Universal measurement model

Abstract

A metrology system can organize metrology measurement values regarding a plurality of metrology targets distributed within a plurality of fields on one or more specimens into the form of a signal vector, such that the metrology measurement values related to the metrology targets within each of the plurality of fields are grouped within that signal vector. The system can further decompose the signal vector into reconstruction vectors related to separate spectral components of the signal vector. The system can further classify a subset of those reconstruction vectors as components of a metrology model, such that the sum of those components is associated with a metrology model that describes the metrology measurement values on the one or more specimens. The system can further generate a control signal for controlling one or more processing tools based on the metrology model.

Description

Technical Field

[0001] Cross - Reference to Related Applications This application claims the benefit under 35 U.S.C. § 119(e) of U.S. Provisional Patent Application No. 63 / 347,133, filed May 31, 2022, entitled "MODELLING OVERLAY ON WAFER WITHOUT POLYNOMIALS: A UNIVERSAL MODEL", the entire content of which is incorporated herein by reference.

[0002] The present disclosure generally relates to metrology, and more specifically to modelling metrology measurements across multiple metrology targets.

Background Art

[0003] Metrology is commonly used in semiconductor processing for monitoring and / or controlling various manufacturing processes. For example, a sample may be provided with a plurality of exposed fields, and measurement values may be collected at various metrology targets within those fields. Then, it would be desirable to generate a model of those metrology measurement values to account for variations both within and between the fields.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, existing modelling techniques have various drawbacks, including but not limited to inaccuracies and lack of robustness. Therefore, there is a need to develop a system and method that eliminates the above - mentioned drawbacks.

Means for Solving the Problems

[0006] A metrology system is disclosed in accordance with one or more exemplary embodiments of the present disclosure. The system of one exemplary embodiment has a controller. Also, in one exemplary embodiment, the controller arranges metrology measurements regarding metrology targets distributed within various fields on one or more specimens into the form of a signal vector, and the metrology measurements related to the metrology targets within each field are grouped within that signal vector. Also, in one exemplary embodiment, the controller decomposes the signal vector into reconstruction vectors related to separate spectral components of the signal vector. Also, in one exemplary embodiment, the controller classifies a subset of those reconstruction vectors as components of a metrology model, and the sum of those components is associated with a metrology model that describes the metrology measurements on the one or more specimens. Also, in one exemplary embodiment, the controller generates control signals for controlling one or more processing tools based on the metrology model.

[0007] A metrology system is disclosed in accordance with one or more exemplary embodiments of the present disclosure. The system of one exemplary embodiment has a metrology subsystem that generates metrology measurements regarding metrology targets distributed within various fields on one or more specimens according to a metrology recipe. The system of one exemplary embodiment also has a controller. Also, in one exemplary embodiment, the controller arranges the metrology measurements into the form of a signal vector, and the metrology measurements regarding the metrology targets within each field are grouped within the signal vector. Also, in one exemplary embodiment, the controller decomposes the signal vector into reconstruction vectors regarding separate spectral components of the signal vector. Also, in one exemplary embodiment, the controller classifies a subset of the reconstruction vectors as components of a metrology model, and the sum of the components is associated with a metrology model that describes the metrology measurements on the one or more specimens. Also, in one exemplary embodiment, the controller generates control signals for controlling one or more processing tools based on the metrology model.

[0008] A metrology method is disclosed in accordance with one or more exemplary embodiments of the present disclosure. In one exemplary embodiment of the method, metrology measurements are generated regarding metrology targets distributed within various fields on one or more specimens. Also, in one exemplary embodiment of the method, the metrology measurements are arranged into the form of a signal vector such that the metrology measurements regarding the metrology targets within each field are grouped within the signal vector. Also, in one exemplary embodiment of the method, the signal vector is decomposed into reconstruction vectors regarding separate spectral components of the signal vector. Also, in one exemplary embodiment of the method, a subset of the reconstruction vectors is classified as components of a metrology model such that the sum of the components is associated with a metrology model that describes the metrology measurements on the one or more specimens. Also, in one exemplary embodiment of the method, control signals for controlling one or more processing tools are generated based on the metrology model.

[0009] It should be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not necessarily restrictive of the invention as claimed. The accompanying drawings, which are incorporated herein and constitute a part hereof, illustrate embodiments of the invention and, together with the general description, serve to explain the principles of the invention.

[0010] By referring to the following accompanying drawings, those skilled in the art (so-called persons having ordinary skill in the art) will be able to better understand several advantages of the present disclosure.

Brief Description of the Drawings

[0011]

Fig. 1A

Fig. 1B

Fig. 1C

Fig. 2

Fig. 3

Fig. 4

Fig. 5

Fig. 6

Fig. 7

Embodiments for Carrying Out the Invention

[0012] Reference is now made in detail to the disclosed subject matter, which is illustrated and described in connection with specific embodiments and individual features thereof. The embodiments described herein should be construed as illustrative and not restrictive. As will be readily apparent to those of ordinary skill in the art, various modifications and changes can be made in the form and details without departing from the spirit and scope of the disclosure.

[0013] Embodiments of the disclosure are directed to systems and methods for modeling metrology measurements taken across metrology targets in one or more fields of a specimen, relying on spectral decomposition of those metrology measurements. In view of the considerations herein, such techniques can provide universal modeling techniques for generating models of metrology measurements, and thus can provide a robust and accurate characterization of variations in metrology measurements across multiple length scales.

[0014] Metrology is widely used for monitoring and / or controlling manufacturing processes. By way of example in the semiconductor manufacturing context, a device can be fabricated through the sequential formation of patterned layers. For example, and without limitation, a series of steps such as deposition of a specimen layer and photoresist, lithographic exposure of the photoresist in a desired pattern, generation of a desired pattern in the specimen layer by etching the photoresist and specimen layer, and removal of the photoresist can be used to form a patterned layer. Typically, since many devices are fabricated in parallel on a single specimen, scribeline separation of those portions of the specimen that relate to different devices (or components thereof), called dice, can facilitate subsequent singulation. Further, lithography tools (e.g., scanners, steppers, etc.) typically have an exposure field (e.g., field size) that is much smaller than the specimen itself, and thus typically expose multiple fields throughout the specimen, allowing patterns for multiple dice to be placed within each field.

[0015] Typically, it is desirable to monitor and / or control the lithography exposure process based on metrology measurements at various metrology targets throughout the specimen. Such metrology targets can include dedicated metrology targets (e.g., overlay targets, etc.) and / or device features. And, one or more metrology models can characterize the variation in metrology measurements throughout a single specimen or multiple specimens. For example, since the various metrology measurements within a field typically follow a certain consistent pattern, considering multiple fields, the pattern will cycle periodically. Such a pattern can be captured by a field model, i.e., the field contribution to the full model. Also for example, the various metrology measurements throughout the specimen can exhibit relatively slow variations / variations (e.g., those at a length scale greater than the field size), which can be captured by a wafer model, i.e., the wafer contribution to the full model. Also for example, further variations in the metrology measurements can be regarded as noise and associated with the residuals of the field and / or wafer contributions of the model.

[0016] Generally, it would be desirable to generate robust field and / or wafer models that can accurately discriminate field and / or wafer components from noise. In other words, generally, it would be desirable that the residuals of the metrology model actually correspond to the noise within the measurements and do not contain field and / or wafer components.

[0017] In various embodiments, the measurement values of interest are hierarchically organized into the form of a signal vector. For example, the measurement values related to the measurement targets within each field can be grouped within that signal vector based on spatial proximity. Further, the measurement values related to different fields can be arranged within that signal vector based on spatial proximity. In various embodiments, that signal vector is decomposed (e.g., using techniques such as singular value decomposition (SVD)) into reconstruction vectors related to different spectral components (e.g., different length scales) of that signal vector. By doing so, the signal vector can be made equal to the sum of the reconstruction vectors. In various embodiments, moreover, and not limited to this, those reconstruction vectors are classified into various groups such as field components, wafer components, and residual components. For example, the sum of the reconstruction vectors classified as field components can be associated with a field model that describes the variation in measurement values among the measurement targets within each field. Similarly, the sum of the reconstruction vectors classified as wafer components can be associated with a wafer model that describes the variation in measurement values between fields (e.g., relatively low-frequency variation).

[0018] Upon careful consideration, such techniques can provide a robust and universal metrology model suitable for a wide range of applications and processes. In particular, by generating reconstruction vectors using spectral decomposition techniques, the dominant length scales (e.g., spectral components) in the signal vector can be efficiently identified and associated with field components, wafer components, and / or residuals. In other words, in the techniques disclosed in the present application, universality is brought about based on the analysis of metrology measurements within a given set, so that no external basis such as regression to a polynomial is required. Specifically, in regression-based techniques (e.g., polynomial-based regression techniques), a relatively large number of fitting parameters are usually required to perform proper fitting, and thus may be troubled by relatively low robustness against variations. Furthermore, a different polynomial basis may be required for accurate modeling of another application or process. In contrast, according to the spectral decomposition technique disclosed in the present application, relevant bases (e.g., reconstruction vectors) can be identified based on the data itself, and only a relatively small number of basis elements can be made sufficient for accurate characterization of field and / or wafer components. Thus, the techniques disclosed in the present application can be universal and robust.

[0019] Furthermore, such techniques can provide advantages over techniques that rely on principal component analysis. For example, an adaptive model that relies on principal component analysis is outlined in U.S. Patent Application No. 17 / 705077, filed on March 25, 2022, entitled "ADAPTIVE MODELING MISREGISTRATION MEASUREMENT SYSTEM AND METHOD," the entire contents of which are incorporated herein by reference. In such techniques, unnecessary assumptions can be made about the lack of correlation between various modeling terms (e.g., wafer terms and field terms). However, in the systems and methods disclosed in the present application, reconstruction vectors related to different length scales and classifiable into different models or their components (e.g., field models and wafer models) are provided without any such assumptions.

[0020] Additional embodiments of the present disclosure are directed to controlling one or more processing tools (e.g., lithography tools) based on fields and / or wafer components generated through spectral decomposition as disclosed herein.

[0021] Next, with reference to FIGS. 1A - 7, a metrology modeling system and method will be described in detail according to one or more embodiments of the present disclosure.

[0022] FIG. 1A is a block diagram of a metrology system 100 according to one or more embodiments of the present disclosure.

[0023] The metrology system 100 of the embodiments includes a metrology subsystem 102 suitable for generating metrology measurements of metrology targets 104 distributed throughout one or more specimens 106.

[0024] Specimen 106 can include all objects that are sources of metrological measurement values. For example, what can be included in specimen 106 is a substrate (e.g., wafer, etc.) formed of a semiconductor or non-semiconductor material. What can be included in a conductor or non-semiconductor material includes, but is not limited to, single-crystalline silicon, gallium arsenide, and indium phosphide. Specimen 106 may have one or more layers. For example, what can be included in such layers includes, but is not limited to, resist, dielectric material, conductive material, and semiconductor material. A wide variety of such layers are known in the art, and the term specimen as used in the present application is intended to include specimens on which various such layers can be formed. One or more layers formed on specimen 106 may or may not be patterned. For example, specimen 106 may have a plurality of dies, and each die may have a repetitive patterned feature. Through the formation and processing of such material layers, ultimately, a finished device can be obtained. A wide variety of devices can be formed on specimen 106, and the term specimen 106 as used in the present application is intended to include specimens 106 on which various devices known in the art are fabricated. Further, in view of the purpose of the present disclosure, the terms specimen 106 and wafer should be construed as interchangeable.

[0025] The metrology measurements of the specimen 106 generated by the metrology subsystem 102 result in some kind of data and may be generated using any technique known in the art. In certain embodiments, the metrology measurements generated by the metrology subsystem 102 are overlay measurements that characterize registration (or registration error) between separate lithographic exposures on one or more layers of the specimen 106. In certain embodiments, the metrology measurements generated by the metrology subsystem 102 characterize lithography exposure conditions such as, but not limited to, the focus position and illumination dose of the specimen 106. In certain embodiments, the metrology measurements generated by the metrology subsystem 102 characterize additional aspects of the manufacturing process such as, but not limited to, the etching process and the polishing process.

[0026] The metrology target 104 may be located anywhere on the specimen 106 suitable for metrology measurements by the metrology subsystem 102. In certain embodiments, the metrology target 104 has features (outer shape features) related to the device to be manufactured, i.e., those referred to as device features in the present application. In certain embodiments, the metrology target 104 has dedicated features designed to facilitate metrology measurements. According to an example in the overlay metrology context, dedicated overlay metrology targets 104 may be provided at overlapping or non-overlapping positions on the specimen 106 in an array that enables robust and accurate overlay measurements by the metrology subsystem 102 of features related to individual lithographic exposures of interest.

[0027] The metering system 100 of the embodiments has a controller 108 communicatively coupled to any member inside thereof. The controller 108 can be assumed to have one or more processors 110. For example, the one or more processors 110 can be configured to execute a set of program instructions held in a storage device 112 or memory. The one or more processors 110 of the controller 108 can include various processing elements known in the art. In that sense, the one or more processors 110 can include various microprocessor-type devices configured to execute algorithms and / or instructions.

[0028] One or more processors 110 of the controller 108 can include various processors or processing elements known in the art. For the purposes of the present disclosure, the terms “processor” or “processing element” can be defined broadly to include all devices having one or more processing or logic elements (e.g., one or more microprocessor devices, one or more application specific integrated circuit (ASIC) devices, one or more field programmable gate array (FPGA), or one or more digital signal processors (DSP)). In that sense, the one or more processors 110 can include various devices configured to execute algorithms and / or instructions (e.g., program instructions stored in memory). According to certain embodiments, in implementing the one or more processors 110, a desktop computer, a mainframe computer system, a workstation, an image computer, a parallel processor, a network-connected computer, or some other computer system can be configured to execute a program and, as described throughout the present disclosure, configure the program to operate or cooperate with the metering system 100. Further, processors or logic elements suitable for performing at least a portion of the steps described in the present disclosure can be provided within various subsystems of the metering system 100. Accordingly, the foregoing description should be construed as illustrative rather than as a limitation to the embodiments of the present disclosure. Further, the steps described throughout the present disclosure can be performed by a single controller or, alternatively, by a plurality of controllers. Additionally, the controller 108 can include one or more controllers housed within a common housing or within a plurality of housings. By doing so, various controllers or controller combinations can be individually packaged as modules suitable for integration into the metering system 100.

[0029] The memory device 112 can include various storage media known in the art suitable for storing a set of program instructions executable by one or more associated processors 110. For example, the memory device 112 can include a non-transitory storage medium. According to another example, the memory device 112 can include, but is not limited to, a read-only memory (ROM), a random access memory (RAM), a magnetic or optical storage device (e.g., disk), a magnetic tape, a solid state drive, etc. It should be further noted that the memory device 112 may be housed within a common controller housing together with one or more processors 110. According to certain embodiments, the memory device 112 can also be remotely located with respect to the physical location of one or more processors 110 and the controller 108. For example, one or more processors 110 of the controller 108 may access a remote memory (e.g., a server) accessible via a network (e.g., the Internet, an intranet, etc.).

[0030] The controller 108 can issue commands to and / or receive data from (e.g., via control signals) any component or subsystem of the metrology system 100, such as, but not limited to, the metrology subsystem 102, and can also do so with other systems, such as one or more process tools (e.g., a lithography tool, etc.) for fabricating the specimen 106. The controller 108 can further be configured to execute any of the various process steps described throughout the present disclosure.

[0031] In one embodiment, the metering system 100 has a user interface 114 communicatively coupled to a controller 108. According to one embodiment, the user interface 114 can include, but is not limited to, one or more desktop machines, laptop machines, tablets, and the like. Also, in one embodiment, the user interface 114 has a display, and the display is used to display data of the metering system 100 to the user. The display of the user interface 114 can include various displays known in the art. For example, the display can include, but is not limited to, a liquid crystal display (LCD), an organic light emitting diode (OLED) display, or a CRT display. It should be recognized by those skilled in the art that any display device that can be integrated with the user interface 114 is suitable for implementation in the present disclosure. Also, according to one embodiment, the user can input selections and / or commands in response to the data displayed to the user via a user input device provided in the user interface 114.

[0032] Hereinafter, with reference to FIGS. 2 to 7, the metering measurement modeling technology will be described in detail according to one or more embodiments of the present disclosure.

[0033] FIG. 2 is a flowchart depicting steps performed in a method 200 according to one or more embodiments of the present disclosure. According to the applicant's note, the embodiments and enabling technologies described above in the context of the metering system 100 should be understood to be extended to the method 200. That being said, according to a further note, the method 200 is not limited to the architecture of the metering system 100.

[0034] In the method 200 of the embodiments, at step 202, metering measurement values are generated for a metering target 104 distributed across fields 306 on one or more specimens 106 (e.g., according to a metering recipe).

[0035] FIG. 3 is a conceptual top view of the specimen 106 depicting the distribution of the metrology targets 104 according to one or more embodiments of the present disclosure.

[0036] As shown in FIG. 3, various die 302 with features created can be distributed throughout the specimen 106 and partitioned by scribe lines 304. Further, the various die 302 throughout the specimen 106 may be identical to each other. Additionally, the die 302 can be associated with separate fields 306 (e.g., exposure fields) related to a lithography tool (e.g., scanner, stepper, etc.). For example, the features on the die 302 can be created by a series of processing steps including, but not limited to, deposition of a specimen layer (e.g., dielectric layer, conductive layer, etc.), deposition of a photoresist, lithographic exposure of a pattern into the photoresist (e.g., through sequential exposure of the various fields 306), and etching of the photoresist and specimen layer.

[0037] FIG. 3 further depicts various metrology targets 104 distributed within each field 306. The metrology targets 104 can generally be located at any suitable location within the field 306, such as within the scribe line 304 and / or within any of the die 302. In embodiments, each field 306 has a common number and distribution of metrology targets 104.

[0038] The metrology measurement values measured in step 202 can include all kinds of data including, but not limited to, overlay measurement values, illumination dose measurement values, or focus position measurement values of the specimen 106 in the lithographic exposure process. Further, the metrology measurement values may be generated using any technique known in the art of the present technology. For example, the metrology measurement values may be generated using a metrology subsystem 102, although this is not essential.

[0039] The metrology measurements of the metrology targets 104 distributed throughout one or more specimens 106 may be captured in any order (e.g., using any sampling plan). In certain embodiments, the metrology measurements for the metrology targets 104 within a field 306 are captured prior to those for the metrology targets 104 within another field 306. In certain embodiments, the measurements for the metrology targets 104 throughout the specimen 106 are captured independent of the fields 306. For example, through the selection of a sampling plan, efficient generation of measurements for the metrology targets 104 throughout the specimen 106 can be achieved. Further, the metrology measurements of the metrology targets 104 may be captured either through parallel measurement of two or more metrology targets 104 or sequentially.

[0040] It is noted that FIG. 3 and the associated description are presented solely for illustrative purposes and should not be construed as limiting. For example, FIG. 3 is not drawn to true scale, and the number, type, size, and / or arrangement of features (e.g., die 302, metrology targets 104, fields 306, etc.) are merely illustrative.

[0041] In the method 200 of the embodiments, at step 204, the metrology measurements are organized into the form of a signal vector, and the metrology measurements for the metrology targets 104 within each field 306 are grouped within that signal vector. The signal vector may generally have any number of dimensions (e.g., one-dimensional, two-dimensional, etc.). Further, the order within the signal vector of the metrology measurements for the metrology targets 104 within each field 306 may be made consistent. By doing so, various systematic variations among the metrology measurements of the metrology targets 104 within each field 306 can result in a periodic signal vector. As will be shown later in connection with step 208, such systematic variations among the metrology measurements of the metrology targets 104 within each field 306 can be associated with the field model of the process.

[0042] According to an example of a one-dimensional signal vector, the signal vector S is

Number

[0043] The measurement values related to the measurement target 104 in each field 306 may generally have any order as long as the order is constant in all fields 306 in the signal vector (e.g., T 1 ,…,T n may correspond to any order of the individual measurement targets 104 in the field 306). However, depending on the embodiment, the order may be based on the physical layout of those measurement targets 104 in the field. For example, the order T 1 ,…,Tn The associated metrology targets 104 can be selected such that they are physically close to each other (e.g., adjacent to each other, nearest, etc.). According to the consideration in the present application, based on the physical layout of the metrology targets 104, by arranging the order of the metrology measurement values within each field 306 of the signal vector, the physical significant relationship between the data points within the signal vector can be promoted. For example, the systematic variation of the metrology measurement values within each field (e.g., the pattern captured by the field model) can be associated with the spatial variation within the manufacturing process across the specimens, and such an arrangement can be quickly captured.

[0044] According to certain embodiments, the signal vector can include metrology measurement values from two or more specimens 106. In this case, the metrology measurement values within the signal vector can be further grouped by the specimen 106. For example, the pattern in equation (1) can be repeated for each specimen 106. As will be described in detail later, such a configuration enables modeling of the specimen-to-specimen variation in the metrology measurement values.

[0045] In the method 200 of various embodiments, at step 206, the signal vector is decomposed into reconstruction vectors related to different spectral components of the signal vector. Various suitable spectral decomposition techniques can be used, such as, but not limited to, SVD, principal component analysis techniques, or some eigenvalue-based decomposition techniques. In certain embodiments, different spectral components of the signal vector (e.g., different length scales of the signal vector) are described by those reconstruction vectors. In such a case, the signal vector S can be expressed as the sum of those reconstruction vectors. For example, the signal vector S can be expressed as

Number

Number

[0046] According to the considerations in the present application, the reconstruction vectors generated using the spectral decomposition technology as disclosed in the present application can be based on the analysis of the measurement data itself (e.g., using eigenvalue-based technology, etc.). That is, the various technologies disclosed in the present application can provide a universal model (e.g., a universal measurement model) suitable for a wide range of applications and data sets. As a result, the various technologies disclosed in the present application can outperform, but not limited to, regression-based technologies such as polynomial regression technology that uses polynomials as basis vectors to represent measurement data. In particular, the accuracy of regression-based technologies can be limited by the ability of the selected basis to accurately represent the variations within the measurement data.

[0047] In the method 200 of the various embodiments, at step 208, a subset of those reconstruction vectors is classified as components of a measurement model, and the sum of those components is associated with a measurement model that describes the measurement values on one or more of those specimens.

[0048] At step 208, generally, those reconstruction vectors can be classified into any number of subsets, and those subsets can be associated with different aspects of the measurement model.

[0049] In the various embodiments, at step 208, a subset of those reconstruction vectors is classified as wafer components, and the sum of the reconstruction vectors classified as wafer components is associated with a wafer model that describes the variation in measurement values between fields 306. Such patterns can be characterized as low-frequency (LF) patterns or large-length scales where the variation in measurement between adjacent measurement targets 104 (e.g., T j versus T j+1 ) will be non-significant.

[0050] In various embodiments, at step 208, a subset of those reconstruction vectors is classified as field components, and the sum of the reconstruction vectors classified as field components is associated with a field model that describes the variation in metrology measurements between metrology targets 104 within each field 306. For example, the field model can describe that part of the variation of the signal vector S whose periodicity is on the order of the magnitude of the number (n) of sampled metrology targets 104 within each field 306. Such patterns can be characterized as mid-frequency (MF) patterns or mid-length scales, in which the pattern related to the metrology, whether the metrology measurements vary between various metrology targets 104 within the field 306, will be consistent between fields 306.

[0051] More generally, in method 200, the reconstruction vectors can also be classified into any number of significant-difference groups (e.g., classifications). In the case of multiple-sample metrology, a subset of those reconstruction vectors can also be classified as inter-sample components that can be associated with the drift of the process across multiple samples 106.

[0052] For example, in an example where both a field model and a wafer model are incorporated into the metrology model, the variation in metrology measurements can be described at various length scales (e.g., in-field length scale, between-field length scale, full-sample-width length scale, etc.). Further, a priori information or assumptions are not required in identifying the field and / or wafer model and related components.

[0053] In method 200 of various embodiments, at step 210, a subset of those reconstruction vectors is classified as residuals. For example, the remaining reconstruction vectors that were not classified as part of the metrology model (e.g., at step 208) can be regarded as residuals. Such patterns can be characterized as high-frequency (HF) patterns or small-length scales that do not fall into the above categories.

[0054] The resulting classified reconstruction vectors and / or associated models (e.g., parts of the metrology model) can be used for various suitable purposes.

[0055] In method 200 of the embodiments, at step 212, a control signal for controlling one or more processing tools is generated based on the metrology model. For example, the field model (and / or wafer model) can be used as control data for controlling some kind of processing tool, such as, but not limited to, a lithography tool (e.g., scanner, stepper, etc.). Thus, at step 212, collectables (correction variables) can be generated for one or more processing tools to compensate for overall unwanted deviations in those metrology measurements. Further, those collectables can be associated with feedback control related to the formation of similar structures on subsequent specimen 106 and / or feedforward control related to the formation of additional structures on additional layers of the same specimen 106 (e.g., one or more specimens related to its signal vector).

[0056] Next, referring to equations (3)-(4), the decomposition of the signal vector into the reconstruction vector will be detailed according to one or more embodiments of the present disclosure. Specifically, equations (3)-(4) represent non-limiting examples of singular spectrum analysis (SSA) techniques that utilize SVD. As shown in the present application, such techniques can be used to achieve the decomposition of the signal vector into both field components and wafer components (e.g., more generally, low-frequency, medium-frequency, and high-frequency patterns related to the number (n) of metrology targets 104 for each field 306).

[0057] In certain embodiments, the signal vector is mapped to matrix O in terms of

Number

Number

[0058] For example, each row (or column) of matrix O can incorporate a series of lagged vectors related to a subset of signal vector S. Such a matrix can be characterized as a transfer matrix or a Hankel matrix, and the elements O on the diagonal i + j = C ij can be made to have the property that they are equal.

[0059] Embedding dimension

Number

Number

Number

Number

Number

Number

[0060] In certain embodiments, SVD is performed on matrix O such that O can be written as

Number

[0061] Moreover, the reconstruction vector R k (see Equation (2) above) and each has the same length as the signal vector S

Number

[0062] Next, referring to FIGS. 4 to 7, an example of spectral decomposition relying on the non-limiting techniques represented by Equations (1) to (4) will be described. The sample 106 in this example has 27 fields 306 (N = 27), and each field has 10 measurement targets 104 (n = 10). Furthermore, the individual measurement values in this example are overlay (OVL) measurement values.

[0063] FIG. 4 is a plot of the signal vector S as formulated by Equation (1) according to one or more embodiments of the present disclosure. Specifically, FIG. 4 depicts the OVL measurement values for each site (e.g., measurement target 104) arranged according to Equation (1). The box 402 in FIG. 4 represents the first field 306 (e.g., F 1 ).

[0064] FIG. 5 relates to one or more embodiments of the present disclosure, and the embedded dimension is

Number

[0065] FIG. 6 relates to one or more embodiments of the present disclosure, and shows a series of plot diagrams depicting the reconstructed vector R k generated by the diagonal average of O k , where k = 1, …, 10.

[0066] As depicted in FIG. 6, each separate reconstructed vector R k contains separate frequency components (e.g., spectral components, length scale information, etc.) related to the signal vector S. It should be noted that according to equation (4) and FIG. 5, as k increases, the relative contribution of the reconstructed vector R k decreases.

[0067] Conversely, according to FIG. 2 (e.g., steps 208 and 210), by classifying those reconstructed vectors R k into various groups, various behaviors of the metrological measurement values at various locations of the specimen 106 can be quantified.

[0068] For example, R 3 may be classified as a wafer component (and thus a wafer model). Also for example, R 1 , R 2 , R 4 and R 5 may be classified as field components, and the sum of those components may be associated with a field model. Also for example, the reconstructed vector R k , where k = 6, …, 10, may be classified as a residual component, and the sum of those components may be taken as the residual of the model.

[0069] FIG. 7 relates to one or more embodiments of the present disclosure, and is a series of plots 702, 704, 706 in which a wafer model, a field model, and a residual are each depicted in comparison with a raw signal vector based on those classifications. As shown in plot 704, the field model captures well the periodic systematic variations in the metrology data in each field, the wafer model captures relatively large length-scale variations (e.g., inter-field variations), and the residual captures the measurement noise. Further, such an analysis can be performed without any a priori knowledge or assumptions about the field model, and without imposing any specific polynomial basis for the regression. Rather, the classification relies on the eigenvalue-based decomposition of the metrology data itself after being meaningfully organized (e.g., step 204) into a signal vector as described in equation (1). Thus, according to the techniques of the present disclosure, a universal model for characterizing metrology measurements can be provided.

[0070] It should be noted that the definition of the residual in the techniques of the present disclosure is different from that in the regression-based approach. In this case, the residual only relates to pure noise associated with non-significant eigenvalues, and it is not guaranteed that the average of the residuals will be exactly 0.

[0071] As previously described in the present application, the embedding length

Number

Number

[0072] Furthermore, according to the considerations in the present application, any suitable technique may be used to classify the reconstruction vectors (e.g., relying on step 208 and / or step 210). In certain embodiments, the reconstruction vectors are classified based on the relationship between the length scales captured by those reconstruction vectors and the resolution and / or operating range (e.g., scanning range) of the processing tool (e.g., lithography tool, etc.). For example, the lithography tool can control the placement of field 306 on specimen 106 within a certain operating range and within the frame of a certain resolution. Therefore, if the reconstruction vector includes the length scale related to its operating range and / or resolution, it can be classified as a field component. As an example, a reconstruction vector that includes a length scale smaller than the resolution of the lithography tool (e.g., high-frequency signal) can be regarded as a residual because all the control signals required to correct such signals exceed the capabilities of the lithography tool. According to certain embodiments, the reconstruction vectors can be classified at least in part through user input. For example, those reconstruction vectors can be displayed to the user via user interface 114 (e.g., in the form of a plot similar to those depicted in FIG. 6) for evaluation and input.

[0073] Here, according to the considerations in the present application with comprehensive reference to FIGS. 5 - 7, it is possible to describe the spectral components of signal vector S in a more efficient form than that directly provided by the reconstruction vectors. As an example, FIG. 5 shows several pairs of degenerate or nearly degenerate eigenvalues λ k (e.g., k = 1 and k = 2, k = 4 and k = 5, k = 6 and k = 7, k = 8 and k = 9). As also demonstrated in FIG. 6, the reconstruction vectors forming each of these pairs include similar frequencies and have a non-uniform envelope showing additional structure.

[0074] According to the method 200 of certain embodiments, by applying a rotation of the constituent reconstruction vector (or a portion thereof), a more efficient description of the spectral components of the signal vector S can be provided. Various suitable rotation techniques can be utilized, such as but not limited to, varimax rotation, quartimax rotation, or equimax rotation. Such steps can be performed, although not necessarily, as part of or prior to the classification of the reconstruction vector (e.g., that depicted in steps 208 and / or 210 of FIG. 2) as field components, wafer components, residuals, etc.

[0075] Here, referring to FIG. 2 and considering the present application, according to what has been contemplated in the present application, the spectral decomposition technique for metrology modeling as disclosed in the present application can also notify a metrology sampling plan (e.g., step 202). As shown in Equation (4) and FIG. 5, the matrix C = O T O (e.g., covariance matrix) has rank

Number

Number

Number

[0076] A sampling plan that describes which metrology targets 104 to sample to generate metrology data (e.g., that by the metrology subsystem 102) can be downsized by skipping metrology measurements (e.g., at step 202) and attempting to randomly select metrology targets 104. Then, matrix reconstruction techniques (e.g., gap filling techniques, etc.) can be used to fill in values where those skipped metrology measurements were. In other words, all of the metrology measurements according to step 202 can be generated based on the metrology data generated by sampling the corresponding metrology targets (e.g., by the metrology subsystem 102), so that all of the metrology targets 104 related to the signal vector S (e.g., the repeating pattern of metrology targets 104 within each field 306) can be included in a full sampling plan. However, in some embodiments, only a portion of the metrology targets 104 related to the signal vector S are included in the sampling plan. In this configuration, the first portion of the measurement results is generated based on the metrology data generated by sampling the corresponding metrology targets 104, while the second portion of the metrology measurement results is generated by matrix reconstruction based on the first portion of the metrology measurement results.

[0077] According to the considerations in this application, the specific number of skipped metrology measurements is not limited to this, but for a given application, the matrix C = O T The sparsity of O, the inaccuracies or artifacts introduced by replacing the skipped measurements with reconstructed values, and other tolerances, etc., can be determined for a given application based on various factors.

[0078] As an example, spectral decomposition of the metrology measurements based on the initial sampling plan can be performed (e.g., relying on method 200 or some subset of the steps included therein). By analyzing the covariance matrix C, if it is revealed that one or more eigenvalues λ k are 0 or have values that can be ignored (e.g., are less than a specified threshold), this can be taken as a sign of oversampling. Consequently, eigenvalues λ that are 0 or negligible kThe sampling plan can be selected and / or adjusted to reduce the number of. In other words, the eigenvalue λ having a value exceeding the specified threshold k The sampling plan can be selected and / or adjusted to provide a specified number. This selection and / or various adjustments can be made prior to the high volume manufacturing (HVM) phase (e.g., in a test or evaluation phase), or during the HVM phase (e.g., the adjustment affects the sampling plan for future measurements).

[0079] Referring again to FIGS. 1A-1C, various aspects of the metrology system 100 are described in detail in accordance with one or more embodiments of the present disclosure.

[0080] In certain embodiments, the metrology subsystem 102 is an optical subsystem that generates metrology measurement values by illuminating the metrology target 104 with light and collecting the light coming from the metrology target 104 as a measurement basis. FIG. 1B is a conceptual diagram of an optical metrology subsystem 102 according to one or more embodiments of the present disclosure.

[0081] The optical metrology subsystem 102 generally includes an illumination source 116 that generates an illumination beam 118 having some spectral or temporal profile, an illumination path 120 that directs the illumination beam 118 towards the metrology target 104, an objective lens 122 and other suitable members that collect the light from the metrology target 104 (referred to as sample light 124 in the present application), a detector 126, and a collection optical path 128 that directs the sample light 124 towards the detector 126. One or more polarizers, one or more beam splitters 134, one or more homogenizers, or one or more apodizers, etc., and some or any type of lenses 130 and other optical elements 132 suitable for operating the illumination beam 118 and / or the sample light 124 can be provided in the illumination path 120 and / or the collection optical path 128.

[0082] Various suitable techniques can be used to generate metrology measurement values by the optical metrology subsystem 102. In certain embodiments, the optical metrology subsystem 102 generates metrology measurement values based on an image of the metrology target 104 (e.g., the field-of-view image when the detector 126 is in the field of view conjugate to the specimen 106). In certain embodiments, the optical metrology subsystem 102 generates metrology measurement values using scatterometry (scattering measurement method) techniques based on the collected diffracted lights (e.g., those measured by the detector 126 in the pupil plane). Further, the specimen 106 may be stationary or moving during measurement.

[0083] In certain embodiments, the metrology subsystem 102 is a particle beam subsystem that generates metrology measurement values by irradiating the metrology target 104 with a particle beam (e.g., an electron beam, an ion beam, a neutral particle beam, etc.) and collecting, as a measurement basis, any combination of particles or light coming from the metrology target 104. FIG. 1C is a conceptual diagram of the particle beam metrology subsystem 102 according to one or more embodiments of the present disclosure.

[0084] The particle beam metrology subsystem 102 can be assumed to have a particle source 136 (e.g., an electron beam source, an ion beam source, etc.), and thus the illumination beam 118 can include a particle beam (e.g., an electron beam, a particle beam, etc.). The illumination source 116 can include various particle sources known in the art suitable for generating the illumination beam 118. For example, what can be included in the illumination source 116, but not limited to, is an electron gun or an ion gun. The particle beam metrology subsystem 102 can further be assumed to have an illumination path 120, and one or more particle beam lenses 138 (e.g., electrostatic, magnetic, single potential, double potential, etc. lenses) provided therein, or other beam manipulation elements (not shown), such as, but not limited to, one or more stigmators or aberration control elements, etc., can control one or more sides of the illumination beam 118. The particle beam metrology subsystem 102 can further be assumed to have a collection optical path 128 that directs some combination of particles or light towards at least one detector 126. For example, what can be included in the detector 126 is an electron collector (e.g., a secondary electron collector, a backscattered electron detector, etc.). Also, for example, what can be included in the detector 126 is a photon detector (e.g., a photodetector, an X-ray detector, a scintillation element coupled to a photomultiplier tube (PMT) detector, etc.) that detects electrons and / or photons coming from the specimen surface. As an example, in FIG. 1C, a detector 126 is depicted that is arranged to capture secondary electrons emitted from the metrology target 104.

[0085] The subject matter described in this application is sometimes depicted with various members included in, connected to, or coupled with other members. It should be understood that these illustrated configurations are merely exemplary, and in fact, many other configurations can be implemented to achieve the same functions. Conceptually, members arranged in some way to achieve the same function achieve their desired function through effective "cooperation." Therefore, any two members in this application that are combined to be able to achieve a specific function can be regarded as "cooperating" with each other to achieve their desired function, regardless of their configuration or intervening members. Similarly, any two members that are cooperating in this way can be regarded as being "connected," "coupled," or "bonded" to each other so that their desired function is achieved, and any two members that can be made to cooperate in this way can be regarded as being "bondable" to each other so that their desired function is achieved. Specific examples of "bondable" include, but are not limited to, members being physically capable of and / or physically interacting with each other, and / or members being wirelessly capable of and / or wirelessly interacting with each other, and / or members being logically capable of and / or logically interacting with each other.

[0086] Many of the present disclosure and its attendant advantages will be understood from the foregoing description, and it will also become apparent that various modifications can be made to the form, construction, and arrangement of the members without departing from the disclosed subject matter or sacrificing all of its main advantages. The forms described are for illustrative purposes only, and the intent of the claims hereinafter set forth is to cover, embrace such modifications. Further, it should be understood that it is the claims in a separate section that define the invention.

Claims

1. A metrology system, comprising a controller having one or more processors configured to execute program instructions, through execution of which the one or more processors collate metrology measurement values regarding a plurality of metrology targets distributed within a plurality of fields on one or more specimens into the form of a signal vector, such that the metrology measurement values regarding the metrology targets within each of the plurality of fields are grouped within that signal vector, decompose the signal vector into reconstruction vectors regarding separate spectral components of the signal vector, classify a subset of the reconstruction vectors as components of a metrology model, such that the sum of those components is associated with the metrology model that describes the metrology measurement values on the one or more specimens, and generate control signals for controlling one or more processing tools based on the metrology model. A metrology system.

2. The metrology system according to claim 1, wherein the metrology model corresponds to a field model that describes the variation in the metrology measurement values among the metrology targets within each of the plurality of fields.

3. The metrology system according to claim 1, wherein the one or more processing tools include a lithography tool that is controllable within a certain length scale range based on the control signal, and when classifying a subset of the reconstruction vectors as components of the metrology model, a subset of the subset of the reconstruction vectors that have spectral components related to the length scale range is classified as components of the metrology model.

4. The metrology system according to claim 1, wherein the metrology model corresponds to a wafer model that describes the variation in the metrology measurement values among the plurality of fields.

5. The metrology system according to claim 1, wherein those of the reconstruction vectors that are not classified as components of the metrology model are associated with residuals.

6. The metrology system according to claim 1, wherein the metrology model includes a field model and a wafer model, and when classifying a subset of the reconstruction vectors as components of the metrology model, Classify a first subset of the reconstruction vectors as field components, associate the field model with the sum of those field components, and classify a second subset of the reconstruction vectors as wafer components, and associate the wafer model with the sum of those wafer components, A metrology system. **Claim 7** The metrology system according to claim 6, wherein those of the reconstruction vectors that are not classified as components of the field model or the wafer model are associated with a residual. **Claim 8** The metrology system according to claim 6, wherein the control signal for controlling the one or more processing tools is based on at least one of the field model and the wafer model. **Claim 9** The metrology system according to claim 1, wherein all of the metrology measurement values are generated based on metrology data generated by sampling the corresponding metrology target among the plurality of metrology targets. **Claim 10** The metrology system according to claim 1, wherein a first portion of the metrology measurement values is generated based on metrology data generated by sampling the corresponding metrology target among the plurality of metrology targets, and a second portion of the metrology measurement values is generated by matrix reconstruction based on the first portion of the metrology measurement values. **Claim 11** The metrology system according to claim 1, wherein the order of the metrology measurement values related to the metrology targets within each of the plurality of fields in the signal vector is constant. **Claim 12** The metrology system according to claim 1, wherein the one or more samples include two or more samples, the metrology measurement values related to each of the two or more samples are grouped in the signal vector, and the order of the metrology measurement values related to the metrology targets within each of the two or more samples is constant. **Claim 13** The metrology system according to claim 1, wherein when decomposing the signal vector into reconstruction vectors related to different spectral components of the signal vector, the reconstruction vectors are generated by performing a singular spectral analysis of the signal vector. **Claim 14** The metrology system according to claim 1, wherein the signal vector is one-dimensional. **Claim 15** The metrology system according to claim 1, wherein the signal vector is two-dimensional.

16. The metrology system according to claim 1, wherein when decomposing the signal vector into the reconstruction vectors corresponding to the respective spectral components of the signal vector, the signal vector is mapped into a matrix O 【Number 1】 with an embedding dimension and O is decomposed to 【Number 2】 Sparse matrix O ｋ is generated, and Based on the matrix O ｋ generate the reconstruction vector R ｋ thereof. obtain a metrology system.

17. The metering system according to claim 16, wherein elements O on the diagonal i + j = C ｉｊ are equal.

18. The metering system according to claim 16, when generating the matrix O by decomposing O ｋ ​ A metering system that generates the matrix O by performing singular value decomposition of O ｋ ​

19. The metering system according to claim 16, wherein when generating the matrix O by decomposing O ｋ ​ A metering system that generates the matrix O by performing principal component analysis of O. ｋ ​

20. The metering system according to claim 16, wherein the reconstruction vector R ｋ is C=O Ｔ is a metering system related to k eigenvalues of O, and when organizing the metering measurement values regarding the plurality of metering targets distributed in the plurality of fields on the one or more specimens into the state of the signal vector A measurement system that converts the measurement value into the state of the signal vector based on the sampling plan, and the eigenvalue λ ｋ A measurement system in which the sampling plan is selected so that a specified number of values having a value exceeding the specified threshold are provided.

21. The metrology system according to claim 16, wherein the embedding dimension 【Number 3】 is equal to the number of metrology targets in each of the plurality of fields.

22. The metrology system according to claim 21, wherein the signal vector is [Number 4] is formed as, MET corresponds to a specific one of the metrological measurement values, N is the number of the plurality of fields, and the superscript F １ , …, F Ｎ corresponds to the plurality of fields, 【Number 5】 is the number of the measurement targets in each of the plurality of fields, and the subscript T １ , …, T ｎ is a measurement system corresponding to the measurement targets in each of the plurality of fields.

23. The metrology system according to claim 22, wherein 【Number 6】 and where 【Number 7】 to obtain a metrology system.

24. The metering system according to claim 16, wherein, based on the matrix O ｋ when generating the reconstruction vector R ｋ therefrom The matrix O ｋ Based on the diagonal average of ｋ A measurement system that generates the reconstruction vector R

25. The metrology system according to claim 1, wherein the metrology measurement values include overlay metrology measurement values, and the plurality of metrology targets include overlay metrology targets.

26. A metrology system, comprising: a metrology subsystem configured to generate metrology measurement values for a plurality of metrology targets distributed in a plurality of fields on one or more specimens according to a metrology recipe; a controller having one or more processors configured to execute program instructions; wherein, through execution of the program instructions, the one or more processors reorganize the metrology measurement values into the form of a signal vector such that the metrology measurement values for the metrology targets in each of the plurality of fields are grouped within the signal vector; decompose the signal vector into reconstruction vectors corresponding to the respective spectral components of the signal vector; classify a subset of the reconstruction vectors as components of a metrology model such that the sum of those components corresponds to the metrology model that describes the metrology measurement values on the one or more specimens; and generate control signals for controlling one or more processing tools based on the metrology model. A metrology system.

27. The metrology system according to claim 26, wherein the metrology subsystem includes an optical metrology subsystem.

28. The metrology system according to claim 26, wherein the metrology subsystem includes a particle beam metrology subsystem.

29. The metering system according to claim 1, wherein all of the metering measurement values are generated based on metering data generated by sampling, by the metering subsystem, a corresponding metering target among the plurality of metering targets.

30. The metering system according to claim 1, wherein a first portion of the metering measurement values is generated based on metering data generated by sampling, by the metering subsystem, a corresponding metering target among the plurality of metering targets, and a second portion of the metering measurement values is generated by matrix reconstruction based on the first portion of the metering measurement values.

31. The metering system according to claim 26, wherein the metering model corresponds to a field model that describes the variation of the metering measurement values among the metering targets within each of the plurality of fields.

32. The metering system according to claim 26, wherein the metering model corresponds to a wafer model that describes the variation of the metering measurement values among the plurality of fields.

33. The metering system according to claim 26, wherein those of the reconstruction vectors that are not classified as components of the metering model are associated with the residuals.

34. The metering system according to claim 26, wherein the metering model includes a field model and a wafer model, and when classifying a certain subset of the reconstruction vectors as components of the metering model, classify a first subset of the reconstruction vectors as field components and associate the field model with the sum of those field components, and classify a second subset of the reconstruction vectors as wafer components and associate the wafer model with the sum of those wafer components. Metering system.

35. A metering method, comprising: generating metering measurement values for a plurality of metering targets distributed in a plurality of fields on one or more specimens, reorganizing the metering measurement values into the form of a signal vector such that the metering measurement values related to the metering targets within each of the plurality of fields are grouped within the signal vector, decomposing the signal vector into reconstruction vectors related to separate spectral components of the signal vector. Classify a certain subset of the reconstruction vectors as components of a measurement model, and the sum of these components is in a state associated with the measurement model that describes the measurement values on the one or more specimens, and generate control signals for controlling one or more processing tools based on the measurement model, Measurement method.

36. The measurement model according to claim 35, wherein the measurement model corresponds to a field model that describes the variation of the measurement values among the measurement targets within each of the plurality of fields.

37. The measurement model according to claim 35, wherein the measurement model corresponds to a wafer model that describes the variation of the measurement values among the plurality of fields.

38. The measurement model according to claim 35, wherein those of the reconstruction vectors that are not classified as components of the measurement model are associated with residuals.

Images