Paul B. Chou

Automatic defect classification for semiconductor manufacturing

Abstract.Visual defect inspection and classification are important parts of most manufacturing processes in the semiconductor and electronics industries. Defect classification provides relevant information to correct process problems, thereby enhancing the yield and quality of the product. This paper describes an automated defect classification (ADC) system that classifies defects on semiconductor chips at various manufacturing steps. The ADC system uses a golden template method for defect re-detection, and measures several features of the defect, such as size, shape, location and color. A rule-based system classifies the defects into pre-defined categories that are learnt from training samples. The system has been deployed in the IBM Burlington 16 M DRAM manufacturing line for more than a year. The system has examined over 100 000 defects, and has met the design criteria of over 80% classification rate and 80% classification accuracy. Issues involving system design tradeoff, implementation, performance, and deployment are closely examined.

The Design and Implementation of a Smart Building Control System

A significant proportion of total worldwide energy is consumed by buildings. For example, buildings in the US account for about 40 percent of total energy consumption and greenhouse gas emission. Making buildings more energy-efficient is an important step to reduce our energy consumption and carbon emission in the combat with global climate change. Broad participation by consumers, business owners, and governments is required to continuously improve on energy efficiency for new and existing buildings and to achieve the global greenhouse gas emission reduction objectives. This paper provides a software system perspective of improving energy efficiency for buildings. It proposes an architecture that allows for phased investments in technologies to capture the returns from energy savings in various use cases. In addition, it addresses the needs and objectives of different stakeholders, including owners, operators, users, and utility providers. A proof-of-concept implementation of the architecture is used to demonstrate the support for building-wide energy conservation policies using real-time energy pricing and individual occupants’ locations and preferences. It shows that the proposed architecture enables fine-grained building control and reduces energy consumption while maximizing its occupants’ comfort.

Principal Component Analysis of Optical Emission Spectroscopy and Mass Spectrometry: Application to Reactive Ion Etch Process Parameter Estimation Using Neural Networks

We report on a simple technique that characterizes the effect of process parameters (i.e., pressure, RF power, and gas mixture) on the optical emission and mass spectra of CHFJO2 plasma. This technique is sensitive to changes in chamber contamination levels (e.g., formation of Teflon-like thin-film), and appears to be a promising tool for real-time monitoring and control of reactive ion etching. Through principal component analysis, we observe that 99% of the variance in the more than 1100 optical and mass spectra channels are accounted for by the first four principal components of each sensor. Projection of the mass spectrum on its principal components suggests a strong linear relationship with respect to chamber pressure. This representation also shows that the effect of changes in thin-film levels, gas mixture, and RF power on the mass spectrum is complicated, but predictable. To model the nonlinear relationship between these process parameters and the principal component projections, a feedforward, multi-layered neural network is trained and is shown to be able to predict all process parameters from either the mass or the optical spectrum. The projections of the optical emission spectrum on its principal components suggest that optical emission spectroscopy is much more sensitive to changes in RF power than the mass spectrum, as measured by the residual gas analyzer. Model performance can be significantly improved if both the optical and mass spectrum projections are used (so called sensor fusion). Our analysis indicates that accurate estimates of process parameters and chamber conditions can be made with relatively simple neural network models which fuse the principal components of the measured optical emission and mass spectra. In the reactive ion etching (RIE) process, plasma characteristics depend on many parameters; some of these parameter values are set by the tool operator, e.g., chamber pressure, RF power, and gas flow, while others are internal to the condition of the chamber, e.g., thin-film thickness on the chamber walls, or the amount of material etched. Plasma characteristics can be observed using in situ measurements, e.g., via optical emission spectroscopy (OES) or residual gas analysis (RGA). How these measurements can be used to estimate the process parameters is the question

Identifying prospective customers

We describe data mining techniques designed to address the problem of selecting prospective customers from a large pool of candidates. These techniques cover a number of different scenarios, namely whether the marketing researchers have demographic information on the current customers, or the general market population, or people with propensity to become customers We also present a novel approach to the problem by exploiting the availability of a data sample from the general market population. Finally, we describe an on-line lead management and delivery system that uses the mining approach described in this paper for insurance agents to obtain qualified customer leads.

An Ecosystem Approach for Healthcare Services Cloud

Patient-centric healthcare and evidence-based medicine with the emphasis on prevention and wellness promise to deliver better and more affordable healthcare. At minimal, they require health related information to be shared among a community including patients, providers, payers, and regulators. It is important for IT systems to facilitate information sharing within such communities. Furthermore, we argue that it is highly valuable to develop IT technologies that can foster sustainable healthcare ecosystems for collaborative, coordinated healthcare delivery. The emerging cloud computing appears well-suited to meet the demand of a broad set of health service scenarios. In particular, the concept of shared infrastructure and services provides the foundation for supporting healthcare service ecosystems. This paper proposes an ecosystem approach to identify high-level requirements for cloud computing technologies to provide hosting environments for sustainable healthcare ecosystems. We draw the lessons and principles from the sustainable ecological ecosystems, review some of the existing IT-enabled healthcare ecosystems, and provide our view on the imperatives for cloud computing research to support future healthcare IT needs.

Automatic defect classification for integrated circuits

While initial detection of defects is the most critical function of inspection, automatic classification of detected defects is becoming increasingly desirable. The key to better process control is reliable process measurement. The classification of defects provides valuable process diagnosis information. The hope is that machines can perform this task more reliably than humans. However, there are many problems in automating defect classification, and many of these are related to the central problems in artificial intelligence, such as knowledge representation, inferencing, and dealing with uncertainty. In this paper we pay special attention to the issues arising in the Automatic Defect Classification (ADC) of integrated circuits. We first discuss technical and system requirements, followed by an outline of the technical challenges to be overcome to develop flexible and powerful ACD tools which can be quickly customized on a user level for diverse applications.

A model-driven approach to RFID application programming and infrastructure management

This paper describes a model-driven methodology to provide a systematic means of programming RFID applications and managing a distributed RFID infrastructure. The centerpiece of the approach is a model that captures the device infrastructure, business operation structure, and their relationship. Building on the model, a management system can be designed to support various management tasks effectively, such as the enrollment, provisioning, and monitoring of devices. Similarly an application framework can be created for solution developers to create distributed RFID applications. Applications can be written against a business operation model and deployed on an evolving device infrastructure. A prototype shows that the approach facilitates the management and programming of an RFID infrastructure

DRIVE: a tool for developing, deploying, and managing distributed sensor and actuator applications

This paper introduces Distributed Responsive Infrastructure-Virtualization Environment (DRIVE), a tool that provides both an integrated development environment (IDE) and an execution environment and thus supports the entire life cycle of sensor/actuator applications. Developers are only responsible for implementing the core event-handling logic, whereas DRIVE generates the necessary code for message passing and invocation, thus reducing the development skills required. The development methodology, which is component based and model driven, separates the solution model, which captures the business logic, from the deployment model, which reflects the physical computing infrastructure. This allows the administrators to configure and deploy applications on various infrastructure topologies. To illustrate the benefits of DRIVE, we present an example application, dock-door receiving, and show the ways in which DRIVE supports the modeling and development of the application logic and the multiphase deployment of the resulting application in a production environment.

Extending SOA/MDD to Sensors and Actuators for Sense-and-Respond Business Processes

While the practice of service oriented architecture (SOA) and model-driven development (MDD) has brought efficiency gains to software development, building responsive, scalable business applications enabled by distributed sensor and actuator (S&A) infrastructure remains challenging, often involving extensive effort and skill. The results are typically one-of-a-kind, difficult to replicate or modify for different environments, and brittle in the face of changes to the physical plant and equipment. This paper examines the challenges presented by distributed S&A infrastructure in the context of business information systems, and introduces an architecture that applies SOA/MDD methodologies to the modeling, development, deployment, and management of S&A systems. In particular, the architecture focuses on the separation of concerns between logical application development and infrastructure capability management, and the support of component assembly with an event-based programming model. The paper then briefly describes a prototype implementation of the architecture, including integrated tooling and runtime support, and illustrates its role relative to existing business process modeling, integration and execution environments with examples based on industry applications.Although some relatively mature products and systems about Dynamic service composition are proposed, their manipulations are mostly not trivial and intuitive for the common users. Therefore, this paper proposes a dataflow driven dynamic service composition architecture aiming at astronomy applications, to simplify the requirements for users to compose a service. The architecture utilizes Dataflow Driven conception to discover an Original Topology Graph, which includes the final composite service with the shortest length and good parallel structure, and uses a Converse Composition algorithm to ascertain the final result quickly. This architecture is applied in the PMGrid prototype for astronomy data processing, and the experiment results show the architecture satisfies the parallelism and asynchrony in distributed system, avoids manual errors, increases the service configuring efficiency, provides the composite service with better structure to increase the executing parallelism, and finally improve the scientific discovery.

A framework for managing the solution life cycle of event-driven pervasive applications

Event-driven, embedded applications that embody the composition of many disparate components are emerging as an important class of pervasive applications. For such applications, realizing solutions often requires a breadth of expertise. Consequently, managing the solution life cycle can be a very complex, time-intensive process. In this paper, we present a framework that eases the complexity of managing the life cycle of event-driven, pervasive solutions. We call this framework Rapid Integrated Solution Enablement or RISE. Component composition and software reuse are two central concepts of RISE, where solutions are graphically composed from reusable components using a visual editor. We describe the RISE architecture and discuss an initial prototype implementation that leverages open source technologies, such as Eclipse. Additionally, we illustrate the efficacy of RISE with an example solution for RFID supply chain logistics