JP2013231648A - Pattern measurement method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To measure a structure of an upper pattern (a measured pattern) without being affected by a base pattern.SOLUTION: A pattern measurement method of measuring the structure of the measured pattern serving as an upper pattern in an actual cell having a base pattern and the upper pattern formed on a substrate comprises: detecting an actual spectrum waveform resulting from a measurement of a pattern of the actual cell; detecting a dummy spectrum waveform resulting from a measurement of a dummy pattern without the base patten; comparing and collating the actual spectrum waveform with the dummy spectrum waveform; creating a measurement model so as to allocate the number of wavelength divisions of a theoretical spectrum only to a wavelength range of a common part where the actual spectrum waveform and the dummy spectrum waveform overlap with each other; performing a numerical calculation by a waveform fitting of the theoretical spectrum derived from the measurement model to the actual cell spectrum; and calculating the structure of the measured pattern on the basis of a result of the numerical calculation.

Description

  The present invention relates to a pattern measurement method for measuring the structure of a pattern to be measured formed on a substrate using a scatterometry (light wave scattering measurement) method.

  When manufacturing a semiconductor device, it is necessary to form a fine pattern on a substrate. In the manufacturing process of a semiconductor device, when forming a fine repetitive pattern on a substrate, a sidewall type double patterning method (double patterning process) is used (for example, refer to Patent Document 1 and Patent Document 2).

  Here, the double patterning method (double patterning process) is a method of forming a pattern by performing patterning twice or more. In the side wall type double patterning method (double patterning process), a pattern is formed using a side wall.

  After such a sidewall-type double patterning method (double patterning process) is used to form a fine repetitive pattern on a substrate, a CD-SEM (Critical Dimension-Scanning Electron) which is an ultra-fine dimension measuring apparatus using an electron beam is used. Measuring the dimensions of the pattern with a Microscope. Then, feedback is performed for dimension correction from the measured dimension result.

  The principle of CD-SEM is that the measurement sample is irradiated with primary electrons emitted from the electron gun, the secondary electrons emitted from the sample are captured by a detector, and the size of the target pattern is determined from the obtained intensity distribution. It is a method of measuring.

  On the other hand, even in the case of manufacturing a DRAM (dynamic random access memory) memory cell as a semiconductor device, a double patterning process is used to form a word trench with miniaturization.

  In manufacturing such a DRAM memory cell, an element isolation region is formed on a semiconductor substrate by a well-known STI (Shallow Trench Isolation) method, and a pad oxide film (matt oxide film) and a mask nitride film are formed on the semiconductor substrate in this order. After film formation, an MLD (Molecular Layer Deposition) film is formed as a mask pattern (MLD pattern) of the word trench by a double patterning method. At this time, since the MLD pattern is formed by a double patterning process, one side surface of the MLD pattern has a side-etched shape (concave shape).

  For this reason, since CD-SEM cannot be sufficiently managed, an actual cell monitor based on scatterometry, which is an optical and shape measuring instrument, has been studied (for example, see Patent Document 3).

  The pattern measurement method disclosed in Patent Document 3 irradiates a measurement pattern formed on a substrate with a light beam, detects an actual spectrum that is a spectrum of reflected diffracted light by the measurement pattern of the light beam, and detects the actual spectrum. Waveform fitting between the spectrum and a theoretical spectrum calculated from a model pattern prepared in advance is performed for each of a plurality of wavelength regions in the actual spectrum to measure the structure of the pattern to be measured.

Japanese Unexamined Patent Publication No. 2009-53547 (paragraph [0151], FIG. 1) JP 2011-107596 A (FIGS. 1 to 4) JP 2012-018097 A

  In a semiconductor device such as the above-described DRAM memory cell, an element isolation region formed by the STI method is formed as a base pattern, and a pattern such as an MLD pattern is formed as an upper pattern (pattern to be measured). Become.

  In the case of real cell measurement, the pattern measurement method disclosed in Patent Document 3 analyzes sensitivity, correlation, accuracy, etc. in all wavelength regions for all parameters including not only the ground pattern but also the ground pattern. And create.

  As a result, there is a problem of being affected by the base pattern.

  A pattern measurement method according to the present invention is a pattern measurement method for measuring the structure of a pattern to be measured, which is a ground pattern, in a real cell having a base pattern and a ground pattern on a substrate, and the measurement result of the pattern of the real cell The actual spectrum waveform is detected, the dummy spectrum waveform that is the measurement result of the dummy pattern without the base pattern is detected, the actual spectrum waveform is compared with the dummy spectrum waveform, and the actual spectrum waveform and the dummy spectrum waveform overlap. Create a measurement model so that the number of wavelength divisions of the theoretical spectrum is assigned only to the wavelength band of the common part, and perform numerical calculation by fitting the theoretical spectrum derived from the measurement model and the spectrum of the real cell spectrum. Based on the result, the structure of the pattern to be measured is obtained.

  According to the present invention, the structure of the upper pattern (pattern to be measured) can be measured without being affected by the base pattern.

It is a top view which shows the shape in the middle of manufacture of the semiconductor device used as the object of the pattern measuring method of this invention. It is sectional drawing about line A-A 'of FIG. It is sectional drawing which shows the theoretical shape for demonstrating the scatterometry which is an optical dimension and shape measuring device. It is sectional drawing which shows the actual shape for demonstrating the scatterometry which is an optical dimension and shape measuring device. It is a graph which shows the theoretical spectrum with respect to an ideal shape, and the real spectrum with respect to an actual shape for demonstrating the scatterometry which is an optical size and shape measuring device. It is sectional drawing which shows the 1st process of the double patterning process for demonstrating double patterning. It is sectional drawing which shows the 2nd process of the double patterning process for demonstrating double patterning. It is sectional drawing which shows the 3rd process of the double patterning process for demonstrating double patterning. It is sectional drawing which shows the 4th process of the double patterning process for demonstrating double patterning. It is sectional drawing which shows the 5th process of the double patterning process for demonstrating double patterning. It is sectional drawing which shows the 6th process of the double patterning process for demonstrating double patterning. It is sectional drawing of the real cell for demonstrating the problem of the prior art. It is sectional drawing of an ideal pattern for demonstrating the problem of the prior art. It is a graph which shows the real spectrum with respect to a real cell, the ideal spectrum with respect to an ideal pattern, and the dummy spectrum with respect to a dummy pattern for demonstrating the problem of the prior art. It is sectional drawing of the real cell for demonstrating the pattern measuring method which concerns on one Example of this invention. It is sectional drawing of a dummy pattern for demonstrating the pattern measuring method which concerns on one Example of this invention. It is a graph which shows the real spectrum with respect to a real cell, and the dummy spectrum with respect to a dummy pattern. It is sectional drawing of an ideal pattern for demonstrating the pattern measuring method which concerns on one Example of this invention. It is a graph which shows the real spectrum with respect to the real cell, the dummy spectrum with respect to a dummy pattern, and the ideal spectrum with respect to an ideal pattern for demonstrating the pattern measurement method which concerns on one Example of this invention.

  First, in order to facilitate understanding of the present invention, the background until the present invention is made will be described.

  With the miniaturization of semiconductor devices, a double patterning process is used to form word trenches. The core gap difference of the word trench is determined by the dimension of the MLD pattern during DP1 MLD (Molecular Layer Deposition) etching.

  In the following drawings, the scale and number of each structure are different from each other in order to make each configuration easy to understand. In addition, an XYZ coordinate system is set and the arrangement of each component will be described. In this coordinate system, the Z direction is a direction perpendicular to the surface of the semiconductor substrate, the X direction is a direction perpendicular to the Z direction in a plane parallel to the surface of the semiconductor substrate, and the Y direction is horizontal to the surface of the semiconductor substrate. This is a direction orthogonal to the X direction on a smooth surface. Thus, the X direction and the Y direction are orthogonal to each other. The X ′ direction is a direction inclined obliquely with respect to the X direction.

  FIG. 1A and FIG. 1B are diagrams showing a shape during the manufacture of a semiconductor device (target of the pattern measurement method of the present invention) during DP1MLD etching. FIG. 1A is a plan view showing the shape of the semiconductor device during manufacture, and FIG. 1B is a cross-sectional view taken along line A-A ′ of FIG. 1A.

  As shown in FIGS. 1A and 1B, the semiconductor device manufacturing method shown in FIG. 1 forms an element isolation region 2 extending in the X ′ direction inclined with respect to the X direction on the semiconductor substrate 1. After the pad oxide film 3a and the mask nitride film 3b are formed in this order, an MLD film is formed on the mask pattern (MLD pattern) 3c of the word trench by a double patterning method.

  At this time, since the MLD pattern 3c is formed by a double patterning process, the side surface on one side has a side-etched shape (concave surface shape) 3c ′. For this reason, CD-SEM (Critical Dimension-Scanning Electron Microscope) cannot be sufficiently managed, and an actual cell monitor using scatterometry, which is an optical dimension / shape measuring instrument, is being studied.

  Next, scatterometry will be described with reference to FIGS. 2A, 2B, and 2C. 2A is a cross-sectional view showing a theoretical shape, and FIG. 2B is a cross-sectional view showing an actual shape. FIG. 2C is a graph showing the theoretical spectrum for the theoretical shape and the actual spectrum for the actual shape. In FIG. 2C, the horizontal axis indicates the wavelength λ (nm), and the vertical axis indicates the power.

  As shown in FIG. 2A, it is assumed that when a reference object 6 is irradiated with a theoretical object 4, reflected light 7 associated with the theoretical shape of the object 4 is obtained. The spectrum is calculated theoretically. The theoretical spectrum calculation point 8 is shown in the graph of FIG. 2C. In addition, although connected with a line for easy understanding, it is actually a discrete value.

  Next, as shown in FIG. 2B, the actual measurement object 5 is irradiated with the reference light 6 to obtain reflected light 7 associated with the actual measurement object 5. The actual spectrum 9 is also shown in the graph of FIG. 2C.

  The difference between the theoretical spectrum calculation point 8 and the actual spectrum 9 is calculated as a difference between the actual shape of the measured object and the theoretical shape of the measured object 4, and the actual measured object shape ( Derive the structure.

  Next, a double patterning process used for forming the word trench will be described with reference to FIGS. 3A to 3F. 3A to 3F are diagrams sequentially showing the steps of the double patterning process.

  First, as shown in FIG. 3A, an element isolation region 2 embedded with an insulating film made of a silicon oxide film is formed on a semiconductor substrate 1 by a well-known STI (Sallow Trench Isolation) method.

  Subsequently, a mat oxide film (pad oxide film) 3a, a mask nitride film 3b, a first BARC (Bottom Anti-Reflection Coating) film 3e, and a resist R are sequentially formed and applied on the surface of the semiconductor substrate 1 by lithography. Then, a repetitive pattern of line width F and line spacing 3F is patterned.

  Next, as shown in FIG. 3B, the core pattern of the first BARC film 3e is formed by BARC etching.

  Next, as shown in FIG. 3C, an MLD-CVD (Molecular Layer Deposition-Chemical Vapor Deposition) film 3f is formed on the entire surface of the semiconductor substrate 1 so that the side film thickness is half the pitch width (1 / 2P). To do.

  Next, as shown in FIG. 3D, a second BARC film 3g is applied so as to bury the irregularities of the MLD-CDV film 3f on the entire surface of the semiconductor substrate 1. Next, as shown in FIG.

  Next, as shown in FIG. 3E and FIG. 3F, the second BARC film 3g, the MLD-CVD film 3f on the horizontal plane, and the first BARC film 3e are etched back by etching back to obtain a line width F and a line spacing. An MLD pattern 3c that repeats at a pitch of F is formed.

  FIG. 3E shows a state in the middle of etch back.

  Next, referring to FIG. 4A, FIG. 4B, and FIG. 4C, when measuring the pattern on pattern shape by scatterometry, that is, when measuring the dimension of the MLD pattern 3c during DP1MLD etching, The problem of will be described. 4A is a cross-sectional view of a real cell, and FIG. 4B is a cross-sectional view of an ideal pattern. FIG. 4C is a graph showing an actual spectrum for an actual cell, an ideal spectrum for an ideal pattern, and a dummy spectrum for a dummy pattern. In FIG. 4C, the horizontal axis indicates the wavelength λ (nm), and the vertical axis indicates the power.

  The ideal pattern shown in FIG. 4B is formed in the order of the mat oxide film 3a and the mask nitride film 3b on the semiconductor substrate 1, and an ideal MLD pattern (hereinafter also referred to as “ideal pattern”) 3d is used as a mask for the word trench. It is formed in a pattern.

  First, theoretical spectrum calculation points 12 (hereinafter, calculation points 12) of the ideal pattern 3d are calculated in the entire frequency region and plotted in the graph of FIG. 4C.

  Next, the real spectrum 10 (hereinafter referred to as “real waveform A”) of the real cell shown in FIG. 4A is measured and plotted on the graph of FIG. 4C. At this time, since the reflected light is affected by variations in the CD and depth of the element isolation region 2, the actual waveform A is distorted.

  A dummy spectrum 10 ′ (hereinafter referred to as “dummy waveform A ′”) when not affected by the base pattern is also plotted in the graph of FIG. 4C.

  If the shape (structure) of the MLD pattern 3c is calculated from the difference between the dummy waveform A 'and the calculation point 12, the correct shape can be obtained. However, since the actual waveform A is obtained, variations in the CD / depth of the element isolation region 2 cause a decrease in the measurement accuracy of the MLD pattern 3c.

  Since the calculation point 12 is calculated in the entire frequency region, it takes time to create a measurement model, and variations such as the CD and depth of the element isolation region 2 must be taken into consideration. There is a possibility that the model needs to be corrected every time the shape, CD, film thickness, etc. of the base pattern changes.

  That is, if the element isolation region 2 is generalized as a ground pattern and the MLD pattern 3c is a ground pattern as a pattern to be measured, the following three problems are concerned.

1) Variations in the CD and depth of the base pattern 2 cause a decrease in the measurement accuracy of the top pattern 3c.
2) Since the analysis including the ground pattern 2 is performed, it takes time to create a measurement model.
3) Each time the shape, CD, film thickness, etc. of the underlying pattern 2 changes, it may be necessary to modify the model.

  Next, a pattern measurement method according to an embodiment of the present invention will be described with reference to FIGS. 5A to 5E. FIG. 5A is a cross-sectional view of a real cell. FIG. 5B is a cross-sectional view of the dummy pattern. FIG. 5C is a graph showing a real spectrum for a real cell and a dummy spectrum for a dummy pattern. FIG. 5D is a cross-sectional view of an ideal pattern. FIG. 5E is a graph showing an actual spectrum for a real cell, a dummy spectrum for a dummy pattern, and an ideal spectrum for an ideal pattern. 5C and 5E, the horizontal axis indicates the wavelength λ (nm), and the vertical axis indicates the power.

  In the real cell shown in FIG. 5A, an element isolation region 2 is formed on a semiconductor substrate 1, a mat oxide film 3a and a mask nitride film 3b are formed in this order, and then an MLD film is formed as a mask pattern ( MLD pattern) 3c.

  First, the real cell portion shown in FIG. 5A is measured, and the real spectrum 10 (real waveform A) is plotted on the graph of FIG. 5C.

  The dummy pattern shown in FIG. 5B is formed by forming a mat oxide film 3a and a mask nitride film 3b in this order on the semiconductor substrate 1 in a region without the element isolation region 2, and then forming an MLD film as a mask pattern for a word trench by a double patterning method. (MLD pattern) formed on 3c.

  Next, the dummy pattern shown in FIG. 5B is measured, and the dummy spectrum 11 (hereinafter referred to as “dummy waveform B”) of the dummy pattern is also plotted on the graph of FIG. 5C.

  When the real waveform A and the dummy waveform B are compared, the spectrum common region 13 of the real cell and the dummy pattern (hereinafter simply referred to as “common region 13”) and the non-common region 14 of the spectrum of the real cell and the dummy pattern ( Hereinafter, it is simply referred to as “non-common area 14”.

  It can be considered that the non-common region 14 is affected by the element isolation region 2.

  Therefore, in this embodiment, the measurement model is such that the wavelength division number of the theoretical spectrum is assigned only to the wavelength band of the common portion 13 where the actual spectrum waveform (actual waveform A) and the dummy spectrum waveform (dummy waveform B) overlap. Create

  4D, the ideal pattern shown in FIG. 5D is formed in the order of the mat oxide film 3a and the mask nitride film 3b on the semiconductor substrate 1, and then an ideal MLD pattern (ideal pattern) 3d is formed as a mask for the word trench. It is formed in a pattern.

  Subsequently, assuming the ideal pattern 3d shown in FIG. 5D, the spectrum is theoretically calculated only in the frequency band corresponding to the common region 13, and as shown in FIG. “Calculation point 12”) is plotted on the same graph as the real waveform A and the dummy waveform B.

  In this way, numerical calculation is performed by waveform fitting of the theoretical spectrum (calculation point 12) derived from the measurement model and the real spectrum 10 (real waveform A) of the real cell.

  Next, the shape (structure) of the MLD pattern 3 c is calculated from the difference between the actual waveform A and the calculation point 12 in the range of the common region 13. That is, the structure of the measured pattern 3c is obtained based on the numerical calculation result.

  As described above, by not performing the calculation in the frequency band affected by the element isolation region 2, that is, the non-common region 14, the possibility of being affected by the element isolation region 2 is reduced, and the measurement accuracy is improved. In addition, since the amount of calculation is reduced, the time for creating the measurement model can be shortened, and the frequency band affected by the element isolation region 2 is excluded from the calculation, so that it is not necessary to consider the change of the element isolation region 2.

  According to the pattern measurement method of the present embodiment described above, the following effects can be obtained.

  The influence of the underlying pattern 2 on the measurement result can be reduced, the model analysis time can be shortened, and a reduction in the number of model corrections associated with changes in the shape, CD, film thickness, etc. of the underlying pattern 2 is expected. The reason is that the measurement result 10 of the actual cell pattern (actual waveform A) and the measurement result 11 of the dummy pattern without the base pattern 2 (dummy waveform B) are compared and collated, This is because the wavelength band is easily changed under the influence of the base pattern 2 and is excluded from the measurement model creation.

  The preferred embodiments of the present invention have been described above. However, the present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the spirit of the present invention. Needless to say, it is included in the range.

1 Semiconductor substrate 2 Element isolation region (underlying pattern)
3a Matte oxide film (pad oxide film)
3b Mask nitride film 3c MLD pattern (upper pattern)
3c 'side edge 3d ideal pattern (ideal MLD pattern)
3e First BARC film 3f MLD-CVD film 3g Second BARC film 4 Ideal measurement object 5 Actual measurement object 6 Reference light 7 Reflected light 8 Ideal spectrum calculation point 9 Actual spectrum 10 Actual spectrum of actual cell (Actual waveform A)
10 'dummy spectrum in the case of the influence of the ground pattern (dummy waveform A')
11 Dummy pattern dummy spectrum (dummy waveform B)
12 Theoretical spectrum theoretical spectrum calculation points (calculation points)
13 Common area of spectrum of real cell and dummy pattern (common area)
14 Real cell and dummy pattern spectrum non-common area (non-common area)
R resist F Line width (word trench groove width)
X X direction X 'X' direction Y Y direction Z Z direction

Claims (3)

A pattern measuring method for measuring a structure of a pattern to be measured which is the above ground pattern in a real cell having a ground pattern and a ground pattern on a substrate,
Detecting a real spectrum waveform which is a measurement result of the pattern of the real cell;
Detecting a dummy spectrum waveform that is a measurement result of the dummy pattern without the base pattern;
The actual spectrum waveform and the dummy spectrum waveform are compared and verified,
Create a measurement model so as to assign the number of wavelength divisions of the theoretical spectrum only for the wavelength band of the common part where the actual spectrum waveform and the dummy spectrum waveform overlap,
Performing a numerical calculation by waveform fitting of the theoretical spectrum derived from the measurement model and the spectrum of the real cell,
Based on the numerical calculation result, the structure of the measured pattern is obtained.
Pattern measurement method.   The pattern measurement method according to claim 1, wherein the upper pattern is an MLD pattern formed by a double pattern process.   The pattern measurement method according to claim 1, wherein the base pattern includes an element isolation region formed by an STI method.

Images