Chia-Yu Hsu

Prof. Chia-Yu Hsu


Research Focus & Metrics

Academic Career Length

22Years (2005-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
29 August 2012 in Flexible Services and Manufacturing Journal
Verified
Journal Article
Closed access
PDF file
CCChen-Fu Chien
Chen-Fu Chien
CHChia-Yu Hsu
Chia-Yu Hsu
PCPei-Nong Chen
Pei-Nong Chen

23 July 2011 in Journal of Intelligent Manufacturing
Verified
Journal Article
Closed access
PDF file
CCChien Chen-Fu
Chien Chen-Fu
CHChia-Yu Hsu
Chia-Yu Hsu
CHChih-Wei Hsiao
Chih-Wei Hsiao

20 November 2013 in IEEE Transactions on Semiconductor Manufacturing
Verified
Journal Article
Closed access
CCChien Chen-Fu
Chien Chen-Fu
CHChia-Yu Hsu
Chia-Yu Hsu
As global competition continues to strengthen in semiconductor industry, semiconductor companies have to continuously advance manufacturing technology and improve productivity to maintain competitive advantages. Die cost is significantly influenced by wafer productivity that is determined by yield rate and the number of gross dies per wafer. However, little research has been done on design for manufacturing and productivity enhancement through increasing the gross die number per wafer and decreasing the required shot number for exposure. This paper aims to propose a novel approach to improve overall wafer effectiveness via data mining to generate the optimal IC feature designs that can bridge the gap between integrated circuit (IC) design and wafer fabrication by providing chip designer with the optimal IC feature size in the design phase to increase gross dies and reduce the required shots. An empirical study was conducted in a leading semiconductor company for validation. The results have shown that the proposed approach can effectively enhance wafer productivity. Indeed, the developed solution has been implemented in the company to provide desired IC features to IC designers to enhance overall wafer effectiveness.


1 January 2015 in Mathematical Problems in Engineering
Verified
Journal Article
Open access
PDF file
CHChia-Yu Hsu
Chia-Yu Hsu
Wafer bin map (WBM) represents specific defect pattern that provides information for diagnosing root causes of low yield in semiconductor manufacturing. In practice, most semiconductor engineers use subjective and time-consuming eyeball analysis to assess WBM patterns. Given shrinking feature sizes and increasing wafer sizes, various types of WBMs occur; thus, relying on human vision to judge defect patterns is complex, inconsistent, and unreliable. In this study, a clustering ensemble approach is proposed to bridge the gap, facilitating WBM pattern extraction and assisting engineer to recognize systematic defect patterns efficiently. The clustering ensemble approach not only generates diverse clusters in data space, but also integrates them in label space. First, the mountain function is used to transform data by using pattern density. Subsequently,k-means and particle swarm optimization (PSO) clustering algorithms are used to generate diversity partitions and various label results. Finally, the adaptive response theory (ART) neural network is used to attain consensus partitions and integration. An experiment was conducted to evaluate the effectiveness of proposed WBMs clustering ensemble approach. Several criterions in terms of sum of squared error, precision, recall, andF-measure were used for evaluating clustering results. The numerical results showed that the proposed approach outperforms the other individual clustering algorithm.

30 March 2015 in Journal of Industrial and Production Engineering
Verified
Journal Article
Closed access
CHChia-Yu Hsu
Chia-Yu Hsu
SLSheng-Chiao Lin
Sheng-Chiao Lin
CCChen-Fu Chien
Chen-Fu Chien
Variations in customer demands directly affect the total quantity of customer orders processed in semiconductor fabrication. This causes variability in wafer start input, affecting the work-in-process bubble and cycle time. The finished goods stored in fab warehouse are called semiconductor vendor-managed inventory. The increase of inventory leads to low inventory turnover ratio. In this study, a back-propagation neural network model with a distributed lag structure is developed to predict customer demands based on customer order behaviors, and to extract useful information for supporting the production plan for VMI based on uncertain demand and market fluctuation. An empirical study was conducted at a leading fab to validate the method. The results have shown the practical viability of the proposed approach. The derived empirical rules can assist decision-makers in making timely production decisions, given various order situations, ensuring that adequate fab utilization and cycle times are maintained.