Applied Mechanics and Materials Vols. 10-12 (2008) pp 248-252 online at http://www.scientific.net © (2008) Trans Tech Publications, Switzerland Online available since 2007/Dec/06
Manufacturing Process Flow Reasoning of Micro Device Based on Ontology Z. Liu1,2,a, P.Y. Jiang1,2,b and S.Y. Zhou1,2,c 1
State Key laboratory for Manufacturing Systems Engineering
2
CAD/CAM Institute, School of Mechanical Engineering, Xi’an Jiaotong University, China a
[email protected],
[email protected],
[email protected]
Keywords: Feature modeling, Process reasoning, Ontology, Expert system
Abstract. Based on feature modeling technology, a method of manufacturing process flow reasoning of micro device is presented to support the top-down design flow. The knowledge of micro device is hierarchically organized to construct the ontology-based knowledge database that is combined with the expert system by an interface. With the support of process databases, the manufacturing process flow is derived by process reasoning. Introduction Nowadays the micro device is more and more complex. However, the design flow of current design tools is still from the process designing to 3D model constructing. It is unintuitive to designers. As an advanced theory, the Top-down design method starts at the system level and makes designers pay less attention to the fabricating issues [1]. The more reasonable design flow can be characterized briefly as “function-to-shape-to-process”. Therefore, problem-solving in the sector of converting the design model to the manufacturing process flow is the guarantee of good manufacturability. To achieve this goal, some researches have been underway. For a function-oriented modeling process, feature model was presented [2]. Besides the above feature modeling, the “forward” and “inverse” design flow problems were also studied as key issues [3]. Another approach accomplished the mask creation by investigating the vertical topology of the model [4]. Especially in bulk micromachining, the process reasoning technology to support top-down design flow is still need to study further. An essential difficulty lies in the standardization of the processes. Some achievements have been registered like the standard processes about bonding and sacrificial layer technology [5]. Along with the standardization of processes, the expert system is applied to manufacturing flow reasoning. Ontology has a bright future for application in information sharing, system integration and software development based on knowledge, etc. Especially for the hierarchical description of the knowledge in micro device manufacturing, it has advantage. Based on feature modeling technology, this paper concentrates on manufacturing process flow reasoning of micro device with a hierarchically organized knowledge database by ontology. Currently, it is used in bulk micromachining. Feature Modeling Technology in Bulk Micromachining The design model of micro device is constructed with design features. First of all, the manufacturing features are obtained to confirm the necessary processes. On account of the characteristics of the top-down design flow, the information framework of micro device is three levels as shown in Fig. 1. The functional model is constructed with functional features at the first level. The mapping process of function/geometric features can be formalized by adopting the macro script language. The 3D design model is obtained at the second level. It is CSG-represented and made up of geometric features.
All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of the publisher: Trans Tech Publications Ltd, Switzerland, www.ttp.net. (ID: 130.203.133.34-16/04/08,09:00:55)
Applied Mechanics and Materials Vols. 10-12
Functional feature level
Geometric feature level Cuboid
Functional feature base on defined “script”
Geometric feature
System level modeling based on bonds graph
249
Manufacturing feature level Substrate feature
Cylinder
Layer feature
Extrude
Bonding feature
Draft
Anisotropic etching feature
Blind hole Through hole
Masks based hole feature
Shell
Doping feature
CSG compound
Compound feature
Mirror
Circuit and other feature
Pattern
Fig.1 The information framework of micro device
By manufacturing features mapping, the process model is derived at the third level. Besides the features shown, it includes some non-geometric attributes like material and manufacturing accuracy, etc. Manufacturing features information is the basis for process reasoning. The Framework of the Reasoning System After feature modeling, the mapping relations among different levels are confirmed. The reasoning process is on the basis of manufacturing features and carried out in the expert system. The framework of the reasoning system is shown in Fig. 2. Manufacturing Design features features Df Df Mf Mf 3D design model Features mapping
Process model Material database Equipment database
Features extraction
Knowledge conversion
Expert system
Ontology tools
Manufacturing process flow
Process database
Fig.2 The framework of the reasoning system
Firstly, the 3D model constructed with design features is converted to the process model by features mapping. Then, the hierarchically organized knowledge is constructed by the ontology tools. Next, it is transmitted into the expert system through the interface. The reasoning rules are also defined. Finally, the information is extracted from the manufacturing features to produce the reasoning sell, which perform the reasoning process in the expert system. As a result, the detailed manufacturing process flow is obtained with support of the material and equipment database The main factors are listed as follows: (1) The transition of the design model into the process model. This transmitting procedure is performed on the basis of the relations between features described in Fig. 1. (2) The construction of the standardized process database. As the requirement of top-down design method, standardization is the trend. The process database is constructed base on the representative processes in bulk micromachining. Some descriptions of the processes are listed in Table 1.
250
e-Engineering & Digital Enterprise Technology
Table 1 Some descriptions of the processes in the database Index I
II
III …
Brief description This process is based on wet etching method and electrostatic bonding of Si ／ Glass technology. It is applied to the manufacturing of micro accelerometer, micro valve and micro pump, etc. This process is based on Si／Glass aligned bonding technology, ICP etching for high aspect ratio and wafer thinning. It is applied to the manufacturing of micro gyroscope and lateral capacitive silicon micro accelerometer, .etc. This process is based on silicon direct bonding and deep reactive ion etching technology. It is applied to the manufacturing of electrostatic micro resonator and micro thermal actuator, etc. …
(3) The hierarchical organization of the knowledge of micro device and the combination of the expert system with ontology. It is the basis for the reasoning procedure and will be discussed in the next section. Process Reasoning Based on Ontology In expert system, the knowledge and the rules are separated so as to make the knowledge reusable and sharable. At the same time, the hierarchical relations among the knowledge are difficult to express with current storing manner. For a more reasonable organizing manner, ontology is used to construct the knowledge database. With this method, the relations among the conceptions are fixed. Further more, the knowledge is hierarchically organized to construct a closer relation with the feature modeling. The advantages of the ontology for the process reasoning are presented as follows: (1) The structure of the knowledge is clearer by ontology. It is reusable to avoid the iterative analysis of domain knowledge. (2) Ontology is the basis for the mutual operation and sharing of knowledge. By certain interface, they are transmitted in the domain. (3) It improves the standardized description of the processes. The micro device modeling is combined with ontology. The main factors of ontoloty are class, slot, instance and axiom. Class is a concept. For bulk micro machining, wet etching and dry etching are all classes. They are subclassse of etching that is a high level class. Slot indicates the attribute and relation of classes. For example, the etching class has the attributes of etchant and equipment, etc. The relations among other factors are constructed similarly. Protégé is adopted as the ontology tool, which is developed by Stanford Medical Informatics. The expert system utilizes Jess (Java Expert System Shell) as the rule engine. JessTab (a plugin used in Protégé) is adopted as the interface between them to implement the transmission of knowledge. The construction of the hierarchical knowledge with ontology is described as follow steps. Step 1: The domain of ontology is confirmed. It includes the feature and the process information of micro device. Step 2: The class and property of ontology are defined. For example, “Base” and “Wet-Etching” are classes. Accordingly, “Base Material” and “Slot Angle” are properties. Step 3: The hierarchical framework of the classes is constructed. It is shown in Fig. 3. The process reasoning by the expert system is performed with following steps: Step 1: The knowledge based on ontology is imported into the expert system by an interface.
Applied Mechanics and Materials Vols. 10-12
251
Step 2: The reasoning rules are defined and imported into the rule database, which are on the basis of the processes commonly used in bulk micromachining. For example, the and-or graph and rules related to the process flow I is shown in Fig. 4. Features
Shape features
Main feature
Tech features Subclass of Managing features
Array feature Aux Feature
Hole Slot Doping
Processes Bond Subclass of Etching
Substrata Wet Etching Dry Etching
ICP RIE
Fig.3 The hierarchical organization of the classes Process Ⅰ Wet etching Slot feature α =54.7°
SFB Bonding Si-Si
Rules in Jess (defrule wetEtchRule?f (unmake-instance ?f) (make-instance wetEtch1 of WetEtch) (printout t "wetetch" crlf)
Fig.4 The and-or graph and relevant rules about the process flow I
Step 3: The manufacturing features are extracted and converted into the facts. They are imported into the system as the reasoning cells. The rules are activated to produce new facts. Step 4: The facts are organized to generate the detailed description of the manufacturing process flow with support of the process databases. Design model of micro silicon pressure sensor
Process model deriving
Manufacturing features mapping
Coordinate management module
Ognized process flow
Protégé based manufacturing process flow reasoning
Fig.5 The reasoning process for the micro silicon pressure sensor
252
e-Engineering & Digital Enterprise Technology
Implementation Taking the micro silicon pressure sensor as example, Fig. 5 illustrates the implementation of this method. Firstly, the feature based model is constructed. Then, process model is derived by features mapping procedure. Next, with support of the knowledge database and process database, the process reasoning is carried out by the expert system. The snapshot is about the reasoning of the compound feature. Finally, the detailed manufacturing process flow is obtained. Conclusion With hierarchical organization of knowledge of micro device by ontology, a method of manufacturing process flow reasoning by expert system is presented. Based on feature modeling of three levels, the top-down design flow is well supported. By updating the ontology-based knowledge database, the reasoning system becomes more powerful to provide better manufacturability of the design model. Future work will be emphasized on improving the processes database to make the reasoning system more robust. References [1] Michael S. McCorquodale, Fadi H. Gebara, Keith L. Kraver and et al: Proceedings of the Design, Automation and Test in Europe Conference and Exhibition. IEEE Computer Society , Los Alamitos, CA, USA. Vol. 1 (2003), pp. 292-296. [2] Feng Gao, Y Steve Hong and Radha Sarma: Proceedings of ASME Design Engineering Technical Conferences, Chicago, USA (2003). [3] Venkataraman Ananthakrishnan, Radha Sarma and G K Ananthasuresh: Journal of micromechanics and microengineering, Vol.13 (2003), pp.927－941. [4] R Schiek and R Schmidt: Microsystem technologies, Vol.12 (2006), pp.204-207. [5] Y.Y. Wang, G.Y. Wu, Y.L. Hao and et al: Acta Electronica Sinica,Vol.30 (2002), pp. 1577-1584.
