COMPUTER ARCHITECTURE FOR THROUGH GLASS VIA DEFECT INSPECTION
CROSS REFERENCE TO RELATED APPLICATIONS
[0001] This application claims the benefit of priority under 35 U.S.C. § 119 of U.S.
Provisional Application No. 63/211,717, filed on June 17, 2021, the content of which is relied upon and incorporated herein by reference in its entirety.
TECHNICAL FIELD
[0002] Embodiments pertain to computer architecture. Some embodiments relate to artificial intelligence. Some embodiments relate to a computer architecture for through glass via defect inspection.
BACKGROUND
[0003] Manufactured through glass vias (TGVs) are oftentimes defective. Techniques for identifying defective TGVs may be desirable.
BRIEF DESCRIPTION OF THE DRAWINGS
[0004] FIG. 1 illustrates the training and use of a machine-learning program, in accordance with some embodiments.
[0005] FIG. 2 illustrates an example neural network, in accordance with some embodiments.
[0006] FIG. 3 illustrates the training of an image recognition machine learning program, in accordance with some embodiments.
[0007] FIG. 4 illustrates the feature-extraction process and classifier training, in accordance with some embodiments.
[0008] FIG. 5 is a block diagram of a computing machine, in accordance with some embodiments.
[0009] FIG. 6 illustrates an example schematic of through glass via (TGV) inspection using backlighting, in accordance with some embodiments.
[0010] FIG. 7 illustrates an example TGV defect inspection system, in accordance with some embodiments.
[0011] FIG. 8 illustrates example TGV characterization via waist, via entrance and exit, and via side view with an optical microscope, in accordance with some embodiments. [0012] FIG. 9 illustrates example computed topography (CT) scan images of TGVs showing non-defective vias and vias with inner defects, in accordance with some embodiments.
[0013] FIGS. 10A-10B illustrates individual via images cropped from raw area scan images, their shapes/inner defects classified, and their locations converted and mapped to part coordinates, in accordance with some embodiments.
[0014] FIG. 11 illustrates defective TGVs having different two-dimensional (2D) signatures depending on the defect type and the optical configuration, in accordance with some embodiments.
[0015] FIG. 12 illustrates example via CT scan image cross sections, in accordance with some embodiments.
[0016] FIG. 13 illustrates an example line-scan image with backlighting including non defective vias and defective vias, in accordance with some embodiments.
[0017] FIG. 14 illustrates example images of different types of vias, in accordance with some embodiments.
[0018] FIG. 15 illustrates example TGV images example TGV images of non-metallized and metallized TGVs under oblique angle, in accordance with some embodiments.
[0019] FIG. 16 illustrates first example via classification maps, in accordance with some embodiments.
[0020] FIG. 17 illustrates second example via classification maps, in accordance with some embodiments.
[0021] FIG. 18 illustrates example area scan image machine learning classification results, in accordance with some embodiments.
[0022] FIG. 19 illustrates an example TGV inspection system, in accordance with some embodiments.
[0023] FIG. 20 illustrates an example relationship between internal defect rate and etching delay, in accordance with some embodiments.
[0024] FIG. 21 illustrates an example machine learning procedure, in accordance with some embodiments.
[0025] FIG. 22 is a block diagram of a computing device for TGV defect inspection, in accordance with some embodiments.
[0026] FIG. 23 is a flow chart of a computer-implemented method for TGV defect inspection, in accordance with some embodiments. DETAILED DESCRIPTION
[0027] The following description and the drawings sufficiently illustrate specific embodiments to enable those skilled in the art to practice them. Other embodiments may incorporate structural, logical, electrical, process, and other changes. Portions and features of some embodiments may be included in, or substituted for, those of other embodiments. Embodiments set forth in the claims encompass all available equivalents of those claims.
[0028] Aspects of the present technology may be implemented as part of a computer system. The computer system may be one physical machine, or may be distributed among multiple physical machines, such as by role or function, or by process thread in the case of a cloud computing distributed model. In various embodiments, aspects of the technology may be configured to run in virtual machines that in turn are executed on one or more physical machines. It will be understood by persons of skill in the art that features of the technology may be realized by a variety of different suitable machine implementations.
[0029] The system includes various engines, each of which is constructed, programmed, configured, or otherwise adapted, to carry out a function or set of functions. The term engine as used herein means a tangible device, component, or arrangement of components implemented using hardware, such as by an application specific integrated circuit (ASIC) or field-programmable gate array (FPGA), for example, or as a combination of hardware and software, such as by a processor-based computing platform and a set of program instructions that transform the computing platform into a special-purpose device to implement the particular functionality. An engine may also be implemented as a combination of the two, with certain functions facilitated by hardware alone, and other functions facilitated by a combination of hardware and software.
[0030] In an example, the software may reside in executable or non-executable form on a tangible machine-readable storage medium. Software residing in non-executable form may be compiled, translated, or otherwise converted to an executable form prior to, or during, runtime. In an example, the software, when executed by the underlying hardware of the engine, causes the hardware to perform the specified operations. Accordingly, an engine is physically constructed, or specifically configured (e.g., hardwired), or temporarily configured (e.g., programmed) to operate in a specified manner or to perform part or all of any operations described herein in connection with that engine.
[0031] Considering examples in which engines are temporarily configured, each of the engines may be instantiated at different moments in time. For example, where the engines comprise a general-purpose hardware processor core configured using software, the general-purpose hardware processor core may be configured as respective different engines at different times. Software may accordingly configure a hardware processor core, for example, to constitute a particular engine at one instance of time and to constitute a different engine at a different instance of time.
[0032] In certain implementations, at least a portion, and in some cases, all, of an engine may be executed on the processor(s) of one or more computers that execute an operating system, system programs, and application programs, while also implementing the engine using multitasking, multithreading, distributed (e.g., cluster, peer-peer, cloud, etc.) processing where appropriate, or other such techniques. Accordingly, each engine may be realized in a variety of suitable configurations, and should generally not be limited to any particular implementation exemplified herein, unless such limitations are expressly called out.
[0033] In addition, an engine may itself be composed of more than one sub-engines, each of which may be regarded as an engine in its own right. Moreover, in the embodiments described herein, each of the various engines corresponds to a defined functionality; however, it should be understood that in other contemplated embodiments, each functionality may be distributed to more than one engine. Likewise, in other contemplated embodiments, multiple defined functionalities may be implemented by a single engine that performs those multiple functions, possibly alongside other functions, or distributed differently among a set of engines than specifically illustrated in the examples herein.
[0034] As used herein, the term “model” encompasses its plain and ordinary meaning. A model may include, among other things, one or more engines which receive an input and compute an output based on the input. The output may be a classification. For example, an image file may be classified as depicting a cat or not depicting a cat. Alternatively, the image file may be assigned a numeric score indicating a likelihood whether the image file depicts the cat, and image files with a score exceeding a threshold (e.g., 0.9 or 0.95) may be determined to depict the cat. [0035] This document may reference a specific number of things (e.g., “six mobile devices”). Unless explicitly set forth otherwise, the numbers provided are examples only and may be replaced with any positive integer, integer or real number, as would make sense for a given situation. For example, “six mobile devices” may, in alternative embodiments, include any positive integer number of mobile devices. Unless otherwise mentioned, an object referred to in singular form (e.g., “a computer” or “the computer”) may include one or multiple objects (e.g., “the computer” may refer to one or multiple computers).
[0036] FIG. 1 illustrates the training and use of a machine-learning program, according to some example embodiments. In some example embodiments, machine-learning programs (MLPs), also referred to as machine-learning algorithms or tools, are utilized to perform operations associated with machine learning tasks, such as image recognition or machine translation.
[0037] Machine learning is a field of study that gives computers the ability to learn without being explicitly programmed. Machine learning explores the study and construction of algorithms, also referred to herein as tools, which may learn from existing data and make predictions about new data. Such machine-learning tools operate by building a model from example training data 112 in order to make data-driven predictions or decisions expressed as outputs or assessments 120. Although example embodiments are presented with respect to a few machine-learning tools, the principles presented herein may be applied to other machine-learning tools.
[0038] In some example embodiments, different machine-learning tools may be used. For example, Logistic Regression (LR), Naive-Bayes, Random Forest (RF), neural networks (NN), matrix factorization, and Support Vector Machines (SVM) tools may be used for classifying or scoring job postings.
[0039] Two common types of problems in machine learning are classification problems and regression problems. Classification problems, also referred to as categorization problems, aim at classifying items into one of several category values (for example, is this object an apple or an orange). Regression algorithms aim at quantifying some items (for example, by providing a value that is a real number). The machine- learning algorithms utilize the training data 112 to find correlations among identified features 102 that affect the outcome.
[0040] The machine-learning algorithms utilize features 102 for analyzing the data to generate assessments 120. A feature 102 is an individual measurable property of a phenomenon being observed. The concept of a feature is related to that of an explanatory variable used in statistical techniques such as linear regression.
Choosing informative, discriminating, and independent features is important for effective operation of the MLP in pattern recognition, classification, and regression. Features may be of different types, such as numeric features, strings, and graphs.
[0041] In one example embodiment, the features 102 may be of different types and may include one or more of words of the message 103, message concepts 104, communication history 105, past user behavior 106, subject of the message 107, other message attributes 108, sender 109, and user data 110.
[0042] The machine-learning algorithms utilize the training data 112 to find correlations among the identified features 102 that affect the outcome or assessment 120. In some example embodiments, the training data 112 includes labeled data, which is known data for one or more identified features 102 and one or more outcomes, such as detecting communication patterns, detecting the meaning of the message, generating a summary of the message, detecting action items in the message, detecting urgency in the message, detecting a relationship of the user to the sender, calculating score attributes, calculating message scores, etc.
[0043] With the training data 112 and the identified features 102, the machine-learning tool is trained at operation 114. The machine-learning tool appraises the value of the features 102 as they correlate to the training data 112. The result of the training is the trained machine-learning program 116.
[0044] When the machine-learning program 116 is used to perform an assessment, new data 118 is provided as an input to the trained machine-learning program 116, and the machine-learning program 116 generates the assessment 120 as output. For example, when a message is checked for an action item, the machine-learning program utilizes the message content and message metadata to determine if there is a request for an action in the message.
[0045] Machine learning techniques train models to accurately make predictions on data fed into the models (e.g., what was said by a user in a given utterance; whether a noun is a person, place, or thing; what the weather will be like tomorrow). During a learning phase, the models are developed against a training dataset of inputs to optimize the models to correctly predict the output for a given input. Generally, the learning phase may be supervised, semi-supervised, or unsupervised; indicating a decreasing level to which the “correct” outputs are provided in correspondence to the training inputs. In a supervised learning phase, all of the outputs are provided to the model and the model is directed to develop a general rule or algorithm that maps the input to the output. In contrast, in an unsupervised learning phase, the desired output is not provided for the inputs so that the model may develop its own rules to discover relationships within the training dataset. In a semi -supervised learning phase, an incompletely labeled training set is provided, with some of the outputs known and some unknown for the training dataset.
[0046] Models may be run against a training dataset for several epochs (e.g., iterations), in which the training dataset is repeatedly fed into the model to refine its results. For example, in a supervised learning phase, a model is developed to predict the output for a given set of inputs, and is evaluated over several epochs to more reliably provide the output that is specified as corresponding to the given input for the greatest number of inputs for the training dataset. In another example, for an unsupervised learning phase, a model is developed to cluster the dataset into n groups, and is evaluated over several epochs as to how consistently it places a given input into a given group and how reliably it produces the n desired clusters across each epoch.
[0047] Once an epoch is run, the models are evaluated and the values of their variables are adjusted to attempt to better refine the model in an iterative fashion. In various aspects, the evaluations are biased against false negatives, biased against false positives, or evenly biased with respect to the overall accuracy of the model. The values may be adjusted in several ways depending on the machine learning technique used. For example, in a genetic or evolutionary algorithm, the values for the models that are most successful in predicting the desired outputs are used to develop values for models to use during the subsequent epoch, which may include random variation/mutation to provide additional data points. One of ordinary skill in the art will be familiar with several other machine learning algorithms that may be applied with the present disclosure, including linear regression, random forests, decision tree learning, neural networks, deep neural networks, etc.
[0048] Each model develops a rule or algorithm over several epochs by varying the values of one or more variables affecting the inputs to more closely map to a desired result, but as the training dataset may be varied, and is preferably very large, perfect accuracy and precision may not be achievable. A number of epochs that make up a learning phase, therefore, may be set as a given number of trials or a fixed time/computing budget, or may be terminated before that number/budget is reached when the accuracy of a given model is high enough or low enough or an accuracy plateau has been reached. For example, if the training phase is designed to run n epochs and produce a model with at least 95% accuracy, and such a model is produced before the nth epoch, the learning phase may end early and use the produced model satisfying the end-goal accuracy threshold. Similarly, if a given model is inaccurate enough to satisfy a random chance threshold (e.g., the model is only 55% accurate in determining true/false outputs for given inputs), the learning phase for that model may be terminated early, although other models in the learning phase may continue training. Similarly, when a given model continues to provide similar accuracy or vacillate in its results across multiple epochs - having reached a performance plateau - the learning phase for the given model may terminate before the epoch number/computing budget is reached.
[0049] Once the learning phase is complete, the models are finalized. In some example embodiments, models that are finalized are evaluated against testing criteria. In a first example, a testing dataset that includes known outputs for its inputs is fed into the finalized models to determine an accuracy of the model in handling data that is has not been trained on. In a second example, a false positive rate or false negative rate may be used to evaluate the models after finalization. In a third example, a delineation between data clusterings is used to select a model that produces the clearest bounds for its clusters of data.
[0050] FIG. 2 illustrates an example neural network 204, in accordance with some embodiments. As shown, the neural network 204 receives, as input, source domain data 202. The input is passed through a plurality of layers 206 to arrive at an output. Each layer 206 includes multiple neurons 208. The neurons 208 receive input from neurons of a previous layer and apply weights to the values received from those neurons in order to generate a neuron output. The neuron outputs from the final layer 206 are combined to generate the output of the neural network 204.
[0051] As illustrated at the bottom of FIG. 2, the input is a vector x. The input is passed through multiple layers 206, where weights W, W, ... , W, are applied to the input to each layer to arrive at f(x),f(x),...,f1(x), until finally the output f(x) is computed.
[0052] In some example embodiments, the neural network 204 (e.g., deep learning, deep convolutional, or recurrent neural network) comprises a series of neurons 208, such as Long Short Term Memory (LSTM) nodes, arranged into a network. A neuron 208 is an architectural element used in data processing and artificial intelligence, particularly machine learning, which includes memory that may determine when to “remember” and when to “forget” values held in that memory based on the weights of inputs provided to the given neuron 208. Each of the neurons 208 used herein are configured to accept a predefined number of inputs from other neurons 208 in the neural network 204 to provide relational and sub -relational outputs for the content of the frames being analyzed. Individual neurons 208 may be chained together and/or organized into tree structures in various configurations of neural networks to provide interactions and relationship learning modeling for how each of the frames in an utterance are related to one another.
[0053] For example, an LSTM node serving as a neuron includes several gates to handle input vectors (e.g., phonemes from an utterance), a memory cell, and an output vector (e.g., contextual representation). The input gate and output gate control the information flowing into and out of the memory cell, respectively, whereas forget gates optionally remove information from the memory cell based on the inputs from linked cells earlier in the neural network. Weights and bias vectors for the various gates are adjusted over the course of a training phase, and once the training phase is complete, those weights and biases are finalized for normal operation. One of skill in the art will appreciate that neurons and neural networks may be constructed programmatically (e.g., via software instructions) or via specialized hardware linking each neuron to form the neural network.
[0054] Neural networks utilize features for analyzing the data to generate assessments (e.g., recognize units of speech). A feature is an individual measurable property of a phenomenon being observed. The concept of feature is related to that of an explanatory variable used in statistical techniques such as linear regression. Further, deep features represent the output of nodes in hidden layers of the deep neural network.
[0055] A neural network, sometimes referred to as an artificial neural network, is a computing system/apparatus based on consideration of biological neural networks of animal brains. Such systems/apparatus progressively improve performance, which is referred to as learning, to perform tasks, typically without task-specific programming. For example, in image recognition, a neural network may be taught to identify images that contain an object by analyzing example images that have been tagged with a name for the object and, having learnt the object and name, may use the analytic results to identify the object in untagged images. A neural network is based on a collection of connected units called neurons, where each connection, called a synapse, between neurons can transmit a unidirectional signal with an activating strength that varies with the strength of the connection. The receiving neuron can activate and propagate a signal to downstream neurons connected to it, typically based on whether the combined incoming signals, which are from potentially many transmitting neurons, are of sufficient strength, where strength is a parameter.
[0056] A deep neural network (DNN) is a stacked neural network, which is composed of multiple layers. The layers are composed of nodes, which are locations where computation occurs, loosely patterned on a neuron in the human brain, which fires when it encounters sufficient stimuli. A node combines input from the data with a set of coefficients, or weights, that either amplify or dampen that input, which assigns significance to inputs for the task the algorithm is trying to learn. These input-weight products are summed, and the sum is passed through what is called a node’s activation function, to determine whether and to what extent that signal progresses further through the network to affect the ultimate outcome. A DNN uses a cascade of many layers of non-linear processing units for feature extraction and transformation. Each successive layer uses the output from the previous layer as input. Higher-level features are derived from lower-level features to form a hierarchical representation. The layers following the input layer may be convolution layers that produce feature maps that are filtering results of the inputs and are used by the next convolution layer.
[0057] In training of a DNN architecture, a regression, which is structured as a set of statistical processes for estimating the relationships among variables, can include a minimization of a cost function. The cost function may be implemented as a function to return a number representing how well the neural network performed in mapping training examples to correct output. In training, if the cost function value is not within a pre-determined range, based on the known training images, backpropagation is used, where backpropagation is a common method of training artificial neural networks that are used with an optimization method such as a stochastic gradient descent (SGD) method.
[0058] Use of backpropagation can include propagation and weight update. When an input is presented to the neural network, it is propagated forward through the neural network, layer by layer, until it reaches the output layer. The output of the neural network is then compared to the desired output, using the cost function, and an error value is calculated for each of the nodes in the output layer. The error values are propagated backwards, starting from the output, until each node has an associated error value which roughly represents its contribution to the original output. Backpropagation can use these error values to calculate the gradient of the cost function with respect to the weights in the neural network. The calculated gradient is fed to the selected optimization method to update the weights to attempt to minimize the cost function.
[0059] FIG. 3 illustrates the training of an image recognition machine learning program, in accordance with some embodiments. The machine learning program may be implemented at one or more computing machines. Block 302 illustrates a training set, which includes multiple classes 304. Each class 304 includes multiple images 306 associated with the class. Each class 304 may correspond to a type of object in the image 306 (e.g., a digit 0-9, a man or a woman, a cat or a dog, etc.). In one example, the machine learning program is trained to recognize images of the presidents of the United States, and each class corresponds to each president (e.g., one class corresponds to Barack Obama, one class corresponds to George W. Bush, one class corresponds to Bill Clinton, etc.). At block 308 the machine learning program is trained, for example, using a deep neural network. At block 310, the trained classifier, generated by the training of block 308, recognizes an image 312, and at block 314 the image is recognized. For example, if the image 312 is a photograph of Bill Clinton, the classifier recognizes the image as corresponding to Bill Clinton at block 314.
[0060] FIG. 3 illustrates the training of a classifier, according to some example embodiments. A machine learning algorithm is designed for recognizing faces, and a training set 302 includes data that maps a sample to a class 304 (e.g., a class includes all the images of purses). The classes may also be referred to as labels. Although embodiments presented herein are presented with reference to object recognition, the same principles may be applied to train machine-learning programs used for recognizing any type of items.
[0061] The training set 302 includes a plurality of images 306 for each class 304 (e.g., image 306), and each image is associated with one of the categories to be recognized (e.g., a class). The machine learning program is trained 308 with the training data to generate a classifier 310 operable to recognize images. In some example embodiments, the machine learning program is a DNN.
[0062] When an input image 312 is to be recognized, the classifier 310 analyzes the input image 312 to identify the class (e.g., class 314) corresponding to the input image 312.
[0063] FIG. 4 illustrates the feature-extraction process and classifier training, according to some example embodiments. Training the classifier may be divided into feature extraction layers 402 and classifier layer 414. Each image is analyzed in sequence by a plurality of layers 406-413 in the feature-extraction layers 402.
[0064] With the development of deep convolutional neural networks, the focus in face recognition has been to learn a good face feature space, in which faces of the same person are close to each other, and faces of different persons are far away from each other. For example, the verification task with the LFW (Labeled Faces in the Wild) dataset has been often used for face verification.
[0065] Many face identification tasks (e.g., MegaFace and LFW) are based on a similarity comparison between the images in the gallery set and the query set, which is essentially a K-nearest-neighborhood (KNN) method to estimate the person’s identity. In the ideal case, there is a good face feature extractor (inter-class distance is always larger than the intra-class distance), and the KNN method is adequate to estimate the person’s identity.
[0066] Feature extraction is a process to reduce the amount of resources required to describe a large set of data. When performing analysis of complex data, one of the major problems stems from the number of variables involved. Analysis with a large number of variables generally requires a large amount of memory and computational power, and it may cause a classification algorithm to overfit to training samples and generalize poorly to new samples. Feature extraction is a general term describing methods of constructing combinations of variables to get around these large data-set problems while still describing the data with sufficient accuracy for the desired purpose.
[0067] In some example embodiments, feature extraction starts from an initial set of measured data and builds derived values (features) intended to be informative and non-redundant, facilitating the subsequent learning and generalization steps. Further, feature extraction is related to dimensionality reduction, such as be reducing large vectors (sometimes with very sparse data) to smaller vectors capturing the same, or similar, amount of information.
[0068] Determining a subset of the initial features is called feature selection. The selected features are expected to contain the relevant information from the input data, so that the desired task can be performed by using this reduced representation instead of the complete initial data. DNN utilizes a stack of layers, where each layer performs a function. For example, the layer could be a convolution, a non-linear transform, the calculation of an average, etc. Eventually this DNN produces outputs by classifier 414. In FIG. 4, the data travels from left to right and the features are extracted. The goal of training the neural network is to find the parameters of all the layers that make them adequate for the desired task.
[0069] As shown in FIG. 4, a “stride of 4” filter is applied at layer 406, and max pooling is applied at layers 407-413. The stride controls how the filter convolves around the input volume. “Stride of 4” refers to the filter convolving around the input volume four units at a time. Max pooling refers to down-sampling by selecting the maximum value in each max pooled region.
[0070] In some example embodiments, the structure of each layer is predefined. For example, a convolution layer may contain small convolution kernels and their respective convolution parameters, and a summation layer may calculate the sum, or the weighted sum, of two pixels of the input image. Training assists in defining the weight coefficients for the summation.
[0071] One way to improve the performance of DNNs is to identify newer structures for the feature-extraction layers, and another way is by improving the way the parameters are identified at the different layers for accomplishing a desired task. The challenge is that for a typical neural network, there may be millions of parameters to be optimized. Trying to optimize all these parameters from scratch may take hours, days, or even weeks, depending on the amount of computing resources available and the amount of data in the training set.
[0072] FIG. 5 illustrates a circuit block diagram of a computing machine 500 in accordance with some embodiments. In some embodiments, components of the computing machine 500 may store or be integrated into other components shown in the circuit block diagram of FIG. 5 For example, portions of the computing machine 500 may reside in the processor 502 and may be referred to as “processing circuitry.” Processing circuitry may include processing hardware, for example, one or more central processing units (CPUs), one or more graphics processing units (GPUs), and the like. In alternative embodiments, the computing machine 500 may operate as a standalone device or may be connected (e.g., networked) to other computers. In a networked deployment, the computing machine 500 may operate in the capacity of a server, a client, or both in server-client network environments. In an example, the computing machine 500 may act as a peer machine in peer-to-peer (P2P) (or other distributed) network environment. In this document, the phrases P2P, device-to- device (D2D) and sidelink may be used interchangeably. The computing machine 500 may be a specialized computer, a personal computer (PC), a tablet PC, a personal digital assistant (PDA), a mobile telephone, a smart phone, a web appliance, a network router, switch or bridge, or any machine capable of executing instructions (sequential or otherwise) that specify actions to be taken by that machine.
[0073] Examples, as described herein, may include, or may operate on, logic or a number of components, modules, or mechanisms. Modules and components are tangible entities (e.g., hardware) capable of performing specified operations and may be configured or arranged in a certain manner. In an example, circuits may be arranged (e.g., internally or with respect to external entities such as other circuits) in a specified manner as a module. In an example, the whole or part of one or more computer systems/apparatus (e.g., a standalone, client or server computer system) or one or more hardware processors may be configured by firmware or software (e.g., instructions, an application portion, or an application) as a module that operates to perform specified operations. In an example, the software may reside on a machine readable medium. In an example, the software, when executed by the underlying hardware of the module, causes the hardware to perform the specified operations.
[0074] Accordingly, the term “module” (and “component”) is understood to encompass a tangible entity, be that an entity that is physically constructed, specifically configured (e.g., hardwired), or temporarily (e.g., transitorily) configured (e.g., programmed) to operate in a specified manner or to perform part or all of any operation described herein. Considering examples in which modules are temporarily configured, each of the modules need not be instantiated at any one moment in time. For example, where the modules comprise a general-purpose hardware processor configured using software, the general-purpose hardware processor may be configured as respective different modules at different times. Software may accordingly configure a hardware processor, for example, to constitute a particular module at one instance of time and to constitute a different module at a different instance of time.
[0075] The computing machine 500 may include a hardware processor 502 (e.g., a central processing unit (CPU), a GPU, a hardware processor core, or any combination thereof), a main memory 504 and a static memory 506, some or all of which may communicate with each other via an interlink (e.g., bus) 508. Although not shown, the main memory 504 may contain any or all of removable storage and non removable storage, volatile memory or non-volatile memory. The computing machine 500 may further include a video display unit 510 (or other display unit), an alphanumeric input device 512 (e.g., a keyboard), and a user interface (UI) navigation device 514 (e.g., a mouse). In an example, the display unit 510, input device 512 and UI navigation device 514 may be a touch screen display. The computing machine 500 may additionally include a storage device (e.g., drive unit) 516, a signal generation device 518 (e.g., a speaker), a network interface device 520, and one or more sensors 521, such as a global positioning system (GPS) sensor, compass, accelerometer, or other sensor. The computing machine 500 may include an output controller 528, such as a serial (e.g., universal serial bus (USB), parallel, or other wired or wireless (e.g., infrared (IR), near field communication (NFC), etc.) connection to communicate or control one or more peripheral devices (e.g., a printer, card reader, etc.).
[0076] The drive unit 516 (e.g., a storage device) may include a machine readable medium 522 on which is stored one or more sets of data structures or instructions 524 (e.g., software) embodying or utilized by any one or more of the techniques or functions described herein. The instructions 524 may also reside, completely or at least partially, within the main memory 504, within static memory 506, or within the hardware processor 502 during execution thereof by the computing machine 500. In an example, one or any combination of the hardware processor 502, the main memory 504, the static memory 506, or the storage device 516 may constitute machine readable media.
[0077] While the machine readable medium 522 is illustrated as a single medium, the term “machine readable medium” may include a single medium or multiple media (e.g., a centralized or distributed database, and/or associated caches and servers) configured to store the one or more instructions 524. [0078] The term “machine readable medium” may include any medium that is capable of storing, encoding, or carrying instructions for execution by the computing machine 500 and that cause the computing machine 500 to perform any one or more of the techniques of the present disclosure, or that is capable of storing, encoding or carrying data structures used by or associated with such instructions. Non-limiting machine readable medium examples may include solid-state memories, and optical and magnetic media. Specific examples of machine readable media may include: non-volatile memory, such as semiconductor memory devices (e.g., Electrically Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM)) and flash memory devices; magnetic disks, such as internal hard disks and removable disks; magneto-optical disks; Random Access Memory (RAM); and CD-ROM and DVD-ROM disks. In some examples, machine readable media may include non-transitory machine readable media. In some examples, machine readable media may include machine readable media that is not a transitory propagating signal.
[0079] The instructions 524 may further be transmitted or received over a communications network 526 using a transmission medium via the network interface device 520 utilizing any one of a number of transfer protocols (e.g., frame relay, internet protocol (IP), transmission control protocol (TCP), user datagram protocol (UDP), hypertext transfer protocol (HTTP), etc.). Example communication networks may include a local area network (LAN), a wide area network (WAN), a packet data network (e.g., the Internet), mobile telephone networks (e.g., cellular networks), Plain Old Telephone (POTS) networks, and wireless data networks (e.g., Institute of Electrical and Electronics Engineers (IEEE) 802.11 family of standards known as Wi-Fi®, IEEE 802.16 family of standards known as WiMax®), IEEE 802.15.4 family of standards, a Long Term Evolution (LTE) family of standards, a Universal Mobile Telecommunications System (UMTS) family of standards, peer-to-peer (P2P) networks, among others. In an example, the network interface device 520 may include one or more physical jacks (e.g., Ethernet, coaxial, or phone jacks) or one or more antennas to connect to the communications network 526.
[0080] Substrates, such as silicon, have been used as an interposer disposed between electrical components (e.g., printed circuit boards, integrated circuits, and the like). Metalized Through Glass Vias (TGVs) provide a path through the interposer for electrical signals to pass between opposite sides of the interposer. Glass is a substrate material that is highly advantageous for electrical signal transmission, as it has dimensional stability, a tunable coefficient of thermal expansion (CTE), low electrical loss at high frequencies, high thermal stability, and an ability to be formed at various thicknesses and at large panel sizes. TGV production technology may utilize a laser damage and etching (LD&E) process to make micro-sized holes in thin glass substrates for a variety of applications in the semiconductor and display industries. It may use a pulsed laser beam to create damage tracks in the glass using a series of picosecond bursts. The laser damage tracks are then etched in an acid bath so that through holes can be formed to achieve a specific surface diameter (e.g., between 10 and 200 pm) and a desired internal geometry (straight, tapered, or hourglass shape with a target waist diameter). The three dimensional (3D) via shape is critical for process development, product monitor and control. TGVs with inner defects cannot only affect the final product electrical performance, but also subsequent processes such as metallization. While our large-scale LD&E process enables high-speed TGV formation, some major challenges remain for inspection and characterization of final etched via geometries such as 3D shape defects (bulges, chatter marks, blind vias etc.).
[0081] Some schemes used for characterizing TGV geometry include vision-based surface diameter or waist diameter measurements, 3D computed topography (CT) scans, or destructive tests, which may leverage dicing of the substrate to enable side-view optical microscopy. TGVs with internal via sidewall defects can still have entrance, exit, or waist diameters that are within specification. In some cases, via side wall defects might be identified in CT scans or through destructive tests, which are time consuming, expensive, and can only be done on a small number of samples. Moreover, it might be difficult to quantify characteristics of via side wall defects using these conventional techniques. Accordingly, techniques for inspecting TGVs for internal shape and side wall defects are desirable.
[0082] Some embodiments relate to a non-destructive method that combines machine vision with machine learning to inspect TGV samples for internal via shape and side wall defects. The method, which also characterizes and maps the defects, can be easily scaled up for high speed inline inspection for production of large TGV panels.
[0083] In some embodiments, a computing device coupled with a camera takes images of individual vias from an oblique inspection angle with proper illumination. Some embodiments use an optimized illumination angle (incident from the back or the top, for example), a preferential polarization orientation, and a uniform illumination to improve the contrast and sensitivity of the TGV images. Traditional vision inspection methods include examination with a microscope, either from the top or bottom of the vias, or from the side after dicing of the sample, both of which provide limited information of the vias’ internal shape. Some embodiments can inspect all TGVs inside a substrate by detecting weak light deflection signatures of the TGV shape defects and comparing the detected signatures to the image of a regularly shaped via. Individual TGV images are extracted from raw scan images together with their corresponding via locations, which are converted to XY part coordinates (or other coordinates) and matched up with the nominal via pattern. Some embodiments include a fully automated via characterization. Supervised Deep Learning (DL) / Machine Learning (ML) techniques may be employed to characterize vias by their appearance in the images. In some cases, a set of via images are categorized by users as “good” or “defective” and then used to train a Machine-Learning (ML) algorithm. The trained ML model enables automatic TGV classification. Some embodiments use online learning, providing for a re categorization after classification. Some embodiments aim to get a feedback of image labels after getting information of classification categories and scores. This enables users to make more insightful decisions on via images of low margin scores, new cases or outliers. Furthermore, other measurement data can also be utilized in via ML characterization. A global coordinate transfer and matching algorithm may be used to combine other measurement results, such as the TGV’s top, bottom, or waist diameter corresponding to the particular via. This enables the understanding of correlations between different measurements as well as improving the classification accuracy. Some embodiments also work for inspection of metalized TGVs. Finally, line scan imaging can be used to apply this method to high speed inline applications.
[0084] Some embodiments relate to a non-destructive high-speed method that can inspect millions of TGVs in a flat substrate. Some embodiments enable quantitative interpretation of different internal via defects in metalized or unmetallized TGVs in different transparent substrate such as glass, fused silica, and the like. Some embodiments provide a method to optimize, calibrate and monitor the LD&E and TGV metallization processes. By being non-destructive and fast, some embodiments provide significant cost and time savings for TGV production. Some embodiments relate to a novel data fusion and machine learning technique for rapidly obtaining correlations between measurements and LD&E parameters that provide a fundamental understanding of which aspects of the LD&E interaction influence TGV metrics.
[0085] FIG. 6 illustrates an example schematic 600 of TGV inspection using backlighting, in accordance with some embodiments. As shown in the schematic 600, a camera 602 is pointed at an object 604. The object 604 includes a TGV 606 and a chip light- emitting diode (LED) 608. The object 604 is illuminated by diffusion back light 610.
[0086] FIG. 7 illustrates an example TGV defect inspection system 700, in accordance with some embodiments. As shown, the TGV defect inspection system 700 includes an area camera 702 with a rotary unit 704 for rotation. Light power 706 powers a back light illuminator 708 of a surface configured for X/Y motion 710. The area camera 702 points at the surface illuminated by the back light illuminator 708.
[0087] A TGV process may use a single picosecond laser burst to form a laser damage track in the glass substrate. The laser beam is formed into an extended focus, or quasi- non-diffracting beam, using optics that form a Bessel-like or Bessel-Gauss beam. Due to the quasi-non-diffracting nature of the beam, the light maintains a tight focused intensity over a much longer range than is achieved with more commonly used Gaussian beams, allowing the full thickness of the glass substrate to be damaged by a single burst pulse. The damage track might have a diameter of 300- 500 nm and its length may extend through the whole thickness of the glass (50 to 1000 pm). The laser-damaged glass may be etched in an etchant bath containing either hydrofluoric acid (HF) or Sodium/Potassium hydroxide to form the TGV. The laser and etch process ids disclosed, for example, in US Patent No. 9,517,963, the entire content of which is incorporated herein by reference.
[0088] FIG. 8 illustrates example TGV characterization images 800 via waist 802, via entrance and exit 804, and via side view 806. The images 800 may be generated with an optical microscope, in accordance with some embodiments.
[0089] One conventional characterization scheme for TGVs includes entrance, exit, and waist diameters measured with 2D vision inspection tools as shown in FIG. 8 at 802 and 804, and profile view imaging on diced samples using an optical microscope as shown in FIG. 8 at 806. TGV diameter measurements cannot detect internal defects in the via shape as shown in microscope’s profile view. However, the optical microscope’s side view inspection may be a destructive method and cannot satisfy inline measurement needs. Another conventional measurement tool is the 3D CT scanner, which provides internal details of the TGVs with unparalleled resolution.
[0090] FIG. 9 illustrates example CT scan images of TGVs showing non-defective vias and vias with inner defects, in accordance with some embodiments. However, CT scans are slow, very expensive, and hard to scale up for inline applications.
[0091] FIGS. 10A-10B illustrates individual via images cropped from raw area scan images, their shapes/inner defects classified, and their locations converted and mapped to part coordinates, in accordance with some embodiments.
[0092] FIG. 11 illustrates defective TGVs having different two-dimensional (2D) signatures depending on the defect type and the optical configuration, in accordance with some embodiments.
[0093] FIG. 12 illustrates example via CT scan image cross sections 1200, in accordance with some embodiments. The cross sections 1200 include chatter marks 1202 and bulges 1204. The 2D images 1200 are acquired at an oblique angle and using line- scan imaging.
[0094] FIG. 13 illustrates an example line-scan image 1300 with backlighting including non-defective vias 1302 and defective vias 1304 and 1306, in accordance with some embodiments.
[0095] FIG. 14 illustrates example images 1400 of different types of vias, in accordance with some embodiments. The images 1400 include images of a narrow waist via 1402, a blind via 1404, a bulged via 1406, and a good (non-defective) via 1408.
[0096] FIG. 15 illustrates example TGV images 1500 of non-metallized 1502 and metallized 1504 TGVs, in accordance with some embodiments. The images 1500 are acquired using a line-scan camera and back illumination.
[0097] FIG. 16 illustrates first example via classification maps 1600 measured before 1602 and after 1604 ninety degree rotation of the part, in accordance with some embodiments.
[0098] FIG. 17 illustrates second example via classification maps 1700 as measured from side A 1702 and side B 1704, in accordance with some embodiments.
[0099] FIG. 18 illustrates example area scan image machine learning classification results, in accordance with some embodiments.
[0100] FIG. 19 illustrates an example TGV inspection system 1900, in accordance with some embodiments. The system 1900 allows for X motion 1902 and Y motion 1904. The system 1900 includes a line-scan time delayed integration (TDI) camera 1906, a LED backlight carrier 1908, and a rotary stage 1910.
[0101] FIG. 20 illustrates an example relationship between internal defect rate and etching delay, in accordance with some embodiments. As shown in FIG. 20, with the same laser damage conditions, the internal defect rate is related to the etching delay.
[0102] FIG. 21 illustrates an example machine learning procedure 2100, in accordance with some embodiments. As shown, labeling tool 2102 and training dataset 2108 provide data for machine learning (ML) classification 2104. The ML classification 2104 provides output to a re-labeling and selection tool 2106, which confirms or edits the classification categories and scores generated by the ML classification 2104. The confirmed or edited classification categories and scores are provided to the training dataset 2108 via data upload from the re-labeling and selection tool 2106.
[0103] Some embodiments relate to a high-speed, non-destructive, in-line capable, quantitative method for the detection and classification of internal via defects in transparent substrates. Some embodiments include an inspection apparatus, either in standalone form or integrated into other TGV inspection systems. FIG. 19 shows a design concept of a standalone line-scan based TGV internal defect measurement system 1900 with backlight illumination built into the sample carrier (e.g., LED backlight carrier 1908). The sample carrier is mounted on a rotary stage and linear axes. The camera 1906 is mounted on another linear axis which is orthogonal to the carrier’s linear axis. The motion system (e.g., X motion 1902 and Y motion 1904) described is capable of inspecting via from different angles inside the sample plane as needed based on the via pattern design.
[0104] Some techniques include taking a scattering image of each individual TGV inside a transparent substrate, such as glass, at an oblique angle with the imaging apparatus as illustrated in FIG. 6 or FIG. 7. The illumination sources generate light, which illuminates the vias at specific angles. These include but are not limited to top/back light, edge light, or low angle dark field illumination. The source can be high intensity LED lights with a chosen spectrum or white light. Some embodiments utilize optimized illumination to examine metalized or unmetallized vias for best contrast. For example, a diffused top / back light reduces reflection from the TGVs’ inside surfaces and increases image contrast.
[0105] In some cases, the camera is the detector that collects the signals from the vias. To detect very weak signals deflected from the vias’ internal defects such as bulges or chatter marks, a TDI line-scan camera can be used. To improve the speed of the inspection, the TDI camera can operate at high frame rates. The camera angle must be optimized and is typically between 40 and 50 degrees to the substrate surface.
The minimum pitch between adjacent vias and the camera angle have to be optimized to prevent vias from overlapping in the image. With TDI line-scan imaging all vias will be in focus compared to area-scan cameras, which might make image processing less complicated and faster.
[0106] In terms of optics, the camera objective may have a specific magnification, numerical aperture (NA), and depth of field, and preferably be telecentric to capture the top and bottom of the via in focus. In some examples, a lens with a magnification less than 3x and an NA less than 0.1 provides a large enough depth of field to image through the glass of 500 pm thickness. The magnification may be, for example, 1.5x - 2.5x.
[0107] Some embodiments include a combination of software and hardware to normalize the image pixel response and noise from optical aberrations and electrical sources.
[0108] Some embodiments include extracting, at a computing device, individual via images and correlating the shape attributes with other measurement data such as diameter, roundness, LD&E process information and other available operator classification information. Some embodiments use pattern match or other image correlation algorithms to find individual TGVs, as shown in FIGS. 10A-10B. Some embodiments leverage one-to-one via image correlating with other measurement results or process parameters based on global via patterns defined by the product design.
[0109] Some embodiments include creating, at a computing device, a machine learning model. Via images as well as the other measurement data can be used in supervised machine learning to train the ML algorithm to classify via types automatically. Images from different types of via shapes will be manually sorted and categorized for training purposes first; for example, classified as “good (non-defective),” “bulge,” or “blind,” as shown in FIG. 13 and FIG. 14.
[0110] Machine learning / deep learning algorithms may enable automatic via classification without prior domain knowledge of the via features. The conventional and common ML/DL approach is a supervised learning model that utilizes labeled data to train an ML/DL model. The ML expert designs a model to get a better performance based on the training data. However, some embodiments are directed to obtaining a better training data set with a better ML model, as shown in FIG. 21.
[0111] Cropped via images are collected and labeled (into the training dataset 2108) to train the ML model. Via images after the training procedure might not need to be labeled as shown in the labeling tool 2102. Via images are classified by the trained ML model at ML classification 2104. After the ML classification 2104, users get informed of via defect conditions and classification scores at the re-labeling and selection tool 2106. Users can relabel the via images if the image has strong inference score from ML model and the significant via image data would be stored at the training dataset. Users can sort via images with low margin scores to check the labels and decide how to relabel and whether it needs to be stored at the training dataset. Some embodiments enhance integrity of training data set so that the ML model can outperform and be gradually improved in future training session(s).
[0112] Using the proposed inspection system and ML algorithm, certain internal via defects are linked to specific conditions in the LD&E process. Most vias with internal defects have normal surface attributes like entrance and exit diameters and might not have been identified through surface measurements alone, neither from top nor bottom. FIG. 17 shows an example of a classification map 1700 as measured from the A- and B-side (1702 and 1704, respectively). Some via defects such as bulges and chatter marks have good in plane symmetry. Furthermore, defect maps for inspections from different viewing angles (i.e. after sample rotation) also align well. FIG. 16 shows two via classification maps 1602 and 1604, one measured at zero degrees (1602) and one after a ninety-degree sample rotation (1604). FIG. 20 shows that for the same laser processing but different etching conditions (e.g., a different etching delay) the percentage of internal via defects can be vastly different. The disclosed technique may also be used to inspect vias after metallization. FIG. 15 shows via images 1500 before metallization 1502 and after metallization 1504 using a line-scan and backlight configuration.
[0113] FIG. 22 is a block diagram of a computing device 2200 for TGV defect inspection, in accordance with some embodiments. The computing device 2200 may include one or more of: a server, a laptop computer, a desktop computer, a mobile phone, a tablet computer, a smart watch, a personal digital assistant (PDA) and the like. In addition to the components shown in FIG. 22, the computing device 2200 may include one or more components of the computing machine 500. Further, while a single computing device 2200 is illustrated, it should be noted that, in some embodiments, the components of the computing device 2200 may be distributed across multiple physical or virtual machines. As shown, the computing device 2200 includes processing circuitry 2210 (e.g., one or more of processor 502, a central processing unit, a graphics processing unit, and the like) and memory 2220 (e.g., one or more of main memory 504, static memory 506, drive unit 516, a disk drive, a hard drive, a random access memory, and the like). The memory 2220 stores an image generation engine 2230, a computer vision engine 2240, and a classification engine 2250. Example operations of the image generation engine 2230, the computer vision engine 2240, and the classification engine 2250 are discussed in conjunction with FIG. 23. The engines 2230, 2240, and 2250 may be implemented using artificial intelligence, machine learning, supervised learning, online learning, and/or deep learning techniques, for example, as described in conjunction with FIGS. 1-4.
[0114] FIG. 23 is a flow chart of a computer-implemented method 2300 for TGV defect inspection, in accordance with some embodiments. The method 2300 may be implemented at a computing device (e.g., computing device 2200) or multiple computing devices, which may include one or more components of the computing machine 500.
[0115] At block 2310, the computing device controls an illumination source (e.g., LED backlight carrier 1908) and an imaging device (e.g., camera 1906) to generate images of TGVs, wherein the TGVs comprise etched or metallized micro vias in a transparent workpiece comprising at least two surfaces. Controlling the illumination source and the imaging device comprises, for multiple TGVs, directing the illumination source to shine light onto the TGVs in the workpiece; and directing the imaging device to detect a scattering image signal from light scatter by the TGVs.
An imaging axis of the imaging device extends a non-zero imaging angle relative to an axis of a given TGV from the multiple TGVs. The entirety of the given TGV is within a field of view of the imaging device.
[0116] At block 2320, the computing device generates, using an image generation engine (e.g., image generation engine 2230) at the computing device, images of the TGVs based on the scattering image signals detected by the imaging device.
[0117] At block 2330, the computing device computes, using a computer vision engine (e.g., computer vision engine 2240) at the computing device, coordinates of the TGVs within the images based on the scattering image signals detected by the imaging device.
[0118] At block 2340, the computing device generates for one or more TGVs in the images and based on the computed coordinates, a TGV type classification using a classification engine (e.g., classification engine 2250) at the computing device, wherein the TGV type classification indicated whether the TGV has or lacks a defect and, if the TGV has a defect, a defect type.
[0119] At block 2350, the computing device determines, based on the generated TGV type classifications, that multiple TGVs have a defect associated with a given defect type.
[0120] At block 2360, the computing device identifies, a manufacturing machine setting associated with the defect type.
[0121] At block 2370, the computing device transmits (e.g., to a manufacturing machine) a control signal to adjust the manufacturing machine setting.
[0122] According to some embodiments, the imaging device comprises a camera and an imaging lens disposed on the imaging axis, the camera comprises an area scan camera or a line scan camera, and the non-zero imaging angle is between thirty degrees and sixty degrees.
[0123] According to some embodiments, the imaging lens comprises a magnification factor and a numerical aperture that are dependent on a thickness of the transparent workpiece.
[0124] According to some embodiments, the thickness of the transparent workpiece is between one and one-thousand micrometers, wherein the magnification factor is less than three.
[0125] According to some embodiments, the control signal is transmitted directly to a manufacturing machine for adjusting the manufacturing machine setting thereat.
[0126] According to some embodiments, the computing device accesses additional images of TGVs manufactured after the manufacturing machine setting is adjusted. The computing device determines whether the defect associated with the given defect type is changing in frequency. The computing device transmits, the manufacturing machine, additional control signals to further adjust the manufacturing machine setting based on a change in frequency of the given defect type.
[0127] According to some embodiments, the control signal is transmitted to a user device to instruct an end user to adjust the manufacturing machine setting. [0128] According to some embodiments, the computing device transmits the generated
TGV type classifications to a user device. The computing device receives, from the user device, an indication that a least one TGV type classification is incorrect. The computing device provides further training to the TGV type classifier based on the indication.
[0129] According to some embodiments, the generated images have multiple different image types comprising one or more of: RGB images, black and white images, shadow images, and CT scan images.
[0130] According to some embodiments, the classification engine comprises an artificial neural network (ANN) trained using online supervised learning.
[0131] According to some embodiments, when via depth is larger than the imaging systems depth of focus, multiple camera focus at incremental depth of the via or a multiple inspection cycle at incremental depth can be implemented.
[0132] According to some embodiments, the computing machine directs the light from the illumination source onto the plurality of vias by directing the light from the illumination source through an edge of the transparent workpiece.
[0133] According to some embodiments, the illumination NA could vary from telecentric illumination to angle controlled diffusion surface depending on final via inner shape/geometrics - cylindrical, tape, and the like.
[0134] According to some embodiments, the computing device determines the characteristic of the laser processing system by determining a focusing position of the defect forming laser processing based on via image and class.
[0135] According to some embodiments, the computing device determines the characteristic of the laser processing system by determining a laser power of the defect forming laser processing based on via image and class.
[0136] According to some embodiments, the computing device determines the characteristic of etching system by determining an etching delay before introducing the ultrasonic enhanced etching based on via image and class.
[0137] According to some embodiments, the computing device correlates or fuses other measurement results including one or more of: measurement of via entrance/exit and waist, laser damage and etching processing data, via metallization processing data to improve ANN classification accuracy.
[0138] Some embodiments include a machine coordinate matching algorithm to correlate individual via measurements on different inspection systems. [0139] According to some embodiments, the ANN reduces the impact from optical imaging distortion and cost of optical component and complicate calibration.
[0140] In some embodiments, the computing device performs active learning on a training dataset that includes the labeled dataset by gathering a set of unlabeled data and user (e.g., expert) feedback. The computing device provides a confidence score for each possible category and a marginal score that is the difference between the highest score and the second highest score. The computing device leverages the user feedback to obtain the true labels. In addition to that, the computing device provides other information on via quality that includes the diameter / roundness of inlet and outlet of via shape, waist diameter and circularity, which provides supplementary information to support the user’s decision. The low marginal scored data is provided to users periodically with supplementary information that includes the diameter / roundness of inlet and outlet of via shape, waist diameter and circularity to determine higher probable labels to the unlabeled data. The newly labeled data items are used to train the classification engine (e.g., classification engine 2250) to conform to the user’s knowledge and quality control level.
[0141] In some embodiments, the illumination setup directs the illumination source to shine the light onto a surface of the transparent workpiece at a specific illumination angle. The illumination angle is different from the angle of the imaging axis.
[0142] As described above, the blocks of the method 2300 are performed serially in the given order. However, these blocks may be preformed in any order, not necessarily the order provided. In some cases, two or more blocks may be performed in parallel.
[0143] Some embodiments are described as numbered examples (Example 1, 2, 3, etc.). These are provided as examples only and do not limit the technology disclosed herein.
[0144] Example l is a method comprising: controlling, by a computing device, an illumination source and an imaging device to generate images of through glass vias (TGVs), wherein the TGVs comprise etched or metallized micro vias in a transparent workpiece comprising at least two surfaces, wherein controlling the illumination source and the imaging device comprises, for multiple TGVs: directing the illumination source to shine light onto the TGVs in the workpiece; and directing the imaging device to detect a scattering image signal from light scatter by the TGVs, wherein an imaging axis of the imaging device extends a non-zero imaging angle relative to an axis of a given TGV from the multiple TGVs, wherein the entirety of the given TGV is within a field of view of the imaging device; generating, using an image generation engine at the computing device, images of the TGVs based on the scattering image signals detected by the imaging device; computing, using a computer vision engine at the computing device, coordinates of the TGVs within the images based on the scattering image signals detected by the imaging device; generating, for one or more TGVs in the images and based on the computed coordinates, a TGV type classification using a classification engine at the computing device, wherein the TGV type classification indicated whether the TGV has or lacks a defect and, if the TGV has a defect, a defect type; determining, based on the generated TGV type classifications, that multiple TGVs have a defect associated with a given defect type; identifying, at the computing device, a manufacturing machine setting associated with the given defect type; and transmitting, from the computing device, a control signal to adjust the manufacturing machine setting.
[0145] In Example 2, the subject matter of Example 1 includes, wherein the imaging device comprises a camera and an imaging lens disposed on the imaging axis, wherein the camera comprises an area scan camera or a line scan camera, wherein the non-zero imaging angle is between thirty degrees and sixty degrees.
[0146] In Example 3, the subject matter of Example 2 includes, wherein the imaging lens comprises a magnification factor and a numerical aperture that are dependent on a thickness of the transparent workpiece.
[0147] In Example 4, the subject matter of Example 3 includes, wherein the thickness of the transparent workpiece is between one and one-thousand micrometers, wherein the magnification factor is less than three.
[0148] In Example 5, the subject matter of Examples 1-4 includes, wherein the control signal is transmitted directly to a manufacturing machine for adjusting the manufacturing machine setting thereat.
[0149] In Example 6, the subject matter of Example 5 includes, accessing additional images of TGVs manufactured after the manufacturing machine setting is adjusted; determining whether the defect associated with the given defect type is changing in frequency; and transmitting, from the computing device to the manufacturing machine, additional control signals to further adjust the manufacturing machine setting based on a change in frequency of the given defect type. [0150] In Example 7, the subject matter of Examples 1-6 includes, wherein the control signal is transmitted to a user device to instruct an end user to adjust the manufacturing machine setting.
[0151] In Example 8, the subject matter of Examples 1-7 includes, transmitting the generated TGV type classifications to a user device; receiving, from the user device, an indication that a least one TGV type classification is incorrect; and providing further training to the TGV type classifier based on the indication.
[0152] In Example 9, the subject matter of Examples 1-8 includes, wherein the generated images have multiple different image types comprising one or more of: RGB images, black and white images, shadow images, and CT scan images.
[0153] In Example 10, the subject matter of Examples 1-9 includes, wherein the classification engine comprises an artificial neural network (ANN) trained using online supervised learning.
[0154] Example 11 is at least one machine-readable medium including instructions that, when executed by processing circuitry, cause the processing circuitry to perform operations to implement of any of Examples 1-10.
[0155] Example 12 is an apparatus comprising means to implement of any of Examples 1- 10
[0156] Example 13 is a system to implement of any of Examples 1-10.
[0157] Example 14 is a method to implement of any of Examples 1-10.
[0158] Although an embodiment has been described with reference to specific example embodiments, it will be evident that various modifications and changes may be made to these embodiments without departing from the broader spirit and scope of the present disclosure. Accordingly, the specification and drawings are to be regarded in an illustrative rather than a restrictive sense. The accompanying drawings that form a part hereof show, by way of illustration, and not of limitation, specific embodiments in which the subject matter may be practiced. The embodiments illustrated are described in sufficient detail to enable those skilled in the art to practice the teachings disclosed herein. Other embodiments may be utilized and derived therefrom, such that structural and logical substitutions and changes may be made without departing from the scope of this disclosure. This Detailed Description, therefore, is not to be taken in a limiting sense, and the scope of various embodiments is defined only by the appended claims, along with the full range of equivalents to which such claims are entitled. [0159] Although specific embodiments have been illustrated and described herein, it should be appreciated that any arrangement calculated to achieve the same purpose may be substituted for the specific embodiments shown. This disclosure is intended to cover any and all adaptations or variations of various embodiments. Combinations of the above embodiments, and other embodiments not specifically described herein, will be apparent to those of skill in the art upon reviewing the above description.
[0160] In this document, the terms "a" or "an" are used, as is common in patent documents, to include one or more than one, independent of any other instances or usages of "at least one" or "one or more." In this document, the term "or" is used to refer to a nonexclusive or, such that "A or B" includes "A but not B," "B but not A," and "A and B," unless otherwise indicated. In this document, the terms "including" and "in which" are used as the plain-English equivalents of the respective terms "comprising" and "wherein." Also, in the following claims, the terms "including" and "comprising" are open-ended, that is, a system, user equipment (UE), article, composition, formulation, or process that includes elements in addition to those listed after such a term in a claim are still deemed to fall within the scope of that claim. Moreover, in the following claims, the terms "first," "second," and "third," etc. are used merely as labels, and are not intended to impose numerical requirements on their objects.
[0161] The Abstract of the Disclosure is provided to comply with 37 C.F.R. § 1.72(b), requiring an abstract that will allow the reader to quickly ascertain the nature of the technical disclosure. It is submitted with the understanding that it will not be used to interpret or limit the scope or meaning of the claims. In addition, in the foregoing Detailed Description, it can be seen that various features are grouped together in a single embodiment for the purpose of streamlining the disclosure. This method of disclosure is not to be interpreted as reflecting an intention that the claimed embodiments require more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive subject matter lies in less than all features of a single disclosed embodiment. Thus the following claims are hereby incorporated into the Detailed Description, with each claim standing on its own as a separate embodiment.