Benchmarking software for Machine Health Management

Introduction
The ASM Centre of Competency, located in the Netherlands, is responsible for identifying new high-tech innovation opportunities for the full global ASM Pacific Technology product portfolio [1]. The equipment portfolio of ASM Pacific Technology covers the complete back-end section of the semiconductor equipment market, and offers comprehensive assembly and advanced packaging solutions for its customers in the fields of optoelectronics, electronics, solar energy, automotive and other segments.

Challenges
The semiconductor industry is one of the most technology-evolving and capital-intensive market sectors [2-5]. To strengthen a competitive position in the semiconductor industry, the key challenges a manufacturing plant has to face are the increase of production throughput and yield. Monitoring health conditions of the semiconductor machines is therefore necessary to secure the required product quality. Furthermore, effective prediction of the failures of the semicon equipment is a way to decrease a machine downtime, improve productivity, and reduce production costs and repairing time.
The semiconductor manufacturing equipment is equipped with different sensors for real-time machine control and equipment condition monitoring. Data from these sensors can be used to detect deviations of the machine performance from the nominal one, in order to avoid degradations in the product quality and losses in the production yield. The same data is useful in the process of discovering when the equipment needs a maintenance to avoid machine failures and unscheduled breaks in the production.
To achieve automated machine condition monitoring and predictive maintenance functionalities, we need to address the following challenges [2-7]:

  • make appropriate selection of sensors for machine condition monitoring and choose optimal locations where these sensors have to be installed,
  • develop algorithms for machine health management, starting from extraction of machine health condition indicators (CI’s) out of the sensor signals, up to design of predictive methods for forecasting equipment behaviour and health diagnostics,
  • optimally schedule maintenance operations according to the equipment conditions.

Company ASM Pacific Technology is developing proprietary software for machine health management, which is customized for applications and deployment on own semicon machines. The question how well this software performs in comparison with commercial off-the-shelf software packages for condition monitoring, health diagnostics and predictive maintenance. Therefore, it is instructive to benchmark the proprietary software of ASM Pacific Technology with commercial software tools in terms of available functionalities, calculation efficiency, user friendliness, level of automation, reliability and accuracy of forecasting machine behaviour, etc.

Internship Assignment
In this project, we will benchmark the proprietary software of ASM Pacific Technology with at least one commercial software package for machine health management. An interesting off-the-shelf candidate is the LabVIEW Analytics and Machine Learning Toolkit [8]. If time would allow, we may also include the Predictive Maintenance Toolbox of Matlab [9] in benchmarking.
The Center of Competency has a large collection of data measured on many ASMPT machines operating in the field. These data involve standard readings from motion and current sensors. The idea is to process these data using the company proprietary software and commercial tools, such as to benchmark functionalities, performance and autonomy of the company software with respect to the off-the-shelf solutions. Furthermore, the Center of Competency has several models of a wirebonder machine that can be used for simulation of machine behaviours that might not be captured in the available measurements. If needed, synthetic data could be generated in simulations to make more complete evaluation of the software capabilities.
The investigation objectives are:

  • literature study on methods for machine condition monitoring, health diagnostics and predictive maintenance,
  • learn how to perform machine health management using the proprietary software of ASM Pacific Technology and LabVIEW Analytics and Machine Learning Toolkit,
  • process available machine data using the proprietary and commercial software packages and critically evaluate advantages and limitations of these software tools,
  • if needed, generate synthetic data by simulation of machine behaviours that are not captured in the existing measurements but could be used for evaluation of some specific software functionalities,
  • eventually, include Predictive Maintenance Toolbox of Matlab in the benchmarking,
  • identify pro’s and con’s of the proprietary software for machine health management and give recommendations for its eventual improvements.

The results of this project will be disseminated via the internship report and presentations at the knowledge institution and Center of Competency. The generated knowledge can become a valuable academic know-how and drive technological advantages against other producers of the semiconductor equipment.

 

For more information about this project, please contact dkostic@alsi.asmpt.com or call +31 24 204 2827.

References
[1] ASM Center of Competency, https://alsi.asmpt.com
[2] S. Munirathinam and B. Ramadoss, “Predictive Models for Equipment Fault Detection in the Semiconductor Manufacturing Process,” IACSIT International Journal of Engineering and Technology, Vol. 8, No. 4, pp. 273-285, 2016.
[3] Y. Liu, Predictive modeling for intelligent maintenance in complex semiconductor manufacturing processes, PhD thesis, The University of Michigan, 2008.
[4] J. Iskandar, J. Moyne, K. Subrahmanyam, P. Hawkins, and M. Armacos: “Predictive Maintenance in semiconductor manufacturing,” 26th Annual SEMI Advanced Semiconductor Manufacturing Conference, Saratoga Springs, NY, USA, 2015.
[5] P. Scheibelhofer, D. Gleispach, G. Hayderer, and E. Stadlober: “A Methodology for Predictive Maintenance in Semiconductor Manufacturing,” Austrian Journal of Statistics, Vol. 41, No. 3, pp. 161–173, 2012.
[6] A. Khanore, Predictive Maintenance Algorithms Development for a Wire Bonder, DCT internship report, 2019.
[7] R. van Bussel: Obtaining Machine Health Condition Indicators Using Wavelet Transforms, DCT internship report, 2019.
[8] LabVIEW Analytics and Machine Learning Toolkit,
http://www.ni.com/pdf/manuals/377061a.html
[9] Predictive Maintenance Toolbox of Matlab, https://nl.mathworks.com/products/predictive-maintenance.html