Optimization of AMHS design for a semiconductor foundry fab by using simulation modeling

A 300mm semiconductor fab requires a tremendous amount of investment with a great deal of process tools inside. Those process tools are connected by the automatic material handling system (AMHS), thus requiring an extremely complex AMHS network. Additionally, unlike simple products of an integrated device manufacturer (IDM) or a memory fab, foundry fab manufactures hundreds of consumer products simultaneously, subsequently creating globally an extremely high transportation volume and complex demands on the material handling system. Previously, experiential engineers designed AMHS, possibly taking months of design lead-time. However, the next generation fab AMHS design focuses not only AMHS specifications, but also operational know-how. Meanwhile, a long design lead time fails to fulfill the requirements of the fast ramp up plan. Additionally, design optimization and shrink lead time are achieved by applying simulation modeling as a design platform whenever TSMC constructs an AMHS for a new fab. Via this platform, precise simulation is performed on the AMHS specifications, fab layout, tool configuration, and process flow in-formation to ensure design success and avoid a potential AMHS bottleneck.

In semiconductor manufacturing, most of 200mm fabrication facilities (called fabs) currently existing operate in manual or in semi-automatic material handling system. To remain competitive and to deal with the migration from high-volume/low-mix to high-mix/low-volume, most of these fabs require investment on factory automation especially on handling operations. This paper addresses a problem of Automated Material Handling System (AMHS) design for reticle in a photolithography area of 200mm wafer fab that was not originally designed to have such system. The objective is to define AMHS components: tracks layout, fleet size and management and reticle storing system while taking into account the existing area layout constraints (machine locations, storage area, corridors dimensions). A simulation approach is adopted to model the Automated Reticle Handling System (ARHS) and analyze the effect on key performance indicators. Using real instances, many simulation tests with alternative designs have been conducted which provided practical feedbacks.

To perform a high fidelity simulation study on a 300 mm wafer fabrication plant, modeling of the manufacturing process (MP) alone is not sufficient. Inclusion of the automated material handling system (AMHS) model is necessary due to the high degree of factory automation. There isn't, however, a single tool that is capable of modeling both the AMHS and MP with sufficient accuracy and granularity. A commercial simulation package such as AutoMod is usually used to model the AMHS while AutoSched AP is usually used to model the MP. These packages can be integrated using the supplied interoperation module but flexibility in optimizing the execution performance for different simulation models is lacking. In this paper, we present an approach to interoperation based on the High Level Architecture standard. We note that the typical characteristics of disparity in the models' time granularity and frequent model interactions are the obstacle to good execution performance.

We address the field of internal logistics, embodied in Automated Material Handling Systems (AMHSs), which are complex installations employed in sectors such as Baggage Handling, Physical Distribution, and Parcel & Postal. We work on designing an integral planning and real-time control architecture, and a set of generic algorithms for AMHSs. Planning and control of these systems need to be robust, and to yield close-to-optimal system performance. Currently, planning and control of AMHSs is highly customized and project specific. This has important drawbacks for at least two reasons. From a customer point of view, the environment and user requirements of systems may vary over time, yielding the need for adaptation of the planning and control procedures. From a systems’ deliverer point of view, an overall planning and control architecture that optimally exploits synergies between the different market sectors, and at the same time is flexible with respect to changing business parameters and objectives is highly valuable. An integral planning and control architecture should clearly describe the hierarchical framework of decisions to be taken at various levels, as well as the required information for decisions at each level, e.g., from overall workload planning to local traffic control. In this research, we identify synergies among the different sectors, and exploit these synergies to decompose AMHSs into functional modules that represent generic building blocks. Thereafter, we develop generic algorithms that achieve (near) optimal performance of the modules. As an example, we present a functional module from the Parcel & Postal sector. In this module, we study merge configurations of conveyor systems, and develop a generic priority-based real-time scheduling algorithm.KeywordsControl ArchitectureSpace UtilizationIntegral PlanningSpace AllocationWorkload BalanceThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

In the 300 mm AMHS with a connected loops layout, the vehicle can travel not just in one loop but all around the wide fab to execute tool-to-tool delivery. Determining how far the waiting FOUPs (WFs) or idle vehicles (IVs) should be considered before selecting a dispatching rule might make dispatching more efficient. A two-phase approach with simulation has been developed to assign the search range (SR) for studying this idea. The model of a simplified 300 mm AMHS was built using eM-Plant and Design Expert to design experiments and analyse the results. In phase I, the number of WF and IV in the system at the time of dispatching will affect the distance of the vehicle’s empty trip (DVemp). Further, the SR was assigned and evaluated based on the average and standard deviation of DVemp under different numbers of WF and IV in phase II. The results indicated that the SR significantly affects the performance, and a longer SR used in a light system is feasible; a shorter SR is applicable for a heavy system.

Wafer transport by Automated Material Handling Systems is essential in modern semiconductor front-end factories to efficiently supply the manufacturing tools. Most common are ceiling mounted Overhead Hoist Transport Systems that allow high throughput and fast delivery times. These systems are increasingly supplemented by various types of Automated Guided Vehicle Systems that enable high flexibility. This development requires new approaches for transport system dispatching.We denote such Automated Material Handling Systems that combine several types of transport systems as heterogeneous. Here, vehicles with different characteristics can execute either a complete or a part of a transport order. Until now, this task allocation process for the vehicles is executed based on predefined rules. A dynamic exchange between the transport systems based on the current system state is not yet considered.This paper presents a formalization of the related planning problem by a mathematical model for transport system scheduling. The approach is not suitable for real-time decision making. However, on the one hand, it may help to identify and evaluate the solution quality of heuristic approaches. On the other hand, it allows the evaluation of the effect of transfers.

Throughput analysis in an automated material handling system, such as an Automated Storage and Retrieval System(AS/RS), may be a complex problem. In the past, several approaches have been used for such an analysis. This paperpresents a combinatorial ap proach to evaluating the throughput performance of a mini-load system, with simulation as the primary method of investigation.

We present key approaches and methodologies from Intel's 300 mm factory integration efforts. The topics discussed in this paper are aimed at cross-functional optimization of fab layouts, automated material handling systems (AMHS) and operations. We discuss the key benefits of up front understanding of the interactions and investigate specific attributes in layouts and AMHS configurations where the integrated design/analysis is addressing cost-benefits and flexibility trade-offs needed in the high volume 300 mm factories.

This paper proposes an analytical model useful in the design of conveyor-based automated material handling systems (AMHS) to support semiconductor manufacturing. The objective is to correctly estimate the work-in-process on the conveyor and assess the system stability. The analysis approach is based on a queuing model, but takes into account details of the operation of the AMHS including turntables. A numerical example is provided to demonstrate and validate the queuing model over a wide range of operating scenarios. The results indicated that the analytical model estimates the expected work-in-process on the conveyor with reasonable accuracy.

The automated material handling system (AMHS) in semiconductor industry plays a vital role in reducing wafer cycle times and enhancing fabrication plant productivity. Vehicle allocation for the AMHS is a challenging task because of the complexity of the manufacturing process and the stochasticity introduced by the inherent variability of processing times. This situation is observed especially in 300 mm wafer fabrication plants where the AMHS comprises both interbay and intrabay systems that perform the timely deliveries. To address this issue, we use the optimal computing budget allocation (OCBA) and extend it by adding genetic algorithm (GA). Under this combined approached, the number of iterations of each alternative is determined by OCBA, and then the optimal solution in the domain of feasible solutions is found through GA. This research provides a useful reference for both scholars and practitioners seeking optimal allocation of lithographical resources and number of iterations using random parameters.

Automated material handling systems (AMHS) using conveyors have been recently proposed as a technology option for next generation wafer fabrication facilities. This technology seems to provide an increasing capacity for moving and storing wafers in a continuous flow transport environment. The goal of this research is to design and test conveyor-based AMHS configurations, which include turntables and storage areas near the processing equipment. Simulation models were developed in AutoMod to determine the best conveyor layout, with emphasis in comparing centralized versus distributed storage systems. The AMHS factors under study comprise the number, location, and capacity of the storage areas. Simulation results show that the distributed storage approach provides improved performance; however, these systems require more capital investment than that needed for the centralized storage approach.

The automated material handling systems today are playing ever more important roles in semiconductor/LCD fabrication facilities. Recently they became more flexible, intelligent, and speedy than in the past. The facilities have been fully automated because the size and weight of the unit loads used in the facilities were being increased beyond the limits that a human operator can handle. This research develops an efficient procedure to streamline the delivery of unit loads by the automated material handling system (AMHS). For this task, the research employs the event scheduling theory that has been successfully used in the both academia and industry. The developed procedure was applied to an actual LCD fabrication facility and improved the performance of an existing material handling system.

An automated manufacturing system (AMS) contains a number of versatile machines (or workstations), buffers, an automated material handling system (MHS), and is computer-controlled. An effective and flexible alternative for implementing MHS is to use automated guided vehicle (AGV) system. The deadlock issue in AMS is very important in its operation and has extensively been studied. The deadlock problems were separately treated for parts in production and transportation and many techniques were developed for each problem. However, such treatment does not take the advantage of the flexibility offered by multiple AGVs. In general, it is intractable to obtain maximally permissive control policy for either problem. Instead, this paper investigates these two problems in an integrated way. First we model an AGV system and part processing processes by resource-oriented Petri nets, respectively. Then the two models are integrated by using macro transitions. Based on the combined model, a novel control policy for deadlock avoidance is proposed. It is shown to be maximally permissive with computational complexity of O (n2) where n is the number of machines in AMS if the complexity for controlling the part transportation by AGVs is not considered. Thus, the complexity of deadlock avoidance for the whole system is bounded by the complexity in controlling the AGV system. An illustrative example shows its application and power.

A computationally effective analysis of the throughput performance of a closed-loop multi-vehicle automated material handling system (AMHS) used in highly automated 300 mm wafer fabrication facilities (fabs) is presented. A probabilistic model is developed, based on a detailed description of AMHS operations, and the system is analyzed as an extended Markov chain. The model represents vehicle operations on the closed-loop considering the possibility of vehicle blocking. This analysis provides essential parameters such as the vehicle blocking probabilities and the throughput capacity of the AMHS. A numerical example is analyzed and simulated using AutoMod to demonstrate and validate the stochastic model.

As the demand for ultra-clean areas in semiconductor fabrication has increased, the need for more efficient automated material handling system (AMHS) has become imperative. A diffusion tool is characterized by its long processing time of 6–12h, thus making it a form-batch manufacturing step. Out of necessity with this application, wafer lots are temporarily stored in stocker and are delivered to diffusion tools for processing when the Work-In-Process (WIP) level reaches 4–6 lots. As a result, diffusion tools wait for long wafer delivery time due to the current vehicle assignment method, thus greatly impacting transportation and production efficiency.