Patent Application
20250166161
The embodiments provided herein are related to a system and method for automatic defect classification, and more particularly for improving defect classification nuisance rate.
In manufacturing processes of integrated circuits (ICs), unfinished or finished circuit components are inspected to ensure that they are manufactured according to design and are free of defects. Inspection systems utilizing optical microscopes or charged particle (e.g., electron) beam microscopes, such as a scanning electron microscope (SEM) may be employed. As the physical sizes of IC components continue to shrink, accuracy and yield in defect detection become more and more important.
Some inspection tools may generate a large number of nuisances. A nuisance may be a detection aberration or a wafer irregularity that is not considered a defect of concern. For example, a nuisance could be caused by misidentified background images or minor defects that are not consequential to the device yield of a manufacturing process. As defect review becomes increasingly critical, there is a continuing need to reduce a nuisance rate in inspection processes.
The embodiments provided herein disclose a particle beam inspection apparatus, and more particularly, an inspection apparatus using a plurality of charged particle beams.
Some embodiments of the present disclosure include a method for improving a nuisance rate in image inspection data. The method may comprise obtaining image data comprising a set of candidate defects; developing a plurality of defect review types and a plurality of nuisance review types; classifying the set of candidate defects into one or more defect types based on the plurality of defect review types during a first classification phase; classifying the set of candidate defects into a plurality of nuisance types based on the plurality of nuisance review types during the first classification phase, and applying a machine learning based multi-phase classification to the classified set of candidate defects.
Some embodiments of the present disclosure include a system for improving a nuisance rate in image inspection data, comprising: a charged particle beam apparatus including a detector; an image acquirer that includes circuitry to receive a detection signal from the detector and construct an image including a first feature; and a controller with at least one processor and a non-transitory computer readable medium comprising instructions that, when executed by the at least one processor, cause the system to: obtain image data comprising a set of candidate defects; develop a plurality of defect review types and a plurality of nuisance review types; classify the set of candidate defects into one or more defect types based on the plurality of defect review types during a first classification phase; classify the set of candidate defects into a plurality of nuisance types based on the plurality of nuisance review types during the first classification phase, and apply a machine learning based multi-phase classification to the classified set of candidate defects.
Some embodiments of the present disclosure include a non-transitory computer readable medium storing a set of instructions that is executable by one or more processors of a system to cause the system to perform a method comprising: obtaining image data comprising a set of candidate defects; developing a plurality of defect review types and a plurality of nuisance review types; classifying the set of candidate defects into one or more defect types based on the plurality of defect review types during a first classification phase; classifying the set of candidate defects into a plurality of nuisance types based on the plurality of nuisance review types during the first classification phase, and applying a machine learning based multi-phase classification to the classified set of candidate defects.
Other advantages of the embodiments of the present disclosure will become apparent from the following description taken in conjunction with the accompanying drawings wherein are set forth, by way of illustration and example, certain embodiments of the present invention.
The above and other aspects of the present disclosure will become more apparent from the description of exemplary embodiments, taken in conjunction with the accompanying drawings.
Reference will now be made in detail to exemplary embodiments, examples of which are illustrated in the accompanying drawings. The following description refers to the accompanying drawings in which the same numbers in different drawings represent the same or similar elements unless otherwise represented. The implementations set forth in the following description of exemplary embodiments do not represent all implementations consistent with the invention. Instead, they are merely examples of apparatuses, systems, and methods consistent with aspects related to the subject matter as recited in the appended claims. For example, although some embodiments are described in the context of utilizing electron beams, the disclosure is not so limited. Other types of charged particle beams can be similarly applied. Furthermore, other imaging systems can be used, such as optical imaging, photo detection, x-ray detection, etc.
Electronic devices are constructed of circuits formed on a piece of semiconductor material called a substrate. The semiconductor material may include, for example, silicon, gallium arsenide, indium phosphide, or silicon germanium, or the like. Many circuits may be formed together on the same piece of silicon and are called integrated circuits or ICs. The size of these circuits has decreased dramatically so that many more of them can be fit on the substrate. For example, an IC chip in a smartphone can be as small as a thumbnail and yet may include over 2 billion transistors, the size of each transistor being less than 1/1000th the size of a human hair.
Making these ICs with extremely small structures or components is a complex, time-consuming, and expensive process, often involving hundreds of individual steps. Errors in even one step have the potential to result in defects in the finished IC, rendering it useless. Thus, one goal of the manufacturing process is to avoid such defects to maximize the number of functional ICs made in the process; that is, to improve the overall yield of the process.
One component of improving yield is monitoring the chip-making process to ensure that it is producing a sufficient number of functional integrated circuits. One way to monitor the process is to inspect the chip circuit structures at various stages of their formation. Inspection can be carried out using a scanning charged-particle microscope (SCPM). For example, an SCPM may be a scanning electron microscope (SEM). A SCPM can be used to image these extremely small structures, in effect, taking a “picture” of the structures of the wafer. The image can be used to determine if the structure was formed properly in the proper location. If the structure is defective, then the process can be adjusted, so the defect is less likely to recur.
As the physical sizes of IC components continue to shrink, accuracy and yield in defect detection become more important. Inspection images such as SEM images can be used to identify or classify a defect(s) of the manufactured ICs. To improve defect detection performance, it is important to reduce a nuisance rate in defect detection and classification. A nuisance may be a detection aberration or a wafer irregularity that is not considered a defect of concern in the inspection process. For example, a nuisance could be caused by misidentified background images or minor defects that are not consequential to the device yield of a manufacturing process. A nuisance rate may be a measure of how many nuisances are mistakenly classified as actual defects of concern during an inspection and classification process. For example, if a detection and classification process identifies 100 defects and it is later determined that 10 of those were actually misclassified nuisances, then a nuisance rate may be 10/100=10%. In some detection and classification systems, a nuisance rate may be unacceptably high.
Conventional classification schemes may use machine learning classifiers to sort inspection data into various categories, such as a nuisance category and a plurality of defect type categories. A conventional system may train a classifier using a set of training data. The conventional system may sort a large amount of nuisance data into a single nuisance category. This can cause issues in machine learning algorithms whose optimization methods are built on the assumption that data will be sorted into the different categories with somewhat equal numbers. When a real-world data set strays too far from this assumption, the classification accuracy may suffer. This issue may be called a data imbalance problem, and it may degrade classifier accuracy, causing the classifier to inadvertently classify nuisances as actual defects.
Embodiments of the disclosure may provide an automatic defect classification (ADC) system and method. The method includes acquiring a set of inspection image data from an inspection tool such as a SEM inspection tool. The image data may include, e.g., a set of defect candidates identified from SEM scans. The ADC method develops a plurality of defect review types and a plurality of nuisance review types. A machine learning (ML) classifier classifies inspection image data into bins according to the defect review types and nuisance review types. The use of plural nuisance types enables the ML classifier to better recognize nuisances and reduces the rate at which nuisances are misclassified as defects.
In some embodiments, the ADC may use multiphase classification. In the multiphase classification, the results of a first classification phase are compared to a manual review of the image data. Based on the comparison results, a set of misclassified data selected from the first phase is re-labeled correctly and put into a revised training pool for further training in another phase. Multiple iterations of this process may be performed until the training results meet expectations.
In some embodiments, a nuisance review type may be created based on a relationship or similarity to a defect review type. In some embodiments, a nuisance review type may be designed to identify data that is likely to be misclassified as a certain defect type. In some embodiments, a plurality of nuisance review types may be developed to smooth out a data imbalance problem. Such a problem can be caused by using one disproportionately large nuisance bin.
Relative dimensions of components in drawings may be exaggerated for clarity. Within the following description of drawings, the same or like reference numbers refer to the same or like components or entities, and only the differences with respect to the individual embodiments are described. As used herein, unless specifically stated otherwise, the term “or” encompasses all possible combinations, except where infeasible. For example, if it is stated that a component may include A or B, then, unless specifically stated otherwise or infeasible, the component may include A, or B, or A and B. As a second example, if it is stated that a component may include A, B, or C, then, unless specifically stated otherwise or infeasible, the component may include A, or B, or C, or A and B, or A and C, or B and C, or A and B and C. Reference is now made to
One or more robotic arms (not shown) in EFEM 106 may transport the wafers to load/lock chamber 102. Load/lock chamber 102 is connected to a load/lock vacuum pump system (not shown) which removes gas molecules in load/lock chamber 102 to reach a first pressure below the atmospheric pressure. After reaching the first pressure, one or more robotic arms (not shown) may transport the wafer from load/lock chamber 102 to main chamber 101. Main chamber 101 is connected to a main chamber vacuum pump system (not shown) which removes gas molecules in main chamber 101 to reach a second pressure below the first pressure. After reaching the second pressure, the wafer is subject to inspection by electron beam tool 104. Electron beam tool 104 may be a single-beam system or a multi-beam system. A controller 109 is electronically connected to electron beam tool 104. Controller 109 may be a computer configured to execute various controls of EBI system 100. While controller 109 is shown in
Beam tool 104 comprises a charged-particle source 202, a gun aperture 204, a condenser lens 206, a primary charged-particle beam 210 emitted from charged-particle source 202, a source conversion unit 212, a plurality of beamlets 214, 216, and 218 of primary charged-particle beam 210, a primary projection optical system 220, a motorized wafer stage 280, a wafer holder 282, multiple secondary charged-particle beams 236, 238, and 240, a secondary optical system 242, and a charged-particle detection device 244. Primary projection optical system 220 can comprise a beam separator 222, a deflection scanning unit 226, and an objective lens 228. Charged-particle detection device 244 can comprise detection sub-regions 246, 248, and 250.
Charged-particle source 202, gun aperture 204, condenser lens 206, source conversion unit 212, beam separator 222, deflection scanning unit 226, and objective lens 228 can be aligned with a primary optical axis 260 of apparatus 104. Secondary optical system 242 and charged-particle detection device 244 can be aligned with a secondary optical axis 252 of apparatus 104.
Charged-particle source 202 can emit one or more charged particles, such as electrons, protons, ions, muons, or any other particle carrying electric charges. In some embodiments, charged-particle source 202 may be an electron source. For example, charged-particle source 202 may include a cathode, an extractor, or an anode, wherein primary electrons can be emitted from the cathode and extracted or accelerated to form primary charged-particle beam 210 (in this case, a primary electron beam) with a crossover (virtual or real) 208. For ease of explanation without causing ambiguity, electrons are used as examples in some of the descriptions herein. However, it should be noted that any charged particle may be used in any embodiment of this disclosure, not limited to electrons. Primary charged-particle beam 210 can be visualized as being emitted from crossover 208. Gun aperture 204 can block off peripheral charged particles of primary charged-particle beam 210 to reduce Coulomb effect. The Coulomb effect may cause an increase in size of probe spots.
Source conversion unit 212 can comprise an array of image-forming elements and an array of beam-limit apertures. The array of image-forming elements can comprise an array of micro-deflectors or micro-lenses. The array of image-forming elements can form a plurality of parallel images (virtual or real) of crossover 208 with a plurality of beamlets 214, 216, and 218 of primary charged-particle beam 210. The array of beam-limit apertures can limit the plurality of beamlets 214, 216, and 218. While three beamlets 214, 216, and 218 are shown in
Condenser lens 206 can focus primary charged-particle beam 210. The electric currents of beamlets 214, 216, and 218 downstream of source conversion unit 212 can be varied by adjusting the focusing power of condenser lens 206 or by changing the radial sizes of the corresponding beam-limit apertures within the array of beam-limit apertures. Objective lens 228 can focus beamlets 214, 216, and 218 onto a wafer 230 for imaging, and can form a plurality of probe spots 270, 272, and 274 on a surface of wafer 230.
Beam separator 222 can be a beam separator of Wien filter type generating an electrostatic dipole field and a magnetic dipole field. In some embodiments, if they are applied, the force exerted by the electrostatic dipole field on a charged particle (e.g., an electron) of beamlets 214, 216, and 218 can be substantially equal in magnitude and opposite in a direction to the force exerted on the charged particle by magnetic dipole field. Beamlets 214, 216, and 218 can, therefore, pass straight through beam separator 222 with zero deflection angle. However, the total dispersion of beamlets 214, 216, and 218 generated by beam separator 222 can also be non-zero. Beam separator 222 can separate secondary charged-particle beams 236, 238, and 240 from beamlets 214, 216, and 218 and direct secondary charged-particle beams 236, 238, and 240 towards secondary optical system 242.
Deflection scanning unit 226 can deflect beamlets 214, 216, and 218 to scan probe spots 270, 272, and 274 over a surface area of wafer 230. In response to the incidence of beamlets 214, 216, and 218 at probe spots 270, 272, and 274, secondary charged-particle beams 236, 238, and 240 may be emitted from wafer 230. Secondary charged-particle beams 236, 238, and 240 may comprise charged particles (e.g., electrons) with a distribution of energies. For example, secondary charged-particle beams 236, 238, and 240 may be secondary electron beams including secondary electrons (energies≤50 eV) and backscattered electrons (energies between 50 eV and landing energies of beamlets 214, 216, and 218). Secondary optical system 242 can focus secondary charged-particle beams 236, 238, and 240 onto detection sub-regions 246, 248, and 250 of charged-particle detection device 244. Detection sub-regions 246, 248, and 250 may be configured to detect corresponding secondary charged-particle beams 236, 238, and 240 and generate corresponding signals (e.g., voltage, current, or the like) used to reconstruct an SCPM image of structures on or underneath the surface area of wafer 230.
The generated signals may represent intensities of secondary charged-particle beams 236, 238, and 240 and may be provided to image processing system 290 that is in communication with charged-particle detection device 244, primary projection optical system 220, and motorized wafer stage 280. The movement speed of motorized wafer stage 280 may be synchronized and coordinated with the beam deflections controlled by deflection scanning unit 226, such that the movement of the scan probe spots (e.g., scan probe spots 270, 272, and 274) may orderly cover regions of interests on the wafer 230. The parameters of such synchronization and coordination may be adjusted to adapt to different materials of wafer 230. For example, different materials of wafer 230 may have different resistance-capacitance characteristics that may cause different signal sensitivities to the movement of the scan probe spots.
The intensity of secondary charged-particle beams 236, 238, and 240 may vary according to the external or internal structure of wafer 230, and thus may indicate whether wafer 230 includes defects. Moreover, as discussed above, beamlets 214, 216, and 218 may be projected onto different locations of the top surface of wafer 230, or different sides of local structures of wafer 230, to generate secondary charged-particle beams 236, 238, and 240 that may have different intensities. Therefore, by mapping the intensity of secondary charged-particle beams 236, 238, and 240 with the areas of wafer 230, image processing system 290 may reconstruct an image that reflects the characteristics of internal or external structures of wafer 230.
In some embodiments, image processing system 290 may include an image acquirer 292, a storage 294, and a controller 296. Image acquirer 292 may comprise one or more processors. For example, image acquirer 292 may comprise a computer, server, mainframe host, terminals, personal computer, any kind of mobile computing devices, or the like, or a combination thereof. Image acquirer 292 may be communicatively coupled to charged-particle detection device 244 of beam tool 104 through a medium such as an electric conductor, optical fiber cable, portable storage media, IR, Bluetooth, internet, wireless network, wireless radio, or a combination thereof. In some embodiments, image acquirer 292 may receive a signal from charged-particle detection device 244 and may construct an image. Image acquirer 292 may thus acquire SCPM images of wafer 230. Image acquirer 292 may also perform various post-processing functions, such as generating contours, superimposing indicators on an acquired image, or the like. Image acquirer 292 may be configured to perform adjustments of brightness and contrast of acquired images. In some embodiments, storage 294 may be a storage medium such as a hard disk, flash drive, cloud storage, random access memory (RAM), other types of computer-readable memory, or the like. Storage 294 may be coupled with image acquirer 292 and may be used for saving scanned raw image data as original images, and post-processed images. Image acquirer 292 and storage 294 may be connected to controller 296. In some embodiments, image acquirer 292, storage 294, and controller 296 may be integrated together as one control unit. The raw image data may be processed to identify a set of defect candidates in the image data.
In some embodiments, image acquirer 292 may acquire one or more SCPM images of a wafer based on an imaging signal received from charged-particle detection device 244. An imaging signal may correspond to a scanning operation for conducting charged particle imaging. An acquired image may be a single image comprising a plurality of imaging areas. The single image may be stored in storage 294. The single image may be an original image that may be divided into a plurality of regions. Each of the regions may comprise one imaging area containing a feature of wafer 230. The acquired images may comprise multiple images of a single imaging area of wafer 230 sampled multiple times over a time sequence. The multiple images may be stored in storage 294. In some embodiments, image processing system 290 may be configured to perform image processing steps with the multiple images of the same location of wafer 230.
In some embodiments, image processing system 290 may include measurement circuits (e.g., analog-to-digital converters) to obtain a distribution of the detected secondary charged particles (e.g., secondary electrons). The charged-particle distribution data collected during a detection time window, in combination with corresponding scan path data of beamlets 214, 216, and 218 incident on the wafer surface, can be used to reconstruct images of the wafer structures under inspection. The reconstructed images can be used to reveal various features of the internal or external structures of wafer 230, and thereby can be used to reveal any defects that may exist in the wafer.
In some embodiments, the charged particles may be electrons. When electrons of primary charged-particle beam 210 are projected onto a surface of wafer 230 (e.g., probe spots 270, 272, and 274), the electrons of primary charged-particle beam 210 may penetrate the surface of wafer 230 for a certain depth, interacting with particles of wafer 230. Some electrons of primary charged-particle beam 210 may elastically interact with (e.g., in the form of elastic scattering or collision) the materials of wafer 230 and may be reflected or recoiled out of the surface of wafer 230. An elastic interaction conserves the total kinetic energies of the bodies (e.g., electrons of primary charged-particle beam 210) of the interaction, in which the kinetic energy of the interacting bodies does not convert to other forms of energy (e.g., heat, electromagnetic energy, or the like). Such reflected electrons generated from elastic interaction may be referred to as backscattered electrons (BSEs). Some electrons of primary charged-particle beam 210 may inelastically interact with (e.g., in the form of inelastic scattering or collision) the materials of wafer 230. An inelastic interaction does not conserve the total kinetic energies of the bodies of the interaction, in which some or all of the kinetic energy of the interacting bodies convert to other forms of energy. For example, through the inelastic interaction, the kinetic energy of some electrons of primary charged-particle beam 210 may cause electron excitation and transition of atoms of the materials. Such inelastic interaction may also generate electrons exiting the surface of wafer 230, which may be referred to as secondary electrons (SEs). Yield or emission rates of BSEs and SEs depend on, e.g., the material under inspection and the landing energy of the electrons of primary charged-particle beam 210 landing on the surface of the material, among others. The energy of the electrons of primary charged-particle beam 210 may be imparted in part by its acceleration voltage (e.g., the acceleration voltage between the anode and cathode of charged-particle source 202 in
The images generated by SEM may be used for defect inspection. For example, a generated image capturing a test device region of a wafer may be compared with a reference image capturing the same test device region. The reference image may be predetermined (e.g., by simulation) and include no known defect. If a difference between the generated image and the reference image exceeds a tolerance level, a potential defect may be identified. For another example, the SEM may scan multiple regions of the wafer, each region including a test device region designed as the same, and generate multiple images capturing those test device regions as manufactured. The multiple images may be compared with each other. If a difference between the multiple images exceeds a tolerance level, a potential defect may be identified.
In some embodiments, a computer system may be provided that can identify defects in a wafer image and classify the defects into categories according to the defect type. For example, once a wafer image is acquired, it may be transmitted to the computer system for processing.
Referring to
Wafer inspection system 310 may be any inspection system that generates an inspection image of a wafer. The wafer may be a semiconductor wafer substrate, or a semiconductor wafer substrate having one or more epi-layers or process films, for example. Wafer inspection system 310 may be any currently available or developing wafer inspection system. The embodiments of the present disclosure do not limit the specific type for wafer inspection system 310. Such a system may generate a wafer image having a resolution so as to observe key features on the wafer (e.g., less than 20 nm).
ADC server 320 may include a communication interface 322 that is electrically coupled to the wafer inspection system 310 to receive the wafer image. ADC server 320 may also include a processor 324 that is configured to analyze the wafer image and detect and classify defects that appear on the wafer image and may use a defect knowledge file in this analysis, detection, or classification. The defect knowledge file may be manually provided to ADC server 320 by an operator. Alternatively, according to some embodiments of the present disclosure, the defect knowledge file may be automatically provided to ADC server 320 by knowledge recommendation server 330.
For example, knowledge recommendation server 330 may be electrically coupled to the ADC server 320. Knowledge recommendation server 330 may include a processor 332 and a storage 334. Processor 332 may be configured to build a plurality of defect knowledge files and to store the plurality of defect knowledge files in storage 334. The plurality of defect knowledge files may contain information related to various types of defects generated during various stages of wafer manufacturing processes. The various stages of wafer manufacturing processes may include, but are not limited to, a lithography process, an etching process, a chemical mechanical polishing (CMP) process, or an interconnection forming process.
Processor 332 may be configured to build the plurality of defect knowledge files based on a plurality of defect patch images. The plurality of defect patch images may be generated by a wafer inspection tool, such as electron beam tool 104 illustrated in
Processor 332 may be trained, via a machine learning process, to build a knowledge file related to a specific type of defect based on a plurality of defect patch images of that type of defect. For example, processor 332 may be trained to build a knowledge file related to broken line defects generated in an interconnect forming process based on a plurality of defect patch images of broken line defects.
Processor 332 may also be configured to, in response to a request for knowledge recommendation from ADC server 320, search for a knowledge file that matches a wafer image included in the received request and provide the knowledge file to the ADC server 320.
Storage 334 may store an ADC data center that contains a plurality of defect knowledge files related to various types of defects generated during various stages of wafer manufacturing processes. The plurality of defect knowledge files in the ADC data center may be built by processor 332 of knowledge recommendation server 330. Alternatively, a portion of the defect knowledge files in storage 334 may be preset by a user or an external computer system and may be preloaded into storage 334.
A defect knowledge file may include general information about a single type of defect. The general information may include patch images and feature parameters to be used for later classification (e.g., size, edge roughness, depth, height, etc.) of the single type of defect. Alternatively, according to some embodiments of the present disclosure, a defect knowledge file may include general information about a plurality of types of defects that are present in the same process layer of a wafer. The single process layer may be, for example, a substrate layer, an epitaxial layer, a thin film layer, a photoresist layer, an oxide layer, a metal interconnection layer, etc.
Reference is now made to
Classifier 404 may be a machine learning classifier. Both unsupervised machine learning and supervised machine learning models may be used to predict one or more defects. Without limiting the scope of the claims, applications of supervised machine learning algorithms are described below.
Supervised learning is the machine learning task of inferring a function from labeled training data. The training data is a set of training examples. In supervised learning, each example is a pair consisting of an input object (typically a vector) and a desired output value (also called the supervisory signal). A supervised learning algorithm analyzes the training data and produces an inferred function, which can be used for mapping new examples. An optimal scenario will allow the algorithm to correctly determine the class labels for unseen instances. This requires the learning algorithm to generalize from the training data to unseen situations in a “reasonable” way (see inductive bias).
Given a set of N training examples of the form {(x1, y1), (x2, y2), . . . , (xN, yN)} such that xi is the feature vector of the i-th example and yi is its label (e.g., class), a learning algorithm seeks a function g: X→Y, where X is the input space and Y is the output space. A feature vector is an n-dimensional vector of numerical features that represent some object. Many algorithms in machine learning require a numerical representation of objects, since such representations facilitate processing and statistical analysis. When representing images, the feature values might correspond to the pixels of an image, when representing text perhaps term occurrence frequencies. The vector space associated with these vectors is often called the feature space. The function g is an element of some space of possible functions G, usually called the hypothesis space. It is sometimes convenient to represent function g using a scoring function f: X×Y→R such that function g is defined as returning the y value that gives the highest score: g(x)=arg maxyf(x, y). Let F denote the space of scoring functions.
Although G and F can be any space of functions, many learning algorithms are probabilistic models where g takes the form of a conditional probability model g(x)=P(y|x), or f takes the form of a joint probability model f(x, y)=P(x, y). For example, naive Bayes and linear discriminant analysis are joint probability models, whereas logistic regression is a conditional probability model.
There are two basic approaches to choosing for g: empirical risk minimization and structural risk minimization. Empirical risk minimization seeks the function that best fits the training data. Structural risk minimization includes a penalty function that controls the bias/variance tradeoff.
In both cases, it is assumed that the training set contains a sample of independent and identically distributed pairs, (xi, yi). In order to measure how well a function fits the training data, a loss function L: Y×Y→R≥0 can be defined. For training example (xi, yi), the loss of predicting the value ŷ is L(yi, ŷ).
The risk R(g) of function g is defined as the expected loss of g. This can be estimated from the training data as Remp(g)=1/NΣiL(yi, g(xi))
Exemplary models of supervised learning include decision trees, ensembles (bagging, boosting, random forest), k-NN, linear regression, naive Bayes, neural networks, logistic regression, perceptron, support vector machine (SVM), relevance vector machine (RVM), and/or deep learning. In some embodiments, the classifier is a Classifier 4 (C4) type used in the e-Manager® ADC system from HMI, Inc. C4 includes a set of decision trees, constructed by selecting a subset of features randomly selected during tree construction. Classification is done by majority voting of the decision trees under a chosen number of trees in the forest and number of variables randomly selected. The above machine learning examples are not meant to be limiting, and that other machine learning algorithms are possible within the scope of the present disclosure.
To evaluate result accuracy of a classification tree, a set of SEM image data may be subjected both to the classification and to a manual review. The manual review is a more rigorous and labor-intensive analysis to determine with higher confidence the actual nature of each candidate in the set. An operator can then compare the binning of each candidate under ADC to the actual result as determined by review. For example, with reference to
The results of the manual review may be compared to the classification results to determine performance metrics such as a nuisance rate NR. A nuisance rate NR can be a measure of “false positives,” that is, the percentage of candidates that were labeled as defects under ADC but determined to be nuisances under review. Dividing the number of false positives by the number of total positives gives a nuisance rate NR. These results are illustrated below.
Two rectangular boxes in
This nuisance rate is unacceptably high, showing that classifier 504 fails to sufficiently filter out unwanted candidates. Other nuisance rate calculations may be employed using the information in
To improve the nuisance rate, a multi-phase system is used.
However, feature sizes in IC manufacturing are always shrinking and the packing density of circuit features is constantly increasing. Therefore, a need exists to further reduce the nuisance rate NR to the greatest extent possible.
After the first classification phase, the results are compared to results of a manual review. Based on different pattern features and strength thresholds, some misclassified defects and misclassified nuisances from each type are re-labeled correctly and put into a revised training pool 812 for a second training in phase II. Multiple iterations of this process can be performed until the training result meets expectations, similar to the process discussed above with respect to
In some embodiments, a nuisance review type may be created based on its relationship or similarity to a defect review type. Stated another way, a nuisance review type may be designed to identify a candidate type that has the potential to be misclassified as a certain defect. Referring back to
In some embodiments, nuisance review types are designed to break down the large data imbalance in classification categories. When a classifier lumps all nuisances into one type, the classification count in that category is much higher than the counts in any other category. Referring back to
To counteract this problem, a plurality of nuisance review types may be created for the purpose of smoothing out the data imbalance. These nuisance review types may not be based on any similarity to particular defect review types, but instead can be designed with the goal of breaking down the large nuisance count into a plurality of smaller ones. In some embodiments, the plurality of nuisance review types may result in a plurality of bins having somewhat similar counts. For example, the largest and smallest counts of a plurality of nuisance types may be within ten times of each other. In some embodiments, the plurality of nuisance review types may be designed such that the largest classification count in a nuisance type can be within a multiple (e.g., 5×) of the smallest classification count.
Embodiments of the present disclosure can greatly reduce a nuisance rate in a defect inspection process. By more effectively filtering nuisances out of detection data, embodiments of the present disclosure may more accurately identify and compensate actual defects, leading to increased accuracy and higher yield in a device manufacturing process.
At step 1001, an inspection tool such as a SEM acquires a set of image data. The image data includes a set of candidate defects to be classified by a ML classifier. The inspection tool may be, e.g., a tool 104 as discussed above with respect to
Steps 1002-1007 illustrate a multi-phase multi-nuisance classification method according to some embodiments of the present disclosure. The method may be carried out according to, e.g., the multi-phase multi-nuisance classification 800 of
At step 1002, a plurality of defect review types and a plurality of nuisance review types are developed. In some embodiments, a defect review type may be designed to identify a particular defect type by comparison to a plurality of parameters in a knowledge file for the defect type stored in, e.g., storage 334 of
At step 1003, a machine learning ADC classifier classifies the image data into a plurality of defect types based on the plurality of defect review types and a plurality of nuisance types based on the plurality of nuisance review types. The machine learning classifier may be, e.g., classifier 804 of
At step 1004, the classified image data is compared to a manual review of the image data. The manual review is a more rigorous and labor-intensive analysis to determine with higher confidence the actual nature of each candidate in the set. An operator can then compare the binning of each candidate under ADC to the actual result as determined by review. The comparison may be used to evaluate accuracy of the classification step 1003. For example, the comparison may evaluate accuracy by determining a nuisance rate NR as described above.
At 1005, it is determined whether the most recent phase of the multi-phase multi-nuisance classification is the final phase. The determination may be based on whether the evaluated accuracy, such as a nuisance rate NR, is within prescribed specifications. The determination may be based on whether the evaluated accuracy, such as a nuisance rate NR, has a value within a prescribed proximity to a previous value in a prior classification phase. Alternatively or additionally, the most recent phase may be determined to be a final phase based on a prescribed number of classification phases. If the most recent phase of the multi-phase multi-nuisance classification is determined to be a final phase, then the process moves to step 1006. A selection of misclassified candidates are re-labeled correctly and put into a revised training pool. The revised training pool may include new defect training data or new nuisance training data. The revised training pool may be used for further training in an additional phase beginning at step 1002. If the most recent phase of the multi-phase multi-nuisance classification is determined to be a final phase, the multi-phase classification ends at step 1007. The classified image data may then be used to improve a process, e.g., an integrated circuit or other semiconductor manufacturing process.
A non-transitory computer readable medium may be provided that stores instructions for a processor of controller 109 to carry out charged particle beam inspection, running a classifier network, performing pattern grouping, or other functions and methods consistent with the present disclosure such as method 1000. Common forms of non-transitory media include, for example, a floppy disk, a flexible disk, hard disk, solid state drive, magnetic tape, or any other magnetic data storage medium, a CD-ROM, any other optical data storage medium, any physical medium with patterns of holes, a RAM, a PROM, and EPROM, a FLASH-EPROM or any other flash memory, NVRAM, a cache, a register, any other memory chip or cartridge, and networked versions of the same.
The embodiments may further be described using the following clauses:
The block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer hardware/software products according to various exemplary embodiments of the present disclosure. In this regard, each block in a schematic diagram may represent certain arithmetical or logical operation processing that may be implemented using hardware such as an electronic circuit. Blocks may also represent a module, a segment, or a portion of code that comprises one or more executable instructions for implementing the specified logical functions. It should be understood that in some alternative implementations, functions indicated in a block may occur out of the order noted in the figures. For example, two blocks shown in succession may be executed or implemented substantially concurrently, or two blocks may sometimes be executed in reverse order, depending upon the functionality involved. Some blocks may also be omitted.
It should also be understood that each block of the block diagrams, and combination of the blocks, may be implemented by special purpose hardware-based systems that perform the specified functions or acts, or by combinations of special purpose hardware and computer instructions.
While the present disclosure has been described in connection with various embodiments, other embodiments of the invention will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the invention being indicated by the following claims.
This application claims priority of U.S. application 63/315,054 which was filed on 28 Feb. 2022 and which is incorporated herein in its entirety by reference.
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/EP2023/052663 | 2/3/2023 | WO |
| Number | Date | Country | |
|---|---|---|---|
| 63315054 | Feb 2022 | US |
