Pedro Espadinha-Cruz
Radu Godina
Radu Godina
1,* and
Eduardo M. G. Rodrigues
Eduardo M. G. Rodrigues
UNIDEMI-Research and Development Unit in Mechanical and Industrial Engineering, Faculty of Science and Technology (FCT), Universidade NOVA de Lisboa, 2829-516 Almada, Portugal
Management and Production Technologies of Northern Aveiro—ESAN, Estrada do Cercal 449, Santiago de Riba-Ul, 3720-509 Oliveira de Azeméis, Portugal
Authors to whom correspondence should be addressed. Processes 2021, 9(2), 305; https://doi.org/10.3390/pr9020305Submission received: 31 December 2020 / Revised: 25 January 2021 / Accepted: 3 February 2021 / Published: 6 February 2021
(This article belongs to the Special Issue Advanced Process Monitoring for Industry 4.0)For decades, industrial companies have been collecting and storing high amounts of data with the aim of better controlling and managing their processes. However, this vast amount of information and hidden knowledge implicit in all of this data could be utilized more efficiently. With the help of data mining techniques unknown relationships can be systematically discovered. The production of semiconductors is a highly complex process, which entails several subprocesses that employ a diverse array of equipment. The size of the semiconductors signifies a high number of units can be produced, which require huge amounts of data in order to be able to control and improve the semiconductor manufacturing process. Therefore, in this paper a structured review is made through a sample of 137 papers of the published articles in the scientific community regarding data mining applications in semiconductor manufacturing. A detailed bibliometric analysis is also made. All data mining applications are classified in function of the application area. The results are then analyzed and conclusions are drawn.
The last few decades have seen the birth of a great diversity of products and services associated with electrical and electronic equipment, and witnessed the presence of electronic and electrical equipment in a large number of products and services, which are subject to constant change [1]. During the last few years, since semiconductor manufacturing processes have gradually diminished in size, the number of transistors that can be fabricated on a sole silicon wafer can amount to a billion units [2]. In order to account for the dynamic evolution of production and distribution and the changes caused by technological advances and inventions, companies that operate in this field need to be flexible and to be able to adapt quickly to a constantly changing environment [3].
Semiconductor production is the process that creates integrated circuits, such as transistors, LEDs, or diodes that can be found in electrical devices and consumer electronics. During the front-end process, the crystalline silicon ingot is produced and the wafers are cut, the electrical circuits are created by photolithography and other chemical processes and, finally, they are electronically tested. In the back-end process, the chunks are cut from the wafer, wired (glued), encapsulated, and tested [4]. The semiconductor manufacturing industrial units (known also as fabs) are one of the highest capital-intensive and entirely automated production systems, in which agnate processes and equipment are utilized to manufacture integrated circuits through a wide range of extensive and complex processes with firmly controlled manufacturing processes, reentering process flows, advanced and complex equipment, and demanding deadlines for complying with constantly unpredictable demands of a constantly increasing product mix [5].
The concept Industry 4.0 involves employing artificial intelligence technologies, data mining techniques, big data and deep learning analysis to the current industrial infrastructure for the purpose of developing innovations that are disruptive [6]. The objective is to strive to put into practice this concept, which will allow flexible decision-making and smart manufacturing systems, as anticipated by the Industry 4.0 concept. Therefore, by turning Industry 4.0 a reality, the role of the Internet of Things (IoT) and additional emergent technologies will have a central role [7]. So far, the tendency to have unmanned operations and increasing automation in semiconductor production systems, as in other production technologies, is constantly growing [8].
Conventionally, semiconductor production systems are known for having a highly complex and lengthy manufacturing process. Typically, semiconductor wafers require a number of process steps that could easily surmount half of a thousand to be produced [9,10]. The level of complexity of every step is frequently equated to that of a medium-sized industrial unit, particularly in such areas such as logistics, planning, control, and data volume, among other steps. Consequently, growing requirements and pressure to perform with a high plant productivity pose a difficult challenge for companies operating in semiconductor manufacturing [1].
The ever-growing demand for integrated circuits that are able to deliver higher performances at lower costs is something semiconductor companies are well familiar with. Therefore, wafer metrology tools are employed for designing and producing semiconductors, cautiously monitoring line widths, film properties, and possible defects in order to improve the production process. Data mining techniques together with metrology tools and wafer verification abilities guarantee a close desired result of the electrical and physical properties of produced semiconductors. Data mining with wafer metrology can accurately and quickly recognize surface pattern defects, particles, and additional conditions that are capable of causing adverse effects on semiconductor performance [11].
Data mining is one of the areas of the knowledge data discovery process and is capable of providing innovative avenues for interpreting data. Data mining comprises the extraction of significant and implicit, previously unidentified, and possibly valuable information from data. Data mining offers the ability to detect patterns that are hidden amid a set of data. Data mining is the process of sorting and classifying data, then finding anomalies, patterns, and correlations in large data sets to predict outcomes. Employing a wide variety of techniques, companies can use this information for problem detection, quality control, increase revenue, cut costs, improve customer relationships, and reduce risk, among others [12]. Since modern semiconductor manufacturing processes suffer from a great degree of complexity, and the amount of data is overwhelming, it is still challenging to reach fast yield improvement by discovering manually useful patterns in raw data [11].
Throughout wafer manufacturing, equipment data, process data, and the historic data will be semiautomatically or automatically collected and grouped in a database in order to be able to diagnose faults, to monitor the process, and to effectively manage the production process. Nevertheless, in such advanced manufacturing units such as semiconductor production, numerous aspects and details are interconnected and have an effect on the yield of the produced wafers [13]. Therefore, data mining techniques are a solution for a significant amount of challenges that the semiconductor manufacturing faces, such as yield improvement [5,11], quality control [14], fault detection [15], predictive maintenance [16], virtual metrology [17], scheduling [18], business improvement [19], and market forecasting [20], among others.
Despite the existence of a high number of studies regarding data mining applications in semiconductor manufacturing, a gap was identified in the literature, in which the necessity to compile and analyze in a more comprehensive way through the compilation in a single paper every published study arose, and expressly perform it without restrictions on location or characteristics. With the intention of filling the identified gap in the research, the aim of this paper is to compile all the existing publications on this topic on Scopus and WoS and to classify and compare them. Therefore, one of the goals of this study is to understand the state of the art regarding data mining solution to existing challenges in semiconductor manufacturing. A bibliometric study is presented, in which are analyzed the number of publications over time, the co-occurrence network, the most cited authors, the distribution of keywords by observed frequency, among other bibliometric metrics. This analysis, besides analyzing bibliometric indicators and making a comparison between distinct features, it also has the purpose to frame these indicators in distinct categories and highlighting every case, not only to seek and detect future research pathways, but also to have a better comprehension of data mining applications in semiconductor industry and to endorse it in order to disseminate its use.
This paper is organized as follows. In Section 2, a brief overview of the semiconductor manufacturing process is given. In Section 3, a structured bibliometric analysis is made. In Section 4, a qualitative organization and analysis data mining application studies in semiconductor manufacturing can be found. In Section 5, a brief result analysis and discussion is made. Finally, in Section 6, overall conclusions are given.
According to the literature, a systematic literature review neutralizes the perceived weaknesses of a narrative review [21]. A systematic literature review usually has distinct stages of preparation, direction-finding and publishing, and diffusion. Every stage might comprise numerous steps of the review process by being part of a method or system that is created to precisely and objectively focus on the overall question the review is bound to answer. In this study, the research design applied in [21,22,23,24] was followed, as seen in Figure 1, by comprising five steps: problem conception; literature search; research evaluation; research analysis; and finally result summarizing.
The objective of this bibliometric analysis is to know the state-of-the-art of data mining application in the semiconductor manufacturing. In a scenario where companies store large amounts of data, data mining approaches are used to extract useful information and knowledge automatically [25]. To achieve that, data mining approaches use a combination of algorithms and concepts from artificial intelligence, statistics, machine learning, and data management [26]. Accordingly, in this bibliometric analysis we look for data mining applications in semiconductors where authors attempt to extract information and knowledge in semiconductor manufacturing from large datasets.
After the topic of data mining data mining applications in semiconductor manufacturing was selected as an object of intensive study in this literature review, an extensive bibliographic research was carried out on the subject and its surroundings. The purpose of this analysis is to identify and evaluate the adopted methodologies of data mining applications in semiconductor manufacturing, by taking into account all the scientific studies found.
The research methodology was carefully developed in order to allow the identification of relevant patterns and areas for the study under analysis. The literature research process comprises such characteristics as the collected qualitative and quantitative information being well defined and delimited, a detailed analysis being made based on the evidence and characteristics recognized in the subject of the study, the analyzed papers are organized by application areas, all contents are analyzed in a qualitative manner, which favors the identification of important subthemes and the successful interpretation of results. We considered papers that address the application of data mining to exploit data stored during semiconductor manufacturing processes. So, in the first step, the usefulness of each article was verified by reading its summary and introduction, so that those who seemed to be out of the review due to imprecision and a lack of details were excluded. Additionally, despite that some of the data mining algorithms and techniques may be applied by semiconductor manufacturing authors, we excluded any papers that do not approach its use for information and knowledge extraction. After defining the aforementioned delimitations, a more detailed analysis was made on the articles that effectively added value in their incorporation in the review article. The purpose of data mining application has been carefully revised. This more detailed analysis includes: a selective reading and choice of material that suits the objectives and proposed theme; an analytical reading of the texts grouping them by application areas; and concludes with the interpretative reading and writing of the literature review body.
After the main elements of the research process have been well established, it becomes essential to adopt some essential assumptions for the accomplishment of this analysis. First, following the guidelines from [27], only indexed and peer-reviewed articles were taken into account, and the indexing databases considered were Scopus and Web of Science (WoS). The keywords utilized were “Data Mining” and “Semiconductor Manufacturing”, which garnered the highest number of results. However, also, all the possible variants, such as “Semiconductor Fabrication”, “Semiconductor Production”, and “Semiconductor Packaging” were utilized in order to cover all the possible published papers through this combination. Table 1 shows the results from different combinations of keywords in the database.
The publications considered for this study were publications in English and the type of articles were journal research articles, journal review articles, conference articles, book chapters, and editorials. A few papers were found in Chinese and Polish, but were excluded from this study. In Figure 2 the flowchart of the paper selection process can be observed. In the end, a final sample of 137 papers was used for the article analysis. This sample comprises almost all papers found with the keywords used.
All the selected studies were classified by year and the result can be seen in Figure 3. Three waves can be seen, the first wave that comprises paper from 2004 to 2007 peaked in 2006 with 10 publications and then the interest waned. The second wave peaked in 2014 and comprises the years 2011 until 2015. Finally, the last wave of interest in this topic can be seen, peaking in 2019, with 12 publications. This wave is still ongoing. However, if divided by decades, one can notice that the decade 2010–2020 comprises 64% of all publications, while the previous decade comprises only 33.5%. This interest reveals the growing scientific interest in this topic. This increase coincides with the overall interest in data mining applications for other industries [28,29].
A particular importance has to be given to the papers that garner the highest interest in the community, which is measured by the number of citations that a study has. Figure 4 shows the most cited studies of data mining applications in semiconductor manufacturing, according to Scopus. It can be observed that the first four articles are much more cited than the remaining ones. The most cited paper is proposed by [30] and deals with maintenance. It addresses a multiple classifier machine learning technique for predictive maintenance in the ion implantation process, and, at the time of the writing of this study, it is only 5 years old. The second most cited article is an overview data preprocessing with two examples, with one in semiconductor manufacturing [31]. This study has more than two decades and it is one of the main reasons why it has 185 citations. The third most cited study deals with quality issues and proposes a framework that combines traditional statistical methods and data mining techniques for fault diagnosis and low yield product for the process of wafer acceptance testing and probing [13]. Finally, the fourth most cited study, with 168 citations, addresses a rule-structuring algorithm based on rough set theory to make predictions for the semiconductor industry [32]. This study is focused on decision support systems and has almost two decades. Still, these four studies, which address data mining applications in different contexts and areas of semiconductor manufacturing and distinct subprocesses, are an example of how vast the applications of data mining techniques in this process are. The interest that these studies attracted is a staple in their respective subcategories of semiconductor manufacturing. Lotka’s Law states that the large number of small paper producers bring together about as much as the small number of large paper producers [33]. The frequency distribution of scientific productivity according to Lotka’s law is shown in Figure 5, Chen-Fu Chien being the most productive author. This can also be observed in Figure 4, in which Chen-Fu Chien is the author of nine of the most cited papers, since Chen-Fu Chien is also a coauthor of the fifth [34] and last [5] most cited papers from this figure.
A bibliometric keyword analysis was performed. This analysis was made with the help of VOSViewer software [35] and biblioshiny, which is a web application for Bibliometrix, and R Package [36]. Both have similar but distinct applications. First, the intention was to identify which were the most employed keywords. Therefore, a keyword analysis with VOSViewer software was performed with the main goal to evaluate the specifics of the discussion on how data mining applications in semiconductor manufacturing.
For the goal of this paper, the Keywords Plus function has been employed with the purpose of harmonizing the keywords that other authors have employed in the Abstract and Keyword section of their respective publications. This analysis shows that 2845 keywords were employed in the selected studies. However, only 51 of these terms appear at least 12 times. The six keywords with the highest occurrences are “data” (which appears 264 times), process (which appears 134 times), system (appearing 117 times), approach (appearing 109 times), and, finally, terms “model” and “semiconductor manufacturing” (both appearing 94 times). The network of co-occurrence links between these keywords is also shown in this paper with the intention of complementing the analysis of keywords co-occurrence. The generated keywords co-occurrence network map can be observed in Figure 6. Three different clusters can be observed.
However, another analysis was made with biblioshiny of the Bibliometrix, from the R Package. With this application it is possible to go more in-depth regarding keyword analysis. Here, only keywords inserted by the authors of their respective papers were considered. The top five keywords that are inserted more often are “data mining”, “semiconductor manufacturing”, “machine learning”, “feature selection”, and “yield enhancement”. However, by making just this simplified analysis not enough can be deduced. In Figure 7 the obtained frequency chart with biblioshiny can be observed with the distribution of the 47 most often found keywords in the selected sample of papers. A total of 349 keywords were found through the simplified technique employed in [37] to represent Zipf’s law. This law stated that certain terms occur much more frequently than others and the distribution is similar to a hyperbole 1/n. As the authors from [37], however, the occurrence of the keywords is stratified in decreasing order of frequency and categorized into three areas of analysis. First, the most important zone represents the basic or trivial information area, which shows the most essential terms on the subject. The second zone comprises the terms considered “interesting information”. This zone can comprise potentially innovative information and fringe themes. Finally, the last area is the noise zone. This area could represent concepts not yet emerging or even simply, noise.
The term “semiconductor” refers to a critical component in millions of electronic devices employed in current daily lives in education, research, communications, healthcare, transportation, energy, and other industries. Smartphones, mobile, wearable devices rely on semiconductors for both core operations and advanced functions and are driving global demand for semiconductors and printed circuit boards (PCBs).
The line width of semiconductors has undergone a drastic reduction, passing from the micrometer to the nanometer scale, while, in parallel, the process power and memory have been increased. Integrated circuits, made of a semiconductor material (such as silicon), are an important part of modern electronic devices in both commercial and consumer industries. These circuits must have the ability to act as an electrically controlled on/off switch (transistor) in order to perform basic arithmetic operations in a computer. To achieve this almost instantaneous switching capability, the circuits must be made of a semiconductor material, a substance with electrical resistance that lies between a conductor and an insulator.
The manufacturing process for semiconductor devices requires several steps that take place in highly specialized facilities. Semiconductor production is a considerably complex process with long lead times that are necessary to deliver the capabilities expected from everyday use of our devices. The semiconductor production times vary depending on the complexity; however, on average, it can take three to five years from initial research to final product.
Highly pure silicon is the most important raw material for the production of microelectronic components such as ICs, microprocessors, and memory chips. Figure 8 shows a summarized version of the manufacturing process. The first step in manufacturing a semiconductor device is to obtain semiconductor materials, such as germanium, gallium arsenide, and silicon, of the desired level of impurities [38,39]. Impurity levels of less than one part in a billion are required for most semiconductor manufacturing [40,41]. Due to the microscopic size of semiconductors, even the slightest hint of contamination can compromise their performance. The partly aggressive liquids required in the further manufacturing process of the microchips for metallizing, developing, etching, and cleaning should be safely conveyed, circulated, and processed [42].
The second main step is the crystal growth of monocrystalline silicon and growth of multicrystalline ingots [43]. Then, from these ingots, wafers are cut, and then shaped, polished, and cleaned with the purpose of being ready for further processing or for device manufacturing [44]. To achieve a functional device with predetermined specifications as a final result, it is necessary to carry out a prior design process for each of the manufacturing steps and a mask design, especially, for the masks used in the photolithographic processes that makes semiconductor manufacturing possible. The mask comprises the master copy of the pattern that will be printed on the wafer [45].
The next important step consists of chemical mechanical planarization or chemical mechanical polishing (CMP) is a process in which topographical irregularities can be removed from wafers with a combination of chemical and mechanical (or abrasive) polishing in order to obtain the smoothest surface possible [46,47]. The process is usually used to planarize oxide, polysilicon, or metal layers in order to prepare them for the subsequent lithographic step [48,49]. During ion implantation, high-energy ions are shot onto the substrate to be doped by the doping agent. The distribution of the implanted atoms in the semiconductor can be specifically influenced by the energy, the entry angle, and the use of masks. With multiple implants carried out one after the other, even complex doping profiles can be produced with good accuracy and replicability [50,51].
As seen in Figure 8, one of the most important steps in semiconductor manufacturing is extreme ultraviolet (EUV) lithography a process that allows carving more electrical circuits in semiconductor silicon wafers. In a lithographic system, images are transferred to silicon with light [52,53]. EUV lithography is considered to be essential to semiconductor manufacturing since it is able to produce a shorter wavelength that allows a greater quantity of electrical circuits to enter a chip [54]. Then, an important step is etching, which is utilized in microfabrication to chemically eradicate layers of a material from the surface of a wafer in order to create a pattern of that material on the substrate [55].
The following step is wafer probing, which is the procedure of electrically verifying each die on a wafer. This is accomplished by utilizing an automatic wafer probing system, which is actively searching for functional defects through by employing special test patterns [56,57,58]. The next step, semiconductor packaging and assembly process, involves enclosing ICs and encompasses from die-attach adhesives to liquid and film-shaped encapsulation compounds, sealing, lead forming/trimming, deflash, wirebonding, lead finish to heat-conducting materials, and conductive and non-conductive adhesives for sensors, among others. The encapsulation technology protects the sensitive layers from external influences and maintains their efficiency [59,60]. Finally, the final component is carefully tested in order to verify if it meets the requirements of standard specifications. The testing process is employed to test semiconductors in the context of design verification, specialized production, and quality assurance [61].
Data mining techniques can have a vast array of applications in the semiconductor industry. The obtained articles were classified accordingly to areas of application. Five major areas for data mining applications in semiconductor manufacturing emerged: quality control, maintenance, production, decision support systems, and finally, categorized as a whole, measurement, metrology, and instrumentation. However, other applications also exist, such as for human resources and talent recruitment and retainment [62], patent analysis [63], supply chain and inventory management [64], and stock market analysis [20], proving that data mining techniques can truly be employed for a wide range of applications.
Figure 9 shows the schematic representation of these applications. In some cases, only one article exists, and as such the direct reference is provided. In other cases, the identified five major areas are divided by subsections, in which a more detailed analysis is made. Additionally, this section is also useful for practicing engineers, since they can quickly find the semiconductor process step or data mining model they are looking for. They can also find the study that has been implemented and validated in industrial setting and through corresponding references, access to it.
Misaligned image processing can cause thousands of auxiliary operations and damaged wafers during a machine’s life during the photolithography process, wafer scrutiny and inspection, or wafer mounting and cutting [65]. Inefficient image processing systems cost semiconductor companies market share and contribute significantly to their overall costs [66]. Data mining techniques are able to provide robust, precise, and fast wafer and chip pattern location for wafer inspection, probing, assembly, cutting, and test equipment to avoid such types of problems. These techniques allow manufacturers to control the quality of wafers and chips with high precision and accuracy, ensuring reliable equipment performance during the semiconductor manufacturing process.
The main purpose of quality prediction tools is to forecast the behavior of the product and then to be able to also forecast the trends of values of its critical parameters, typically accomplished by employ learning functions that have the capacity to stem knowledge from the preceding information. Forecasting quality with the help of data mining techniques normally starts by creating a model based on previous data, for instance labeling samples, and then assess and verify the unidentified samples, or to evaluate, from a given sample, the attributes’ value ranges [67].
Table 2 shows the categorized papers by data mining applications for quality control in distinct steps of semiconductor manufacturing. These steps are identified, when possible, and can be found in the summary proposal. The table is subdivided into eight major columns and in a few can be observed the year of publication, reference, and the overall summarized description of the study. One of the remaining columns describes the proposed and/or used data mining algorithm, which can be helpful by quickly identifying a specific algorithm. The next column shows which DM technique is used. The remaining columns show if the sample data is collected from a real production site or if it was simulated, and if it is real, it is identified, when possible, by company and country of origin. Additionally, if experimental validation studies were performed on site, it is also highlighted.
This topic is the most popular one, with 47 publications. By observing Table 2, it can be seen that several applications are made in distinct subprocesses such as wafer probing and testing process, etching process, and photolithography, among others. A high and varied number of algorithms are employed. The majority of articles address challenges of correctly identifying defective patterns in order to improve production yield [68]. Yield is a quantitative measure of the quality of a semiconductor process. It is measured as the number of functioning dies or chips on a wafer and can also be seen as the fraction of dies on the yielding wafers that are not rejected during the production process [107]. However, other applications in quality control can also be found, such as a study addressing a design-of-experiment (DOE) data mining for yield-loss diagnosis for semiconductor manufacturing by detecting high-order interactions, for subprocesses such as lithography and etching, among others [85]. These data mining technique are also used with statistical process control. Cumulative sum control charts, known as CUSUM, are a special type of statistical process control tool that is used in [89] as part of and unified outlier detection framework, which takes advantages of data complexity reduction by employing entropy and sudden change detection through the use of CUSUM charts.
Only a few articles were published addressing maintenance management and prediction, but are important nonetheless. Only five papers were classified and can be observed in Table 3. This table is organized as Table 2. As it can be noticed, these studies are sparse and the majority were published in the last 8 years. However, the most cited article is a study in this area of application. In this study a multiple classifier machine learning methodology for predictive maintenance in the ion implantation subprocess is proposed [30] and a similar study is proposed in [16]. In another study, hidden Markov model-based predictive maintenance for semiconductor wafer production equipment and documented over one year was proposed in [108]. A data mining technique that is able to deliver early warning by identifying tool excursion in real time for advanced equipment control in order to diminish atypical yield loss is proposed in [109] and was validated by practical applications in the field. Finally, the last study addresses spatial pattern recognition in order to improve the resolution and identification of defective and malfunctioning tools in semiconductor manufacturing developed and implemented at Advanced Micro Devices, Inc. (AMD) [110].
The high necessity for always striving to make progress regarding the yield of current semiconductor production processes and decrease the time-to-market for more advanced, innovative, and gradually elaborate designs and processes demands for process tools and wafers to be examined and verified with up-to-date measurement systems and equipment. Several papers, namely 19, are categorized in this topic, as depicted in Table 4. This table is organized as Table 2. The topics addressed in this section range from models comprising a precise semiconductor photolithography process control method through virtual metrology by employing significant correlations between focus measurement data encountered by data mining and tool data [111].
In fact, virtual metrology is a recurring topic, and is defined as a set of methods that allow predicting the properties of a wafer through sensor data and machine parameters in the manufacturing equipment, thus avoiding the highly expensive physical measurement of the wafer properties [112,113,114]. Since machine data is typically sampled much more often when compared to metrology data, and since machine data becomes immediately available when compared to the delays that frequently occur with metrology tools, an accurate virtual metrology is capable of meaningfully developing the process control and monitoring performance through a constantly supply of real-time forecasted metrology data. A few feature extraction methods for virtual metrology with multisensor data are proposed in [17,115,116].
However, other measurement and instrumentation were also proposed and classified. For instance, in [117] a real-time data mining solution with the segmentation, detection, and cluster-extraction (SDC) algorithm that can automatically and accurately extract defect clusters from raw wafer probe test production data is proposed. Additionally, a data mining that employs machine learning methods with the purpose of modeling unknown functional interrelations and to predict the thickness of dielectric layers deposited onto a metallization layer of the manufactured wafers is proposed in [118]. Finally, at IBM, a data mining technique with the purpose of automatically identifying and exploring correlations between inline measurements and final test outcomes in analog and/or radio frequency (RF) devices and by integrating domain expert feedback into the algorithm in order to identify and remove bogus autocorrelations [119]. Practical application and validation of this technique is made.
Another trend in semiconductor manufacturing is the use of decision support systems (DSS). A DSS is a system designed to support in solving unstructured and semistructured managerial problems, throughout all the decision process’ stages [132]. The DSS use in this area is not novel. Earliest publications in this area date to the 1990s (e.g., [133,134]). DSSs are used to support decision-making in activities like production scheduling, simulation, prediction, material selection, fault detection, quality, etc. DSSs may, sometimes, have a knowledge base, which requires artificial intelligence to provide knowledge to support the decision process. However, the earliest uses of DSS required knowledge modeling by knowledge engineers from documented and expert knowledge. Knowledge extraction from unprocessed data allowed one to discover hidden knowledge in large amounts of data. The use of data mining techniques to uncover knowledge to be modeled in DSS is a trend also present in semiconductor literature. Researchers apply data mining techniques to find patterns and hidden relations that may help in semiconductor decision making. Usually, the goal is to determine links between control parameters and product quality, essentially in the form of decision rules [135].
In Table 5 the literature where data mining is used to support the decision-making process in semiconductors’ manufacturing is presented. Analyzing this table, one can see that most contributions address yield management and failure detection issues (see [135,136,137,138,139,140,141,142,143,144,145]). The authors from [146] aim at the same problem, but focus on the development of a computer integrated manufacturing (CIM) system to improve product yield. Other articles provide isolated contributions. In [147], the authors propose the application of data mining techniques to support decision-making in HR management of high-tech companies. In [148], the authors suggest the integration of data mining in semiconductor manufacturing execution systems (MES). Last, in [32] provides a multi-purpose data mining application for predictions in semiconductor manufacturing.
Traditional methods for production planning often require complex calculations and do not always allow a prompt reaction to changes or short-term adjustments that may arise. Given the size of the semiconductor production lines in a factory, sensors within production equipment are capable of delivering enormous amounts of data. This data can be, in turn, used not only for machine control, but also for production analysis purposes, especially real-time production planning. This has the potential to bring great advantages, especially in those industrial units in which the production is affected by frequent dynamic changes in the orders to be processed or technical specifications. Additionally, machine learning processes are able to recognize patterns and automatically learn and operationalize practical forecast models from a wide variety of data sources and large amounts of data. Therefore, in the context of semiconductor manufacturing with its complex and numerous subprocesses, numerous data mining applications are proposed for the production and production planning environment.
Table 6 depicts the articles addressing data mining applications for production in semiconductor manufacturing. A total of 16 papers were found in this category. This table is structured as Table 2. It can be noticed that from 2009 until 2015 is when the bulk of these studies were published, then a four-year hiatus was observed. From 2019 can be noticed some interest in the topic.
Many of the studies concerning production planning are focused on reducing cycle time. In [155], a new approach that is capable of integrating data mining that intends to forecast arrival rates and determining the allocation of interchangeable tool sets in order to reduce the work in process (WIP) bubbles for cycle time reduction is proposed. While in another study [64], a cycle time forecasting model is developed by employing knowledge discovery in databases by following cross industry standards for data mining. A data-mining approach for estimating the interval cycle time of each job in a semiconductor manufacturing system is proposed in [156] and a data mining methodology, which identifies key factors of the cycle time in a semiconductor manufacturing plant, which intends to predict its value is addressed in [157].
Scheduling is another concern in semiconductor manufacturing due to its vast number of steps and jobs [158,159,160], confirmed by the majority of the identified studies in Table 6. Efficient order scheduling structures are required for balancing the production load and capacity throughout all the production stages [161]. A data mining dynamic scheduling strategy selection model that is able to respond to a constantly altering system status for a semiconductor manufacturing system is proposed in [18]. In [162] a data-driven scheduling knowledge life-cycle management for an intelligent shop floor is proposed and validated through a simulation model of the semiconductor production line. As early as in 2004 scheduling challenges were a concern, evidenced by a study proposing an hierarchical clustering method in [163] that is able to discriminate groups according to the similarity of the objects and used to schedule semiconductor manufacturing processes. In [164] a dynamic scheduling model, which is able to optimize the production features subset is proposed, and this model is capable of creating a SVM-based dynamic scheduling strategy classification model for semiconductor manufacturing. A data-based scheduling framework and adaptive dispatching rule for semiconductor manufacturing is addressed in [165] by employing backward propagation neuronetworks (BPNNs). Finally, a shop floor control system in semiconductor production by self-organizing map-based smart multicontroller is given in [166]. This study, as all the scheduling studies, showed a better system performance than the typical fixed decision scheduling rules.
After analyzing all the studies collected in the sample, a few trends begin to be noticed. First, that studies regarding data mining applications in subprocesses such as ICs and mask design are very scarce. The same occurs with studies addressing wafer cutting, cleaning drying, and polishing, while edge rounding and lapping subprocess has no dedicated study. This is better illustrated by Figure 10 in which a representation of several studies depicting data mining applications in several subprocesses of semiconductor manufacturing can be seen. It is noticeable that the majority of studies are concentrated in 5–6 major steps. A few studies do not specify in which subprocess data mining techniques are applied, and these are not represented in Figure 10.
Another trend visible in the analyzed literature is the diverse use of data mining techniques. The application of data mining in semiconductor manufacturing has a different focus depending on the subject areas concerning the manufacturing processes. However, most articles address mainly the issues of quality control, maintenance, and production. Predictive techniques, using algorithms as regression or decision trees, are often used in semiconductor literature to estimate wafer quality [81], fault detection [121,136], or cycle-time [170]. Classification techniques in quality control arise as a way to classify defects [83], failures in bin maps [91], or production lots [131]. The exploration of yield loss causes [84] or failure diagnostics [98] is performed using techniques as rule induction, decision trees, and association rules.
Many opportunities and improvements can still be made. For example, the semiconductor companies could employ the internet of things and sensors to empower industrial units with the capability of interpreting data and transmitting analytics, in real time, to an application that could provide insights and alerts to whom it may concern [174]. This will allow these players to gather a high amount of data. However, even though internet of things and data mining applications represent a key opportunity for semiconductor manufacturing companies—one that they should start to pursue as soon as possible, while the use of data mining in the sector is still developing under the current upgrading environment. Nevertheless, the effectiveness and scale of the internet of things implementation, and with it a comprehensive use of data mining techniques, could depend on how fast industry players can overcome some challenges [175]. In order to persevere and being able to accompany the change speed and challenges, semiconductor companies are required to adapt rapidly. Taking into account this dynamic, industrial units should embrace digitalization in an agile manner as well [176].
Even though employing data mining techniques has been very beneficial for this industry, as shown by all the studies used in this review, several disadvantages of data mining still exist and are as follows:
Data mining systems can violate privacy. Absence of safety and security can be very detrimental to its users and it can create miscommunication between employees, thus leading to genuine privacy concerns [177].
Security is an important factor related to every data-oriented technology, and semiconductor manufacturing is not an exception. Data that is very critical might be a target of malicious attacks [178].
Too much and redundant information collection can be disadvantageous as irrelevant collected information is a challenge [179,180].
There is a possibility of information misuse through the mining process. Data mining system have to evolve in order to diminish the misuse of the information ratio [181].
Accuracy of data mining techniques is another limitation [182]. Accuracy is an evaluation system of measurement on how well a data mining model can perform. Many common accuracy and error scores for regression and classification can occur. Therefore, improving accuracy becomes paramount.
Several challenges of data integration and interoperability in data mining can occur. Data interoperability and data integration affect the performance of an organization. A comprehensive approach has to be made in order to address the challenges in interoperability and integration [183,184].
Missing and imbalanced data is a challenge in this industry. In cases in which data is imbalanced, the majority of classification algorithms have as a consequence a weak performance. Since wafer yield enhancement is a crucial performance index in semiconductor wafer manufacturing, key process steps must be cautiously selected and managed [9].
Data processing time is another limitation that has a significant impact on the available time since data preprocessing very often involves more than 50% of time and effort of the entire data analysis process [185].
This evolution of semiconductor manufacturing relies heavily on the big data explosion in order to cope with the abovementioned data limitations and challenges of the semiconductor industry. Especially, supporting greater volumes and lengthier archives of data has allowed many solutions to correctly portray system dynamics, significantly simplify intricate multivariate interactions of parameters, eliminate disturbances, and clean and overcome data quality challenges. Data mining algorithms in such types of solutions must be rewritten in order to benefit from the parallel computation allowed by the high processing capacity and storage power with the purpose of processing data without consuming too much time. However, an enormous amount of data and a wide range of data mining techniques does not mean necessarily more predictive capability and insights [186]. Researchers and practitioners have to adapt data mining techniques in a manner so that these will be customized to specific applications in terms of data quality available data and objective, among others.
Overall, through this review, some light was shed over the possible applications of data mining techniques in semiconductor manufacturing. Yet, given the sheer number of steps that this production process has, and due to its complexity, the number of studies already made is still scarce. Big data and data mining allowed for original and innovative insights through the analysis of large amounts of data and presenting correlations and opportunities that were not previously noticed. However, decision makers must decide and which data should be collected and employed and which questions must be answered [149]. This signifies that the potential to apply these techniques in other subprocesses is enormous and is still left largely unexplored. Finally, by suffering constant and quick evolution, the need to adapt these techniques to the newer processes in semiconductor manufacturing is another opportunity to explore.
The production of semiconductors is a highly complex process, which entails several subprocesses that employ a diverse array of equipment. The size of the semiconductors signifies a high number of units can be produced, which require huge amounts of data in order to be able to control and improve the semiconductor manufacturing process. Therefore, in this paper a structured review was made through a sample of 137 papers of the published articles in the scientific community regarding data mining applications in semiconductor manufacturing. A detailed bibliometric analysis was made. All data mining applications were classified in function of the application area. Five distinct areas were identified: quality control, maintenance, production, decision support systems, and finally, categorized as a whole, measurement, metrology, and instrumentation. Results showed that quality was the most popular one, with 47 publications, making 34.3% of all publications. Maintenance was an area in which only a few studies were made, highlighting the gap and the opportunity for more studies to be made in this area.
The work performed in this study concerning data mining applications in semiconductor manufacturing can have theoretical implications. The characterization and categorization of several useful and successful cases can positively contribute to future research efforts of employing such a wide range of techniques with the purpose of increasing the application and diffusion of data mining applications in semiconductor manufacturing. Knowledge of different models and algorithms could have positive implications for the development of theory, for understanding all the possible applications in different areas of semiconductor production, but also for the development of practice, since many of these were implemented and validated on the shop floor. However, as the literature review has shown, many applications can still be made since several studies address only a specific step of semiconductor manufacturing and documentation of real-life application are scarce. Additionally, recent data mining techniques and models have a great opportunity to be used since only a few studies exist. Finally, since the semiconductor manufacturing process is always evolving, the need to adapt these techniques to the newer process is another challenge and opportunity to explore.
Overall, as seen from all the comprised studies from distinct steps of semiconductor production, the scope and functions of data mining techniques can be enhanced and disseminated throughout the entire semiconductor manufacturing process in order to provide, in real time, a proactive adjustment and advanced control decisions for the whole process and the smart facilities. Therefore, more research should be made to employ and facilitate smart production for Industry 4.0 in several industries for digital transformation and for upgrading existing manufacturing units. This will allow for an improving capability for optimizing interrelated decisions and improving decision flexibility.
Conceptualization and methodology, R.G. and P.E.-C.; software R.G.; validation and investigation, R.G. and P.E.-C.; review and editing, E.M.G.R. All authors have read and agreed to the published version of the manuscript.
The authors acknowledge Fundação para a Ciência e a Tecnologia (FCT-MCTES) for its financial support via the project UIDB/00667/2020 (UNIDEMI).
