TWI614569B - Method and system of detecting defects in photomasks - Google Patents

Abstract

In some embodiments, a method and / or system may include detecting defects in a photomask. The method may include acquiring a first image of a first die. The method may include acquiring a second image of a second die. In some embodiments, the method may include dividing the first image and the second image into a plurality of first and second portions, respectively. The method may include reducing one or more differences in the sizes of the first and second portions. In some embodiments, the method may include determining a difference in a function derived from an image intensity between the corresponding first and second portions. The method may include summing the differences in the function between the corresponding first and second parts. The method may include detecting meso-scale defects in the second die.

Description

Method and system for detecting defects in photomask

The present invention relates generally to a system and method for detecting defects in reticle or reticle. More specifically, the present invention relates generally to systems and methods for detecting defects in reticle and / or reticle used to generate grains by comparing images (eg, images of the grains).

Semiconductor devices are currently required to have high density and performance associated with increased transistor and circuit speeds and improved reliability. These requirements require greater precision and uniformity in the formation of semiconductor devices, and careful program monitoring.

One program used in the production of semiconductor devices is lithography. In lithography, masks or "reticles" are used to transfer circuit patterns to semiconductor wafers. A reticle includes a set of complex geometric patterns corresponding to a circuit component to be integrated on a wafer. Each graticule in a series is used to transfer its corresponding pattern to a photosensitive layer. The transfer of the reticle pattern to the photosensitive layer is usually performed by an optical exposure tool that directs light or other radiation through the reticle to expose the photoresist. A photoresist is used to form a photoresist mask, and the underlying polycrystalline silicon or metal layer is selectively etched according to the mask to form features such as lines or gates.

It should be understood that any defects on the reticle, such as an additional chromium or a missing chromium, will be transferred to the manufactured wafer in a repeating manner. Therefore, check the markings and detect any defects on them Department is important.

Defects on a reticle or reticle are detrimental to wafer yield in semiconductor manufacturing processes. Traditionally, there are two inspection modes, grain-to-grain (D: D) and grain-to-database (D: DB). Both modes rely on a basic assumption: the number of defective pixels in a processing patch (defined as a small rectangular area on a reticle) is a small fraction of the total number of pixels present in the processing patch. Most algorithms use this assumption to reduce dynamic tool noise and mask noise. For example, there are methods to dynamically compensate for slight feature size differences between a test die and a reference die. Therefore, most defect detection methods are adjusted to find defects on a length scale of approximately 10 ¹ to 10 ² nm.

However, these existing methods do not have the capability or sensitivity to detect defects with a length scale comparable to a processing patch. These so-called mesoscopic defects may be caused by a mask writing error. If undetected, they may result in limited or no output.

In some embodiments, a method and / or system may include detecting defects in a photomask. The method may include acquiring a first image of a first die. The method may include acquiring a second image of a second die. In some embodiments, the method may include dividing the first image and the second image into a plurality of first and second portions, respectively. The method may include reducing (eg, minimizing) one or more differences in the size of the first and second portions. In some embodiments, the method may include determining a difference in a function derived from an image intensity between the corresponding first part and the corresponding second part. The method may include summing differences in the function between the corresponding first part and the corresponding second part. The method may include generating a graphical display based on the positions on the surface associated with the first die and the second die. The method may include detecting meso-scale defects in the second die.

In some embodiments, transmitted light or reflected light is used to acquire the first image and / or the second image.

In some embodiments, the first die includes a reference die. In some embodiments, the second die includes a test die.

導出。在一些實施例中，b在一第 一部分及/或一第二部分內變化，該部分再細分為子部分。在一些實施例中，方法包含在導出b之後映射第二晶粒中之缺陷。 In some embodiments, the method may include dividing the first image and the second image into a plurality of first and second portions, respectively. The method may include reducing one or more differences in the size of the first portion and the second portion. In some embodiments, reducing any difference in the size of the first image and the second image includes using b | ▽ I (x, y) |. b can be linearly proportional to the difference in critical dimension (CD) size. b can be minimized by the following objective function:

Export. In some embodiments, b varies within a first part and / or a second part, which is subdivided into sub-parts. In some embodiments, the method includes mapping defects in the second die after deriving b.

)，其中b指示特徵上之偏差量，且

指示一組模型化參數。校準理論上模型化晶粒可包含藉由最小化

來為各貼片影像凍結

及浮動b。 In some embodiments, the first die includes a theoretically modeled die. In some embodiments, the method includes calibrating theoretically modeled grains. Calibrating modeled grains in theory can include deriving a set of modeled parameters (b;

), Where b indicates the amount of deviation in features, and

Indicates a set of modeled parameters. Calibration theoretically modeling grains can include minimizing

To freeze each patch image

And floating b.

In some embodiments, a system may include a processor and a memory medium. The memory medium may be coupled to a processor that stores program instructions. The processor can execute program instructions to obtain a first image of a first die. The processor can execute program instructions to obtain a second image of a second die. The first image and the second image can be divided into approximately equal first and second parts by executing program instructions by the processor. Any difference in size between the first part and the second part can be reduced by the processor executing program instructions. The processor can execute program instructions to detect defects in the second die using the first die as a reference.

100‧‧‧ Get the first image of one of the first grains

110‧‧‧ Get a second image of a second die

120‧‧‧ Divides the first image and the second image into many first and second parts, respectively

130‧‧‧ Reduce one or more differences in size between part one and part two

140‧‧‧ Judging the image intensity between the corresponding first part and second part Difference in one of the functions

150‧‧‧ produces one of these differences based on the location associated with the first die and the second die

160‧‧‧ Detects mesoscale defects in the second grain

Because the following detailed description of the preferred embodiment and the benefits of referring to the accompanying drawings, familiarize yourself with Those skilled in the art can understand the advantages of the present invention.

FIG. 1 depicts one embodiment of one method of detecting defects in a photomask.

Although the invention is susceptible to various modifications and alternative forms, specific embodiments thereof are shown by way of example in the drawings and may be described in detail herein. The drawings may not be drawn to scale. It should be understood, however, that the additional drawings and detailed description are not intended to limit the invention to the particular form disclosed, but on the contrary, the intention is to cover all within the spirit and scope of the invention as defined by the scope of the accompanying patent Modifications, equivalents, and substitutions.

The headings used herein are for organizational purposes only and are not intended to limit the scope of the description. As used throughout this application, the word "may" is used in a permissive meaning (ie, meaning a possibility) rather than a mandatory meaning (ie, meaning a must). The word "include" indicates an open relationship and therefore means include, but is not limited. Similarly, the word "having" also indicates an open relationship, and therefore means having, but not limiting. As used herein, the terms "first", "second", "third", etc. are used as labels for the preceding nouns and do not imply any type order (e.g., space, time, logic, etc.) unless explicitly stated otherwise Indicate this order. For example, an "third die electrically connected to the module substrate" does not exclude the case where one of the "fourth die electrically connected to the module substrate" is connected before the third die unless otherwise specified. Similarly, a "second" feature does not require a "first" feature to be implemented before a "second" feature unless otherwise specified.

Various components may be described as "configured to" perform a task or tasks. In these contexts, "configured to" is one of the broader interpretations generally meaning "having a structure or tasks performed during operation". Thus, even when a component is not currently performing a task (for example, a set of electrical conductors can be configured to electrically connect one module to another, even when the two modules are not connected), the component can be configured To perform tasks. In some contexts, "configured" may be a broad interpretation of a structure that generally means "having a circuit that performs a task or tasks during operation". Thus, even when the component is not turned on at the time, the component can be configured to perform tasks. Generally speaking, the electricity A circuit can contain hardware circuits.

Various components may be described as performing a task or tasks for convenience of description. These descriptions can be interpreted as including the phrase "configured." Explain that one of the components configured to perform one or more tasks clearly intends not to call 35 U.S.C. § 112, paragraph 6, an explanation of the component.

The scope of the invention includes any feature or combination of features disclosed herein (either explicitly or implicitly) or any generalization thereof, whether or not it addresses any or all of the issues addressed herein. Therefore, during the prosecution of this application, a new patent application scope (or one of its priority applications) to any such combination of features can be formulated. In particular, referring to the scope of the accompanying patent application, the features of the subsidiary technical solutions can be combined with the features of the independent technical solutions, and the features of the respective independent technical solutions can be combined in any suitable manner and not only listed in the scope of the accompanying patent applications Specific combination.

It should be understood that the present invention is not limited to a given device or biological system, and of course, it may vary. It should also be understood that terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting. As used in this specification and the scope of the accompanying patent application, the singular forms "a" and "the" include the singular and the plural referents unless the content clearly indicates otherwise. Thus, for example, reference to "a connector" includes one or more connectors.

Unless defined otherwise, all technical or scientific terms used herein have the same meaning as commonly understood by one of ordinary skill.

The term "connected" as used herein generally refers to items that can be joined or connected together.

As used herein, the term "coupled" generally refers to items that are operated or joined or connected together with or without one or more intervening components.

As used herein, the term "directly" generally refers to a structure that is in physical contact with another structure or when used in connection with a procedure, means that there is no intermediate step or component involved. Circumstances where one program affects another program or structure.

The term "mesoscopy" as used herein generally refers to a length scale of approximately 10 ³ nm to 10 ² μm. For example, reference may be made herein to a mesoscopic defect that means a defect with a lateral dimension of approximately 10 ³ nm to 10 ² μm.

In some embodiments, a method and / or system may include defects in a photomask. FIG. 1 depicts one embodiment of one method of detecting a defect in a photomask. The method may include acquiring a first image 100 of a first die. The method may include acquiring a second image 110 of a second die. In some embodiments, the method may include dividing the first image and the second image into a plurality of first and second portions 120, respectively. The method may include reducing any differences 130 in the size of the first and second portions. In some embodiments, the method may include determining any difference 140 in a function derived from a corresponding image intensity between the first part and the second part. The method may include summing differences in functionality between the corresponding first part and the corresponding second part. The method may include generating a graphical display 150 of substantially all the differences based on the locations associated with the first die and the second die. The method may include detecting meso-scale defects 160 in the second die. In some embodiments, the method may include using a first die as a reference to detect mesoscale defects in the second die. In some embodiments, the method and / or system may be applied in a D: D detection mode.

In some embodiments, the first image and / or the second image are acquired using transmitted light or reflected light. In some embodiments, the first image and / or the second image is acquired using a combination of transmitted light and reflected light. For example, transmitted light and reflected light can be combined at a detector.

In some embodiments, the first die includes a reference die. In some embodiments, the second die includes a test die.

導 出。 In the D: D detection mode, for a specific processing patch, I _test (x, y) and I _ref (x, y) are respectively expressed as optical images of the self-referenced die and the test die. These optical images can be obtained from transmitted or reflected light. In some embodiments, one has performed that may assume image registration. In some embodiments, any difference in reducing the size of the first image and the second image includes using b | ▽ I (x, y) |. b can be linearly proportional to the difference in CD size. b can be the following objective function for self-minimization:

Export.

If b is slowly changing and does not change in a processing patch, the above sum can be used above all effective pixels in the patch. If b changes within a patch, one can divide a patch into several sub-tiles. For the purposes of the following discussion, b should be assumed to be a constant within a patch. In some embodiments, when b changes significantly within a first part and / or a second part of one of the first image and the second image, the part is subdivided into sub-parts. In some embodiments, the method may include mesoscopic defects mapped in the second die after deriving b.

Once b is determined for each patch, a disc level map can be generated. Defect detection thresholds are typically set at one level to capture outliers (e.g., defects) with lower damage or false counts (background noise).

3)及給定n-1量測(b_i)：

。 在以上方程式中，b_i係差分偏差且d_i係i^th晶粒之CD偏差量(非直接可量測)。一個期望的係d_i與平均CD偏差〈d〉之間之差分偏差，其中 〈d〉定義為以d_i’≡d_i-〈d〉形式之

。一者可展示

。使用以上組問題，一者可將 相鄰晶粒差分偏差資訊轉換為差分偏差，其中各晶粒可展示自所有晶粒之平均之CD偏差。 Consider the existence of n grains (n

3) and given n-1 measurement (b _i ):

. In the above equation, b _i d _i and the deviation differential-based system i ^th CD deviation of the grain (not directly measurable). The difference between a desired system d _i and the average CD deviation <d>, where <d> is defined as the form of d _i '≡d _i- <d>

. One can show

. Using the above set of problems, one can convert the differential deviation information of adjacent grains into differential deviations, where each grain can show the average CD deviation from all grains.

以最小化測試影像與 參考影像之間之差異。此處不同的係測試影像係光學的，且參考影像係理論上模型化的。在一些實施例中，方法包含校準理論上模型化晶 粒。校準理論上模型化晶粒可包含導出一組模型參數(b；

)其中b表 示特徵上之偏差量，且

表示一組模型參數。校準理論上模型化晶粒 可包含藉由最小化Σ[I _test (x,y)-I _ref (x,y;b)∣]²為各貼片影像凍結

及浮動b。恆定常數值b係為貼片之介觀偏差量之一近似。 In some embodiments, the method and / or system may be applied to a D: DB detection mode. In some embodiments, the first die includes a theoretically modeled die. One can use the same equation

To minimize the difference between the test image and the reference image. The different images here are optical, and the reference images are theoretically modeled. In some embodiments, the method includes calibrating theoretically modeled grains. Calibrating modeled grains in theory can include deriving a set of model parameters (b;

) Where b represents the amount of deviation in features, and

Represents a set of model parameters. Calibration die may comprise theoretically modeled by minimizing _{Σ [I test (x, y} ) - I ref (x, y; b) |] 2 for the patch image is frozen

And floating b. The constant value b is an approximation of the mesoscopic deviation of the patch.

It should be noted that the embodiments described herein are applied to various imaging modes. For example, both transmitted light and reflected light can be used to acquire images with a high-resolution microscope. These images are obtained by imaging conditions that are similar or identical to one of the imaging conditions of a stepper or scanner. In some embodiments, in some embodiments, inferring a parameter proportional to the change in CD over a mask based on a combination of optical image (D: D) and one of optical and modeled image (D: DB).

In some embodiments, a system may include a processor and a memory medium. The memory medium may be coupled to a processor that stores program instructions. The processor can execute program instructions to obtain a first image of a first die. The processor can execute program instructions to obtain a second image of a second die. The processor can execute program instructions to divide the first image and the second image into approximately equivalent first and second parts. Any difference in size between the first part and the second part can be reduced by the processor executing program instructions. The processor can execute program instructions to detect defects in the second die using the first die as a reference.

In the current method, the embodiments described herein have many advantages. Compared with standard defect detection methods that are only sensitive to microscopic defects, the present invention can detect mesoscopic and macroscopic defects. Compared with the intensity CDU-based method, the present invention is less sensitive to the pattern density effect and is a more direct measurement of one of the feature size differences of a multi-grained photomask. It is reported that there is no known existing method to detect mesoscopic feature size defects in a D: DB detection mode.

In this patent, certain US patents, US patent applications, and other materials (eg, articles) are incorporated by reference. However, the text of these US patents, US patent applications, and other materials is incorporated by reference only to the extent that there is no conflict between this text and other narratives and drawings set forth herein. In the event of this conflict, any such conflicting text incorporated herein by reference to the U.S. patent, U.S. patent application, and other materials is not specifically incorporated by reference herein.

In view of this description, those skilled in the art will appreciate further modifications and alternative embodiments of the various aspects of the present invention. Therefore, this description should be understood only as a drawing and as a general way to teach those skilled in the art to implement the present invention. It should be understood that the form of the invention shown and described herein is to be taken as the presently preferred embodiment. The Components and Materials Division replaces those shown and described herein, which can reverse parts and procedures, and can independently utilize certain features of the invention. Having the benefits of this description of the invention, those skilled in the art will understand all . Changes may be made in the elements described herein without departing from the spirit and scope of the invention as described in the scope of the following patent applications.

100‧‧‧ Get the first image of one of the first grains

110‧‧‧ Get a second image of a second die

120‧‧‧ Divides the first image and the second image into many first and second parts, respectively

130‧‧‧ Reduce one or more differences in size of the first and second parts

140‧‧‧ Judging the difference in a function of image intensity between the corresponding first and second parts

150‧‧‧ produces one of these differences based on the location associated with the first die and the second die

160‧‧‧ Detects mesoscale defects in the second grain

Claims (20)

A method for detecting defects in a photomask, comprising: acquiring a first image of a first die; acquiring a second image of a second die; separately dividing the first image and the second image divided into a plurality of first and second portions; and a reduced difference in the size of one or more of such second portions of such first portion comprising determining b | ▽ I (x, y ) |, and a threshold where b The size difference is linearly proportional; the difference in a function derived from an image intensity between the corresponding first and second parts is determined; the function between the corresponding first and second parts is added up These differences; and detecting mesoscale defects in the second die. The method of claim 1, wherein the first image and / or the second image are acquired using transmitted light and / or reflected light. The method of claim 1, further comprising generating a graphical display of the differences based on the positions associated with the first die and the second die. The method of claim 1, wherein the first die includes a reference die. The method of claim 1, wherein the second die includes a test die. The method as claimed in item 1, wherein b is to minimize the following objective function:

Export. If the method of item 1 is requested, when b is in the first image and the second image When changing in a first and / or a second part, the first and / or the second part is further subdivided into sub-parts. The method of claim 6, further comprising mapping defects in the second die after deriving b. The method of claim 1, wherein the first die includes a theoretically modeled die. The method of claim 9, further comprising calibrating the theoretically modeled grains, including: deriving a set of modeled parameters (b;

), Where b represents the amount of deviation in the feature, and

Represents a set of modeled parameters; and by minimizing

To freeze each patch image

And floating b. A system for detecting defects in a photomask, the system comprising: a processor; a memory medium coupled to the processor storing program instructions executable by the processor: obtaining a first die and a first Image; obtaining a second image of a second die; dividing the first image and the second image into a plurality of first and second portions, respectively; reducing the size of the first and second portions One or more differences, including a determination b | ▽ I (x, y) |, where b is linearly proportional to a critical size difference; determined from the corresponding first part and the corresponding second part An image intensity derivation of a difference in a function; summing the differences in the function between the corresponding first part and the second part; and detecting a mesoscopic scale in the second grain defect. If the system of claim 11, wherein the processor can further execute the program instructions to generate a graphical display of the differences according to the positions associated with the first die and the second die. . The system of claim 11, wherein the first die includes a reference die. The system of claim 11, wherein the second die includes a test die. As in the system of claim 11, wherein b is the minimization of the following objective function:

Export. The system of claim 11, wherein when b changes in a first part and / or a second part, the part is subdivided into sub-parts. The system of claim 16, further comprising the defect mapped in the second die after deriving b. The system of claim 11, wherein the first die includes a theoretically modeled die. The system of claim 18, further comprising calibrating the theoretically modeled grain, which further comprises: deriving a set of modeled parameters (b;

), Where b represents the amount of deviation in the feature, and

Represents a set of modeled parameters; and by minimizing

To freeze each patch image

And floating b. A method for detecting defects in a photomask, comprising: acquiring a first image of a first die, wherein the first die includes a theoretically modeled die; calibrating the theoretically modeled die, It includes: deriving a set of modeled parameters (b;

), Where b represents the amount of deviation in the feature, and

Represents a set of modeled parameters; and by minimizing

To freeze each patch image

And floating b; dividing the first image and the second image into a plurality of first and second parts, respectively; reducing one or more differences in the sizes of the first and second parts; A difference in a function derived from an image intensity between the corresponding first part and the second part; sum up the differences in the function between the corresponding first part and the second part; and detect the first Mesoscale defects in two grains.