheral side and 11 nm on the outer peripheral side, and the change with respect to the position of the substrate was schematically shown in 33 of FIG.

  (Sample C) The buffer sheet was placed in the middle part of the master (concentric), and imprinting was performed so that only the middle part of the master could be pressed relatively strongly. As a result of the above observation, the value of d was 9 nm on the inner circumference side, 8 nm on the outer circumference side, and 3 nm on the middle circumference portion, and the change with respect to the position of the substrate was schematically shown as 32 in FIG.

  (Sample D) Install the buffer sheets on the outer and inner circumferences of the master (the width is narrower than the above), so that the outer and inner circumferences of the master can be pressed relatively strongly. Printed out. As a result of the above observation, the value of d was 3 nm on the inner circumference side, 4 nm on the outer circumference side, and 10 nm on the middle circumference portion, and the change with respect to the position of the substrate was as if 32 in FIG.

  (Sample E) As Comparative Example 1, imprinting was performed by pressing the buffer sheet evenly without placing the buffer sheet on the master. As a result of the above observation, the value of d was constant from the inner circumference side to the outer circumference side and was 6 nm.

  The above five samples were evaluated for pattern shape disturbance after etching. As a result, the pattern error rate was about 5% for samples A, B, and C, 30% for sample D, and 12% for sample E (Comparative Example 1).

  In the region where the in-plane distance between patterns is larger than 300 nm, that is, in the portion where the track width is 400 nm or more, the pattern error rate did not change at 10% in any sample. For comparison, when a similar experiment was performed using a rectangular dot pattern master with an in-plane distance between patterns of 3 μm, there was almost no pattern error and the pattern after processing the magnetic film was missing. The rate was almost 1% of the detection limit. From this, it was found that the present invention is effective when the pattern in-plane distance is 300 nm or less.

  During the above experiment, several samples having imprint patterns having a distribution different from the d distribution shown in FIG. When these samples were evaluated in the same manner as described above, a sample with an in-plane distance of 300 nm or less and containing a portion that can be regarded as d = 0 within the AFM resolution range is a pattern error. The rate was found to be less than 7%.

(Example 2)
In response to the result of Example 1, as a technique for creating a similar imprinted shape, an attempt was made to provide a distribution in the protrusion height of the master. Ar ion etching treatment was performed on a Ni master produced by the same method as in Example 1. The etching power was set to be sufficiently weak so that etching of a pattern (projection height) of about 50 nm occurred in 5 minutes. At this time, a shield plate was installed on the front surface of the Ni master to adjust the height of the protrusion after etching. As a result, the master with the height of the convex portion increased from the inner periphery to the outer periphery (master α), the master with the height of the convex increased from the outer periphery to the inner periphery (master β), the height of the convex Three types of master discs, ie, master discs (master disc γ) whose length increases from the outer peripheral side and the inner peripheral side toward the middle peripheral portion, could be produced.

  Using these masters, an imprint / magnetic layer processing experiment was performed in the same manner as described above. As a result, the pattern error rate is: master α: 3%, master β: 2%, master γ: 4%, compared with 12% in the case of normal imprint using a normal master (Comparative Example 1 above) Also a small value was obtained. This is because, as a method of giving the distribution d as shown in FIG. 3 to the distance d from the bottom of the concave portion to the surface of the processed layer, the height d ′ of the pattern convex portion of the master is distributed and imprinted uniformly. It shows that the method is also effective.

  As in the case of Example 1, several samples having no distribution as shown in FIG. 3 were produced during the ion etching process of the master. Also, when imprinting is performed under various conditions using this master and the above masters α, β, and γ, a portion where d after imprinting becomes 0 within the AFM resolution range may occur. A sample was made. Based on the results of Example 1, the pattern error rate was also evaluated for these samples. As in Example 1, the pattern error rate was 8% or less in the sample in which a portion where d was 0 was generated. It turned out to be smaller.

(Example 3)
About the sample after the imprint produced in Examples 1 and 2, an attempt was made to fill the concave portion with a material having an etching rate lower than that of the resist portion. Specifically, the coating resist was PMMA (polymethylmethacrylate), and after imprinting, PS (polystyrene) resin was spin coated to fill the recesses. Further, when 10 nm of Au was vapor-deposited after imprinting, Au was deposited relatively thickly in the recesses. Using these samples, the magnetic layer was processed and the pattern error rate was evaluated. As a result, it was found that the error rate was improved by 30-50% relative to the cases of Examples 1 and 2 (for example, the error rate of 5% was 3%). This is probably because the occurrence of errors due to etching during processing of the magnetic layer is suppressed.

  In addition, SOG (Spin-On-Glass) was tried as a filling material. As the SOG, a commercially available product generally used in the semiconductor manufacturing process was used. As a result of optimizing the coating conditions, it was found that there are conditions that allow SOG to be selectively filled in the recesses. Using these samples, the magnetic layer was processed and the pattern error rate was evaluated. As a result, similarly, it was found that the error rate was improved by 30-50% relative to the cases of Examples 1 and 2 (for example, the error rate of 5% was 3%).

  In addition, TEOS (tetraethoxysilane) was tried as a filling material. Similarly, TEOS used was a commercial product that is generally used in the semiconductor manufacturing process. TEOS can be deposited by CVD, but for the sake of simplicity, we tried spin-coating a solution diluted with a solvent. As a result of optimizing the coating conditions, it was found that there is a condition that the TEOS can be selectively filled in the recesses. Using these samples, the magnetic layer was processed and the pattern error rate was evaluated. As a result, similarly, it was found that the error rate was improved by 30-50% relative to the cases of Examples 1 and 2 (for example, the error rate of 5% was 3%).

  In addition, C was deposited as a filling material by CVD. Using methane gas as a raw material, it was deposited on the pattern by plasma CVD. C was mostly deposited in the recesses, and when deposited at a set film thickness of 40 nm, the surface roughness decreased compared to before deposition. In this state, the PMMA resist was removed by oxygen ashing. The concave portion where CVD-C is deposited is not etched, but the convex portion is etched from the side wall. As a result, an opposite (negative) pattern to the mask layer was formed on the mask layer. Using these samples, the magnetic layer was processed and the pattern error rate was evaluated. As a result, similarly, it was found that the error rate was improved by 30-50% relative to the cases of Examples 1 and 2 (for example, the error rate of 5% was 3%).

Example 4
The same experiment as in Example 1 was performed. However, d was constant. Α = di / do is defined for the inner diameter di and outer diameter do of the board, and 1.5, 2.0, and 3.0 inch outer diameter boards are processed by reducing the outer diameter based on 2.0, 2.5, and 3.5 inch disk substrates. Was made. In addition, a substrate having an enlarged inner diameter was also produced, and substrates having various values of α and area S were produced.

The same imprint-magnetic layer processing experiment as that of Example 1 was performed on these substrates, and pattern error rate evaluation was performed. The result is shown in FIG. Α is plotted on the horizontal axis and S is plotted on the vertical axis, and the pattern error rate of Comparative Example 1 above is adopted as a reference value. . The straight line 41 is a straight line showing a relationship of S = α × 15000 (mm ² ). As is apparent from the figure, the pattern error rate decreases when S <α × 15000 (mm ² ).

  If α <0.2, the effect of reducing the pattern error rate cannot be obtained. Further, when α> 0.4, the experiment was not performed because the disk substrate was not substantially meaningful as a storage medium such as a hard disk.

  In the same manner as in Example 3, an attempt was made to fill a material having a low etching rate after pattern formation of the mask layer. In the same manner as in Example 3, the PMMA mask layer = PS filling resin, Au deposition, SOG, TEOS, and CVD-C were tested. In either case, it was found that the pattern error rate was improved by 30-50% when S / α <15000 (for example, 5% error rate was 3%). However, in the region where S / α> 15000, the pattern error rate never fell below 12%.

(Example 5)
The same experiment as in Example 1 was performed. However, d was constant. Although the formed patterns have the same shape, the length a of the mask layer recesses in the radial direction and the length b of the mask layer protrusions are substantially constant on the same substrate. Using a 1.0, 2.5, and 3.5 inch disk substrate, the same imprint-magnetic layer processing experiment as in Example 1 was performed, and the pattern error rate was evaluated.

  The result is shown in FIG. The horizontal axis is a, and the vertical axis is γ. Pattern error rates of 12% or more are indicated by black circles, and those with less than 12% are indicated by white circles. A straight line 51 is a curve showing a relationship of aγ = 300 nm. As is apparent from the figure, the pattern error rate decreases when aγ <300 nm.

  In the same manner as in Example 3, an attempt was made to fill a material having a low etching rate after pattern formation of the mask layer. In the same manner as in Example 3, the PMMA mask layer = PS filling resin, Au deposition, SOG, and TEOS were tested. In either case, it was found that the pattern error rate was improved by 30-50% when aγ <300 nm (for example, 5% error rate was 3%). However, the pattern error rate never fell below 12% in the region where aγ> 300 nm.

The figure which shows typically the state which implemented the nanoimprint method which concerns on embodiment of this invention. The figure which shows typically the etching process after nanoimprint. The figure which shows the relationship between the position on a board | substrate, and the depth d of the recessed part formed in the mask layer. The figure which shows the change of the transcription | transfer error with respect to (alpha) and S. FIG. The figure which shows the change of the transcription | transfer error with respect to a and (gamma).

Explanation of symbols

  DESCRIPTION OF SYMBOLS 11 ... Board | substrate, 12 ... Processed layer, 13 ... Mask layer, 14 ... Master disk.

Claims (5)

Depositing a working layer on a circular substrate having a hole in the center;
Depositing a mask layer on the working layer;
A step of pressing a master having a convex portion concentrically formed with the circular substrate against the mask layer to form a concave portion, the distance d from the bottom of the concave portion formed in the mask layer to the surface of the processed layer Forming a recess in the mask layer so that increases or decreases from the inner peripheral side to the outer peripheral side of the circular substrate, or decreases from the outer peripheral side and the inner peripheral side to the middle peripheral portion of the circular substrate. When,
Separating the master from the mask layer;
Etching, and transferring the pattern including the concave portion of the mask layer to the processed layer. Depositing a working layer on a circular substrate having a hole in the center;
Depositing a mask layer on the working layer;
A step of pressing a master having a convex portion concentric with the circular substrate against the mask layer to form a concave portion, wherein the height d ′ of the convex portion increases or decreases from the inner peripheral side toward the outer peripheral side, Or forming a recess in the mask layer using a master that increases from the outer peripheral side and the inner peripheral side toward the middle peripheral part;
Separating the master from the mask layer;
Etching, and transferring the pattern including the concave portion of the mask layer to the processed layer. Depositing a working layer on a circular substrate having a hole in the center;
Depositing a mask layer on the working layer;
A step of pressing a master having a convex portion concentrically formed with the circular substrate against the mask layer to form a concave portion;
Separating the master from the mask layer;
Etching, and transferring the pattern including the concave portion of the mask layer to the processed layer,
When the imprint area of the master is defined as S = α = di / do with respect to the outer diameter do and the inner diameter di of the circular substrate, S / α <15000 mm ² and 0.2 <α <0. 4. A nanoimprinting method characterized by satisfying the requirement of 4. Depositing a working layer on a circular substrate having a hole in the center;
Depositing a mask layer on the working layer;
A step of pressing a master having a convex portion concentrically formed with the circular substrate against the mask layer to form a concave portion;
Separating the master from the mask layer;
Etching, and transferring the pattern including the concave portion of the mask layer to the processed layer,
When γ = a / (a + b) is defined with respect to the radial length a of the concave portion of the mask layer and the radial length b of the convex portion of the mask layer, a and γ are substantially over the entire master. And a nanoimprint method characterized by satisfying the requirement of aγ <300 nm.   5. The nanoimprint method according to claim 1, wherein a material having an etching rate slower than that of the processed layer is filled in the concave portion of the mask layer before the etching step.

Images