JP2007163613A - Inspection method of reticle - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To accurately inspecting fluctuations in the pattern dimension of a reticle mask pattern in a short time to identify the fluctuation section. <P>SOLUTION: The surface of reticle mask pattern 1 is scanned with a laser beam 31; at least either the diffracted light satisfying diffraction conditions or the diffracted light not satisfying the conditions in the laser light 31 diffracted on a pattern edge on the surface of the mask pattern 1 is detected; and the scanning position of the laser light 31 is verified with respect to the satisfying or unsatisfying conditions of the diffracted light so as to detect a section where the pattern dimension is locally fluctuated. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a method for inspecting a mask used in the manufacture of an LSI or the like, and more particularly, to a method for detecting a local portion where a pattern size of a reticle used in a lithography process is fluctuating.

  Currently, in the LSI manufacturing process, the pattern dimension accuracy in critical layers is required to be within about ± 10% of the pattern dimension.

  The pattern dimensional accuracy is determined by the lithography process and the etching process, but in particular, the dimensional accuracy of the reticle is greatly affected by the lithography process.

  By the way, as for the dimensional accuracy of the reticle, since the magnification of the reticle is generally at most 4 or 5 times the pattern on the wafer, the dimensional accuracy on the reticle is required to be the same as that on the wafer. For example, for a 0.5 μm pattern on the wafer, dimensional control of ± 50 nm is required on the wafer and on the reticle.

  In addition, the dimension in the reticle plane needs to be uniformly controlled, and needs to be locally within this controlled accuracy. Of course, the same applies to the wafer after the wafer transfer.

  Conventionally, in order to measure a local dimensional error in the reticle plane, a portion having a large dimensional error is identified by comparing the reticle pattern with the design data of the reticle pattern or comparing it with other patterns on the same chip. It was. According to this method, in order to capture a minute dimensional error, it is necessary to shorten the wavelength of a light source to be inspected or to make a pixel for pattern processing fine.

  Such shortening of the inspection wavelength and miniaturization of the pixels of the inspection image greatly increase the cost of the inspection apparatus, and cannot be an inspection method commensurate with the cost. For example, in order to detect an error from 50 nm required for 0.25 μm technology, an inspection device used for 90 nm technology is required.

  On the other hand, if a dimension measurement with high resolution is performed directly, accuracy can be expected, but a long processing time is required. Therefore, it is impractical to specify a part having a large dimension error locally on the reticle (for example, Non-Patent Document 1).

As another inspection method, there has been proposed an inspection method in which a portion with poor dimensional accuracy but with poor dimensional accuracy is detected as color unevenness by an optical system, and a region with a large dimensional error is specified. In this method, only rough position information can be obtained, and a local detailed part cannot be specified. In addition, the light amount fluctuation region of the defective part needs to be larger than the pixel size of the light receiving element, or high-order diffracted light with a small amount of light is used. It is unsuitable (for example, refer to Patent Document 1).
Byung Gook Kim, et al. Required Performances of Reticle Inspection System for ArF Lithography: Through Analysis of Defect Printability Study, Proceedings of SPIE Vol.4754 p517-525 (2002) Japanese Patent Laying-Open No. 2005-233869 ([0015], [FIG. 1])

  Thus, with the conventional reticle inspection technique, it is impossible to accurately measure and identify a region where the dimensional accuracy is locally degraded within a practical measurement time.

  Accordingly, the present invention provides a method for identifying local variations in pattern dimensions by scanning a mask pattern surface of a reticle with laser light and detecting diffracted light that is diffracted by the pattern edge.

  The above-described problem is that, in claim 1, the step of scanning the mask pattern surface of the reticle with laser light, the step of detecting diffracted light diffracted at the pattern edge of the mask pattern surface, and the diffraction light satisfy one diffraction condition. This is solved by a method for inspecting a reticle configured to include a step of checking whether or not the pattern is satisfied, and a step of detecting a variation portion of the pattern dimension of the mask pattern based on the comparison.

  That is, the laser beam scanned on the mask pattern is diffracted by the mask pattern. The diffraction angle of the diffracted light: θ is represented by an angle that satisfies the diffraction conditional expression: p × sin θ = nλ / 2 (p: pattern pitch, λ: wavelength of laser light, n: natural number).

  If the pattern pitch: p is determined when light of a certain wavelength is irradiated, the diffraction angle of the diffracted light is determined according to the diffraction conditional expression. Therefore, if inspection is performed while comparing the part to be scanned with the laser beam and the change in the diffraction angle of the diffracted light, that is, the change in the pattern pitch: p, the boundary part of the pattern in which the pattern dimension changes can be specified.

  Next, in claim 2, the mask pattern is solved by the reticle inspection method according to claim 1, wherein the mask pattern is formed at a predetermined pitch.

  That is, in the present invention, since the fluctuation of the diffraction angle of the light diffracted by the scanning light depending on the pattern pitch: p of the mask pattern is detected, the pattern pitch: p of the measurement site of the mask is accurately captured in advance. It is essential.

  Next, in claim 3, the laser beam is solved by the reticle inspection method according to claim 1, wherein the laser beam is narrowed down to at least the pitch dimension of the mask pattern.

  That is, since the laser beam that scans the mask pattern is diffracted at the boundary portion of the pattern according to the diffraction condition, it is difficult to detect an accurate change in the diffraction angle with a beam width that exceeds at least two patterns. Therefore, it is preferable that the laser beam to be irradiated is narrowed down to the mask pattern pitch: p.

  Next, in claim 4, the step of collating the diffracted light diffracted at the pattern edge of the mask pattern surface is solved by the reticle inspection method according to claim 1 configured to be performed through a filter.

  That is, the diffracted light is diffracted from a 0th-order diffracted light to a first-order, second-order, and higher-order diffracted light at a plurality of diffraction angles determined by a natural number of the diffraction conditional expression; In the case of the present invention, for example, if only the first-order diffracted light is detected through a filter that transmits only the first-order diffracted light, the light intensity of the diffracted light is large, so that it can be detected more accurately and quickly.

  According to the mask inspection method of the present invention, it is possible to inspect and specify a portion where pattern dimension variation exists in a practical short time while scanning the mask pattern of the reticle with laser light.

  As a result, it is possible to eliminate the conventional problem that it is difficult to specify the local part of the dimensional variation on the mask or that it takes a long time for the inspection, and the present invention is particularly suitable for the mask pattern inspection of the reticle. Can contribute a lot to efficiency.

  FIG. 1 is a diagram showing the positional relationship between the pattern dimension variation portion and the scanning laser beam, FIG. 2 is a schematic diagram of diffracted light at normal / abnormal pattern dimensions, and FIG. 3 is a schematic diagram of local dimension variation and boundary portions. 4 is a diagram showing the pattern dimension variation at the boundary, FIG. 5 is a diagram in which the pattern dimension variation is converted into the pitch variation, FIG. 6 is a relationship diagram between the pattern pitch and the diffraction angle, and FIG. FIG. 8 is an example of a working curve of pattern dimension variation and diffraction angle, and FIG. 9 is a configuration example of an optical system for carrying out the present invention.

  FIG. 1 schematically shows a positional relationship diagram between a pattern dimension variation portion and a scanning laser beam. The laser beam 31 is sequentially scanned in the direction of the arrow so as to straddle adjacent patterns 1 formed on the reticle 10. The case where the pattern 1 of the scanned mask has a normal pitch: p and the case where an abnormal pitch: p ′ exists are schematically shown.

  When the pattern 1 has a normal pitch: p, diffracted light appears in the direction of diffraction angle: θ (°) determined by the diffraction conditional expression: p × sin θ = nλ / 2. However, when the pattern 1 has an abnormal pitch: p ′, diffracted light does not appear in the direction of the diffraction angle: θ without following the diffraction conditional expression. Whether this diffracted light appears or not corresponds to the normal part / abnormal part of the pitch of the pattern 1. Therefore, if the position of these parts is locally detected, a part where pattern 1 is abnormal can be specified.

  FIG. 2 is a schematic diagram of the diffracted light of the normal part / abnormal part of the pattern dimension. FIG. 5A shows that the diffracted light 4 is diffracted at a diffraction angle: θ when the pitch of the pattern 1 is a normal pitch: p.

  On the other hand, as shown in FIG. 5B, the pitch of the pattern 1 is an abnormal pitch at the boundary part 2 between the normal part / abnormal part: p ′ (= p + ΔCD / 2, ΔCD is a variation amount of the pattern dimension). In this case, the diffracted light 4 is not diffracted at the diffraction angle: θ but is diffracted at the diffraction angle: θ ′ (≠ θ).

  By the way, the demonstration of the present invention for recognizing the variation of the pattern dimension by detecting the variation of the pattern pitch on the reticle is the result of examining the diffracted light diffracted at the boundary of the pattern. This is based on the fact that there is a correlation between fluctuations. That is, this is based on the fact that the variation of the pattern pitch can be replaced with the variation of the pattern dimension. Considered below.

  In FIG. 3, local pattern dimension variation (also referred to as CD (Criticai Dimension) variation) in the mask pattern 1 such as a reticle indicates the degree of minute protrusion / recession at the boundary 2 of the pattern edge as indicated by an arrow. Point to.

  That is, if the patterns 1 arranged in parallel are formed at a predetermined pattern pitch, the fluctuations in and out of the boundary portion 2 are within an allowable range. Therefore, this corresponds to a normal state in which the interval between the adjacent patterns 1 does not vary, which is not a problem. The simulation results of such a pattern dimension variation event are shown in FIGS.

  FIG. 4 shows the amount of variation in the pattern size at the boundary, where the horizontal axis indicates a unitless dimension and indicates an arbitrary position on the mask pattern reticle, and the vertical axis indicates the pattern size variation in nm. It is shown.

  In FIG. 4, the curve indicated by ● indicates the case where the pattern dimension fluctuation amount has changed most rapidly, and the curve indicated by the broken line indicates the case where the change in pattern dimension fluctuation amount is gradual, indicated by ○. The curved line shows the case where the change in the pattern dimension variation is the slowest.

  When this pattern dimension variation is replaced with the pitch variation of the mask patterns arranged in parallel on the reticle, the result is as shown in FIG. That is, FIG. 5 shows the pattern dimension variation amount of FIG. 4 converted into the pitch variation amount. The horizontal axis indicates a unitless dimension and indicates an arbitrary position on the mask pattern reticle, and the vertical axis indicates the pattern. The amount of dimensional variation is shown in units of nm.

  In FIG. 5, when the change in the pattern dimension variation amount as shown by the curve indicated by ● in FIG. 4 is the most rapid, the pattern pitch variation amount is locally increased as indicated by the curve indicated by ● in FIG. It has the largest peak. Then, the peak of the pitch fluctuation amount becomes lower as the change of the pattern dimension fluctuation amount is gradually reduced, and when the change of the pattern dimension fluctuation amount is the slowest as shown by the circle, the corresponding pitch This is because the peak amount of fluctuation is the most.

  As described above, the variation amount of the pattern dimension of the mask pattern shown in FIG. 4 can be replaced with the variation amount of the pitch of the mask pattern shown in FIG. In addition, a steep peak appears in the pitch variation amount in a local region where the variation of the pattern dimension is large. Therefore, if the part where the peak of the pitch fluctuation amount appears can be detected, the local part on the mask pattern can be specified.

  FIG. 6 shows a pattern pitch determined by a diffraction conditional expression: p × sin θ = nλ / 2 when a He—Ne laser having a wavelength of λ = 633 nm is used as a laser beam for scanning a mask pattern on a reticle to be inspected: The relationship between p (μm) and the diffraction angle of the first-order diffracted light: θ (°) is shown.

  In FIG. 6, the diffraction angle changes greatly with a slight change in the pattern pitch, for example, a change of 1 μm>. Therefore, if the variation of the pattern pitch defined at the boundary portion of the pattern is captured, the variation of the pattern pitch can be replaced with the variation of the local pattern dimension.

  FIG. 7 shows the relationship between the pattern dimension variation and the diffraction angle obtained by replacing the relationship between the pattern pitch variation and the diffraction angle shown in FIG. Here, assuming a 90 nm pattern on the wafer and a 180 nm pitch and a 4 × reticle, the diffraction angle satisfying the diffraction condition with the 633 nm He—Ne laser beam and the variation of the pattern dimension on the wafer [MEF (Mask Error) (Enhancement Factor) = 1, when the reticle magnification is 4 times). Variation in the pattern pitch in a region where the pattern dimension varies locally is equivalent to this pattern dimension error.

  FIG. 8 shows the region of ΔCD ± 9 nm in FIG. 7 and the amount of diffraction angle fluctuation corresponding thereto, which is used as a so-called working curve. From FIG. 8, it can be seen that when the allowable range of variation in the pattern dimension is ΔCD ± 9 nm, the diffraction angle range is 23.6 ° to 29.3 °.

  FIG. 9 is a structural example of a reticle inspection optical system for carrying out the present invention. In FIG. 9, a reticle 10 serving as an inspection object for examining pattern dimension variation is placed on an inspection stage 5 attached to a laser interferometer or the like and accurately positioned while being moved with high accuracy in a short time. Is done.

  As the laser light source 3, for example, a He—Ne laser with a wavelength of 633 nm is used. The laser beam 31 is irradiated on the surface of the reticle 10 as oblique light having a predetermined angle as shown in FIG. In order to scan the laser beam 31 on the surface of the reticle 10, the inspection stage 5 is moved with high accuracy.

  The laser beam 31 irradiated while being scanned on the surface of the reticle 10 is diffracted at the pattern edge of the mask pattern formed on the reticle 10, and the first-order diffracted light at the boundary 2 between the local dimensionally abnormal portion and the normal portion. The diffraction angle: θ varies. Therefore, the variation amount of the diffraction angle: θ is detected by the light receiving element 6 and, although not shown, the abnormal stage on the surface of the reticle 10 is specified by collating with the position read by synchronizing the inspection stage 5. .

In order to catch the fluctuation of the diffraction angle: θ, it is performed by detecting only the first-order diffracted light satisfying the diffraction condition or detecting the diffracted light not satisfying the diffraction condition in the light receiving element 6. The diffracted light distributed to either one is detected by the filter 7 arranged in (1).
[Example 1]
First-order diffracted light is detected using a He—Ne laser with a wavelength of 633 nm as the laser light. At the boundary between the local dimension abnormal part and the normal part, the dimension locally fluctuates as shown by ● in FIG. For this reason, the pitch defined by the pattern edge at each location varies sharply as shown in FIG. This variation corresponds to p ′ shown in FIG. This pitch variation is scanned with He-Ne laser light, and the first-order diffracted light is detected.

  The first-order diffracted light satisfies the relationship between the pitch and the diffraction angle shown in FIG. 5 according to the condition of the diffraction conditional expression (p × sin θ = nλ / 2).

  The position where the pitch variation showing a steep peak as shown in FIG. 5 appears can be identified by monitoring the first-order diffracted light intensity obtained from the diffraction angle together with the position information.

  The position information can be read, for example, by moving the reticle of the object to be measured on the stage of the laser interferometer.

As shown in FIG. 1, when an active region forming layer of a flash memory is taken as an example of an actual device having parallel line patterns, the pattern to be measured has a 2 μmL & S pattern on the reticle and a pitch of 4 μm.
[Example 2]
The laser beam spot is narrowed down to about the pattern pitch. Other conditions are the same as those in the first embodiment.

When the spot diameter of the laser beam is reduced to, for example, 5 μmφ, the sensitivity is increased with respect to the change in angle satisfying the diffraction condition of the diffracted beam. Incidentally, in FIG. 1, the size of the spindle-shaped laser beam is 5 μm.
Example 3
As shown in FIG. 9, the diffracted light of the laser light 31 is detected by extracting only the diffracted light 4 having a certain diffraction angle by arranging a fixed filter 7 that transmits only incident light from a predetermined angle. . Other conditions are the same as in the first embodiment.

  The position where the filter 7 is arranged is set in accordance with the diffraction angle: θ of the diffracted light 4 diffracted at the pattern edge of the normal portion of the mask pattern (not shown) of the reticle 10. Then, since the diffracted light 4 does not pass through the filter 7 at the boundary portion 2 with the abnormal portion pattern, the boundary portion between the abnormal portion and the normal portion can be specified by detecting the position.

  Here, a simulation result of a 180 nm pitch pattern using a He-Ne laser that is easy to handle as a light source of laser light is shown. However, the light source and the pattern pitch are not limited, and various modifications can be made to the simulation result and the working curve of the pattern dimension variation associated therewith.

  In scanning of the laser beam, the reticle is placed on the stage and moved. For example, the reticle may be moved in the X direction, and the laser beam may be scanned by rotating the polygon mirror in the Y direction. And various modifications are possible.

  Furthermore, although the case where the pattern is formed at a predetermined pitch is illustrated here, the present invention is not limited to the case of mask patterns arranged in parallel at the same pitch, and various modifications are possible.

It is a positional relationship figure of a pattern size fluctuation site and scanning laser light. It is a schematic diagram of the diffracted light of the normal / abnormal part of the pattern dimension. It is a schematic diagram of a local dimensional variation and a boundary part. It is a figure which shows the pattern dimension variation | change_quantity of a boundary part. It is the figure which converted the pattern dimension variation | change_quantity into the pitch variation | change_quantity. FIG. 5 is a relationship diagram between a pattern pitch and a diffraction angle. FIG. 6 is a relationship diagram between pattern dimension variation and diffraction angle. It is an example of the working curve of a pattern dimension fluctuation | variation and a diffraction angle. It is a structural example of the optical system which implements this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Mask pattern 2 Boundary part 3 Laser light source 31 Laser light 4 Diffracted light 5 Inspection stage 6 Light receiving element 7 Filter 10 Reticle

Claims (4)

Scanning the mask pattern surface of the reticle with laser light;
Detecting diffracted light diffracted at the pattern edge of the mask pattern surface;
Checking whether the diffracted light satisfies one diffraction condition;
And a step of detecting a variation portion of the pattern dimension of the mask pattern based on the collation. The reticle inspection method according to claim 1, wherein the mask pattern is formed at a predetermined pitch. The reticle inspection method according to claim 1, wherein the laser beam is narrowed down to at least a pitch dimension of the mask pattern. The method for inspecting a reticle according to claim 1, wherein the step of comparing the diffracted light diffracted at the pattern edge of the mask pattern surface is performed through a filter.

Images