Semiconductor Manufacturing Systems Research Articles

Preface for Semiconductor Manufacturing Systems and Processes

Preface for Semiconductor Manufacturing Systems and Processes

A fast ramp-up framework for wafer yield improvement in semiconductor manufacturing systems

A fast ramp-up framework for wafer yield improvement in semiconductor manufacturing systems

Cycle time analysis for cluster tools with parallel process chambers, processing time variation, and chamber cleaning operations

Cluster tools are the key equipment in semiconductor manufacturing. For high-end chip manufacturing processes, to ensure the wafer quality, periodical cleaning operations for process chambers in such tools are required for eliminating the residual gas and chemicals. In operating a tool, the wafer processing time in a process chamber may vary within a range. It is meaningful to predict the cycle time of cluster tools with chamber cleaning operations and processing time variation. By doing so, automation material handling system can be told when a lot is completed by a cluster tool such that it can assign an overhead hoist transporter to transport this lot at the right time. This plays an important role in improving the productivity of a whole semiconductor manufacturing system. This work conducts the cycle time analysis of cluster tools with chamber cleaning operations and processing time variation. Specifically, it proposes a novel method to calculate the average cycle time of a single-arm cluster tool under which an optimal schedule can be obtained. Then, for a dual-arm cluster tool, an efficient algorithm is developed to approximate the average system cycle time under which an optimal schedule can be obtained. Experiments show that the gap between the average system cycle time obtained by simulations and the average system cycle time obtained by the proposed method is no more than 0.1% which demonstrate the effectiveness of the proposed method.

Hyperparameter Tuning with Gaussian Processes for Optimal Abstraction Control in Simulation-based Optimization of Smart Semiconductor Manufacturing Systems

Smart manufacturing utilizes digital twins that are virtual forms of their production plants for analyzing and optimizing decisions. Digital twins have been mainly developed as discrete-event models (DEMs) to represent the detailed and stochastic dynamics of productions in the plants. The optimum decision is achieved after simulating the DEM-based digital twins under various what-if decision candidates; thus, simulation acceleration is crucial for rapid optimum determination for given problems. For the acceleration of discrete-event simulations, adaptive abstraction-level conversion approaches have been previously proposed to switch active models of each machine group between a set of DEM components and a corresponding lookup table-based mean-delay model during runtime. The switching is decided by detecting the machine group’s convergence into (or divergence from) a steady state. However, there is a tradeoff between speedup and accuracy loss in the adaptive abstraction convertible simulation (AACS), and inaccurate simulation can degrade the quality of the optimum (i.e., the distance between the calculated optimum and the actual optimum). In this paper, we propose a simulation-based optimization (SBO) that optimizes the problem based on a genetic algorithm (GA) while tuning specific hyperparameters (related to the tradeoff control) to maximize the speedup of AACS under a specified accuracy constraint. For each individual, the proposed method distributes the overall computing budget for multiple simulation runs (considering the digital twin’s probabilistic property) into the hyperparameter optimization (HPO) and fitness evaluation. We propose an efficient HPO method that manages multiple Gaussian process models (as speedup-estimation models) to acquire promising optimal hyperparameter candidates (that maximize the simulation speedups) with few attempts. The method also reduces each individual’s exploration overhead (as the population evolves) by estimating each hyperparameter’s expected speedup using previous exploration results of neighboring individuals without actual simulation executions. The proposed method was applied to optimize raw-material releases of a large-scale manufacturing system to prove the concept and evaluate the performance under various situations.

Clogging monitoring and regeneration of filtration film in wafer cleaning circulation systems for semiconductor manufacturing

Clogging monitoring and regeneration of filtration film in wafer cleaning circulation systems for semiconductor manufacturing

A simulation optimization framework to solve Stochastic Flexible Job-Shop Scheduling Problems—Case: Semiconductor manufacturing

This paper addresses a Stochastic Flexible Job-Shop Scheduling Problem (SFJSSP) in the context of semiconductor manufacturing. Semiconductor industry is among the most capital-intensive businesses whose operational excellence is of vital importance. Within the front-end fab of the semiconductor industry, the photolithography workstation is the well-known bottleneck process. To elevate the performance of the whole semiconductor manufacturing system, developing a competent schedule for its bottleneck is essential. However, the re-entrant product flows, high uncertainties in operations times, and rapidly changing products and technologies within the photolithography, make it difficult to develop a schedule for the whole semiconductor fab. Considering Industry 4.0, hybrid methods such as Simulation Optimization (SO) have proven their applicability in addressing complex production scheduling problems. Thus, this paper develops a mathematical model for SFJSSP of the semiconductor manufacturing considering special constraints of the photolithography workstation (machine process capability, machine dedication, and maximum reticles (masks) sharing constraints). Next, we transform the developed model into an SO model integrated with a computer simulation model capable of modeling the photolithography workstation. The simulation model develops an initial schedule based on the Least Work Remaining (LWR) dispatching rule. Moreover, the simulation model calculates the objective function of the SFJSSP. A tailored Genetic Algorithm (GA) is then developed, which attempts to optimize the initially proposed schedule. To validate the superiority of the presented SO methodology in addressing SJSSPs, it is compared with previously proposed methods. Furthermore, to assess the impact of the three special constraints of the photolithography work area on system performance, two sets of experiments are proposed. In the first set of experiments, the performance of two SFJSS environments, one with the special constraints and one without, is compared. The second set of experiments involves observing the system’s performance while systematically varying the severity of the special constraints. The results indicate that improved performance levels can be accomplished by enhancing flexibility within both the operations of individual jobs and the machines within the manufacturing system.

PSO-SVM Based Performance-Driving Scheduling Method for Semiconductor Manufacturing Systems

There are currently many studies on data-driven optimization scheduling, but only a few studies have combined “closed-loop optimization” with “performance-driven”. Therefore, this research proposed a PSO-SVM-based (particle swarm optimization optimized support vector machine) scheduling method that reconciles the composite dispatching rules (CDR), performance-driving ideology, and feedback mechanism ideology. Firstly, the composite dispatching rules coalesce flexible equipment maintenance, multiple process constraints, and dynamic dispatching. Secondly, the performance-driving ideology is carried out through two learning models based on the PSO-SVM algorithm, based on targeted optimizing performances. Thirdly, the feedback mechanism ideology makes the scheduling method realize closed-loop optimizations adaptively. Finally, the superiority of the proposed scheduling method is validated in a semiconductor manufacturing system in China. Compared with CDR, the proposed scheduling method combines MOV, PR, and EU, respectively EU_ O, EU_ P, PCSR and ODR increased by 7.85%, 5.11%, 8.76%, 8.14%, 6.60%, and 7.33%, indicating the superiority of this method.

CHARACTERISTICS AND OVERVIEW OF KNOWLEDGE OF CURRENT RESEARCH IN THE PRODUCTION OF HARMONIC GEARBOX

The range of applications for drives and gears stretches over numerous technological fields. The gearbox is found in every mechanical machine and represents only one small but important sub-application. Industry as a whole has a constant demand for new individual solutions tailored to solve particular problems. But again and again, it is called upon to find new ways of transmitting movements that a particular component is required to conduct so as to attain a particular result. Harmonic drive products are unique precision speed reducers playing important roles in robots, semiconductor manufacturing systems, factory automation equipment and furthermore, in aerospace equipment. The greatest benefits are the zero-backlash characteristics and the weight and space savings compared to other gears because our gear mechanism consists of only three basic parts. They are the: Wave Generator, Flexspline and Circular Spline.

Robust possibilistic programming-based three-way decision approach to product inspection strategy

During product inspection, Type I and Type II errors, widely considered to assess product quality, are represented by erroneously rejecting non-defective products and erroneously accepting defective products, respectively. Meanwhile, the formulation of product inspection strategy is oftentimes deemed as an interdependent process influenced by multiple factors (e.g., production technology and customer demand) from both upstream (e.g., production process) and downstream (e.g., customer zone). Due to the limited understanding and knowledge of decision-makers on imprecise parameters of these factors, it is challenging for decision-makers to formulate an appropriate product inspection strategy. To address the aforementioned challenges, an improved three-way decision (3WD), known as robust possibilistic programming (RPP)-based 3WD is put forth. Such an improvement for 3WD is evident in two aspects (1) the RPP model endows 3WD to address imprecise parameters and chance constraints using flexible attitudes, and (2) the RPP model ensures the formulation of the product inspection strategy of 3WD is robust. Additionally, with the aid of 3WD, an intermediate/uncertain decision, is introduced to simultaneously reduce the aforementioned two types of errors. A case study of a semiconductor manufacturing system, along with discussions and two comparative studies, is presented to demonstrate the effectiveness and performance of this approach. As shown in this study, the proposed approach (1) effectively prevents decision-makers from making an over-conservative attitude in the formulation of the product inspection strategy, (2) has better performance in satisfying customer demand, and (3) has significant advantages in terms of cost and computational time.

Dual-Arm Cluster Tool Scheduling for Reentrant Wafer Flows

Cluster tools are the key equipment in semiconductor manufacturing systems. They have been widely adopted for many wafer fabrication processes, such as chemical and physical vapor deposition processes. Reentrant wafer flows are commonly seen in cluster tool operations for deposition processes. It is very complicated to schedule cluster tools with reentrant processes. For a dual-arm cluster tool with two-time reentering, the existing studies point out that a one-wafer periodical (1-WP) schedule can be found, and it is optimal in terms of productivity. However, for some wafer fabrication processes, wafers should be processed at some PMs more than two times. This gives rise to a question of whether there still exists a 1-WP schedule for dual-arm cluster tools with the number of reentering times being more than two such that the cycle time of a tool can reach the lower bound. This problem is still open, and this is what this work wants to tackle. For a dual-arm cluster tool with the number of reentering times being k (>2) times, if there does not exist a value f ∈ {1, 2 …} such that k = 3f, theoretical proofs are given to show that a 1-WP schedule can be found, otherwise it does not exist. For cases with a 1-WP schedule, the cycle time can be obtained by analytical expressions. For the cases without a 1-WP schedule, two new methods for a three-wafer periodical schedule are proposed to improve the system productivity by comparing it with an existing three-wafer periodical schedule. The applications of the obtained results are demonstrated by examples. Wafer residency time constraints are required for some wafer fabrication processes. Note that the results obtained in this work cannot be directly applied to cluster tools with both reentrant wafer flows and wafer residency time constraints. Nevertheless, schedulablity and scheduling analyses for that applications can be conducted based on the obtained results in this work.

Formulations and an adaptive large neighborhood search for just-in-time scheduling of unrelated parallel machines with a common due window

In this paper, we address an unrelated parallel machine scheduling problem inspired by aspects of semiconductor manufacturing and other production systems. The objective is to minimize the total sum of weighted earliness and tardiness penalties when jobs are subject to a common restrictive due window. We conceptualize the problem, present two mixed-integer linear programming (MILP) formulations, and give necessary but not sufficient conditions for an optimal schedule. Since optimal schedules do not necessarily start at time zero, we derive two constructive heuristics coupled with a shift-forward procedure that determines leading idle times based on the elimination of straddling jobs. We propose a novel earliness–tardiness extension of the adaptive large neighborhood search (ET-ALNS) with several ad hoc removal/insertion operators and an embedded procedure to accelerate the calculation of the objective function. We carry out extensive computational experiments on a new set of 600 benchmark instances and compare our methods against two modern solvers with an incorporated warm-start mechanism as well as other state-of-the-art algorithms designed for closely related problems. A careful statistical analysis is conducted and results show that ET-ALNS outperforms the remaining methods by a significant margin.

Prehľad doterajšieho výskumu výroby harmonického prevodu

Harmonic Drive, also called Harmonic Drive gear, harmonic gear, or strain wave gearing, mechanical speed-changing device, invented in the 1950s, that reduces the gear ratio of a rotary machine to increase torque. It operates on a principle different from that of conventional speed changers. Harmonic drive products are unique precision speed reducers playing important roles in robots, semiconductor manufacturing systems, factory automation equipment and furthermore, in aerospace equipment that may convey our human dreams.

Cooperative Product Agents to Improve Manufacturing System Flexibility: A Model-Based Decision Framework

Due to the advancements in manufacturing system technology and the ever-increasing demand for personalized products, there is a growing desire to improve the flexibility of manufacturing systems. Multi-agent control is one strategy that has been proposed to address this challenge. The multi-agent control strategy relies on the decision making and cooperation of a number of intelligent software agents to control and coordinate various components on the shop floor. One of the most important agents for this control strategy is the product agent, which is the decision maker for a single part in the manufacturing system. To improve the flexibility and adaptability of the product agent and its control strategy, this work proposes a direct and active cooperation framework for the product agent. The directly and actively cooperating product agent can identify and actively negotiate scheduling constraints with other agents in the system. A new modeling formalism, based on priced timed automata, and an optimization-based decision making strategy are proposed as part of the framework. Two simulation case studies showcase how direct and active cooperation can be used to improve the flexibility and performance of manufacturing systems. Note to Practitioners—An intelligent product is a product in a manufacturing system that is able to make decisions based on a set of specifications and affect its own production process. Intelligent products have often been proposed to address the challenges associated with small-batch manufacturing and highly customized production. Specifically, by using intelligent products, manufacturers would be able to complete small orders without the need to reconfigure or reschedule operations in the manufacturing system. However, one of the major challenges in the implementation of this control strategy is the need to develop methods that allow intelligent products to cooperate with machines, robots, and other products in a manufacturing system. In this work, we propose a novel direct and active cooperation framework that allows intelligent products to communicate and cooperate with other resources and products on the shop floor. Using the proposed cooperation framework, intelligent products can resolve scheduling conflicts and work together to meet individual specifications (e.g., deadlines). Two case studies, a small job shop and a large semiconductor manufacturing system, showcase how the proposed cooperation framework can be leveraged for different types of applications.

Scheduling for a flow shop with waiting time constraints and missing operations in semiconductor manufacturing

ABSTRACT Semiconductor manufacturing systems consist of highly complex and constrained processes owing to the requirement for high-quality wafers and dynamic customer orders. This study focuses on waiting time constraints to achieve high-quality wafers, with missing operations, to manufacture different types of orders, which are common scheduling requirements in semiconductor manufacturing systems. Waiting time constraints limit the waiting times between consecutive machines. Although sequential steps are modelled as a flow shop, some jobs can skip some steps in the flow shop. Such jobs can accordingly skip certain machines in the flow shop, which is called ‘missing operations’. This research investigates a scheduling problem for a flow shop with waiting time constraints and missing operations. A mathematical programming formulation and scheduling algorithms are proposed to operate such a flow shop efficiently. Through computational experiments, the proposed algorithms were evaluated, and analysis of various parameters for waiting time constraints and missing operations was conducted.

Learning-Based Grey Wolf Optimizer for Stochastic Flexible Job Shop Scheduling

This work considers a stochastic flexible job shop scheduling with limited extra resources and machine-dependent setup time in a semiconductor manufacturing environment, which is an NP-hard problem. In order to obtain its reliable and high-performance schedule in a reasonable time, a learning-based grey wolf optimizer is proposed. In it, an optimal computing budget allocation-based approach, which is designed for two scenarios from real manufacturing environments, is proposed to intelligently allocate computing budget and improve search efficiency. It extends the application area of optimal computing budget allocation by laying a theoretic foundation. Besides, to obtain proper control parameters iteratively, a reinforcement learning algorithm with a newly designed delay update strategy is used to build a parameter tuning scheme of a grey wolf optimizer. The scheme acts as a guide for balancing global and local search, thereby enhancing effectiveness of the proposed algorithm. The theoretic interpretation of the developed optimal computing budget allocation-based approach and the convergence analysis results of the proposed algorithm are presented. Various experiments with benchmarks and randomly generated cases are performed to compare it with several updated algorithms. The results shows its superiority over them. Note to Practitioners—Meta-heuristic are often deployed to solve semiconductor manufacturing scheduling problems. However, they face to two thorny issues when they face stochastic manufacturing environments. 1) their computational efficiency is quite low, thus requiring substantial improvement, since a stochastic optimization problem requires Monte Carlo sampling to estimate the actual objective function values in a precise manner; and 2) most of them are parameter-sensitive, and choosing their proper parameters is highly challenging in such environments. To address the first issue, we develop an optimal computing budget allocation-based method for deciding the optimal numbers of sampling times based on both prior knowledge and simulation results. To address the second one, we propose a reinforcement learning algorithm to self-adjust the parameters of our proposed method called Learning-based Grey Wolf Optimizer. In addition, we design a delay update strategy to enhance its robustness, and thus, a feasible and high-quality schedule can be founded in a short time for real-time scheduling problems. Theoretic proofs and experimental results show that the proposed method is effective and efficient. Consequently, it can be readily applicable to practical semiconductor manufacturing systems.

Dynamic scheduling for semiconductor manufacturing systems with uncertainties using convolutional neural networks and reinforcement learning

The dynamic scheduling problem of semiconductor manufacturing systems (SMSs) is becoming more complicated and challenging due to internal uncertainties and external demand changes. To this end, this paper addresses integrated release control and production scheduling problems with uncertain processing times and urgent orders and proposes a convolutional neural network and asynchronous advanced actor critic-based method (CNN-A3C) that involves a training phase and a deployment phase. In the training phase, actor–critic networks are trained to predict the evaluation of scheduling decisions and to output the optimal scheduling decision. In the deployment phase, the most appropriate release control and scheduling decisions are periodically generated according to the current production status based on the networks. Furthermore, we improve the four key points in the deep reinforcement learning (DRL) algorithm, state space, action space, reward function, and network structure and design four mechanisms: a slide-window-based two-dimensional state perception mechanism, an adaptive reward function that considers multiple objectives and automatically adjusts to dynamic events, a continuous action space based on composite dispatching rules (CDR) and release strategies, and actor–critic networks based on convolutional neural networks (CNNs). To verify the feasibility and effectiveness of the proposed dynamic scheduling method, it is implemented on a simplified SMS. The simulation experimental results show that the proposed method outperforms the unimproved A3C-based method and the common dispatching rules under the new uncertain scenarios.

A global scheduling approach for cycle time control in complex manufacturing systems

This paper proposes a novel global scheduling approach for cycle time control strategy in large complex manufacturing systems with multiple workcentres, such as semiconductor manufacturing systems. The interaction between workcentres is taken into account by using global information such as release quantities, the Work-In-Process (WIP), cycle time targets and machine capacities. Local scheduling decisions in workcentres are steered by production targets, i.e. quantities of products to complete in each operation and each period on a scheduling horizon. These global production targets are determined by a mathematical model (global scheduling model), which optimises the satisfaction of cycle time targets. One of the major innovations of the proposed model is that it relies on the temporal trace of the WIP. The mathematical model is coupled with a generic multi-method simulation model for evaluation purpose. Computational experiments conducted on industrial data show that our global scheduling approach efficiently controls the cycle times of products.

Metaheuristics for a Flow Shop Scheduling Problem with Urgent Jobs and Limited Waiting Times

This study considers a scheduling problem for a flow shop with urgent jobs and limited waiting times. The urgent jobs and limited waiting times are major considerations for scheduling in semiconductor manufacturing systems. The objective function is to minimize a weighted sum of total tardiness of urgent jobs and the makespan of normal jobs. This problem is formulated in mixed integer programming (MIP). By using a commercial optimization solver, the MIP can be used to find an optimal solution. However, because this problem is proved to be NP-hard, solving to optimality requires a significantly long computation time for a practical size problem. Therefore, this study adopts metaheuristic algorithms to obtain a good solution quickly. To complete this, two metaheuristic algorithms (an iterated greedy algorithm and a simulated annealing algorithm) are proposed, and a series of computational experiments were performed to examine the effectiveness and efficiency of the proposed algorithms.

Simulation and deep reinforcement learning for adaptive dispatching in semiconductor manufacturing systems

Simulation and deep reinforcement learning for adaptive dispatching in semiconductor manufacturing systems

Evaluation System and Correlation Analysis for Determining the Performance of a Semiconductor Manufacturing System

Numerous performance indicators exist for semiconductor manufacturing systems. Several studies have been conducted regarding the performance optimization of semiconductor manufacturing systems. However, because of the complex manufacturing processes, potential complementary or inhibitory correlations may exist among performance indicators, which are difficult to demonstrate specifically. To analyze the correlation between the performance indicators, this study proposes a performance evaluation system based on the mathematical significance of each performance indicator to design statistical schemes. Several samples can be obtained by conducting simulation experiments through the performance evaluation system. The Pearson correlation coefficient method and canonical correlation analysis are used on the received samples to analyze linear correlations between the performance indicators. Through the investigation, we found that linear and other complex correlations exist between the performance indicators. This finding can contribute to our future studies regarding performance optimization for the scheduling problems of semiconductor manufacturing.