The Minitab® client is one of the world’s largest solar technology and renewable energy companies. For more than 20 years, it has been a leading manufacturer of solar PV modules and a provider of solar energy solutions. The company strives to deliver competitive, clean electricity to large energy buyers, and prides itself on its sustainability efforts.
Inconsistencies in manufacturing process due to unidentified factors
The Research and Development manager at a solar power company location in Shanghai, China had a problem. His team discovered that their manufacturing process was producing unacceptable levels of variability in the thickness of their silicon nitride films. They needed to control the thickness of the film to improve the solar cells’ surface energy efficiency levels and to meet internal standards. The ideal film thickness is between 77 – 87nm.
As illustrated above, from the initial process, the team found that the film thickness exceeded the upper limit as seen in the control chart, and there were issues with the process. Most of the data fell above the average, and the process capability was relatively low. As a result, the defect rate in the current process was very high.
The project team needed to analyze the reasons for the large variation of the film thickness and screen out the possible factors
FINDING THE ROOT CAUSE ALONG THE PRODUCTION LINE
The team wanted to determine whether the measurement system or process errors were causing the discrepancies in film thickness. They used the Topcon product testing platform, an industry-standard testing equipment, to measure the film thickness and refractive index.
The team then used Minitab’s Gage R&R measurement system analysis to see if the variability was caused by the measurement system itself. The analysis value of the measurement, Gage R&R, is 9.16, which is less than 10%, indicated that the Topcon measurement system met the requirements and was therefore not causing the issues.
NARROWING DOWN POSSIBILITIES WITH MINITAB BRAINSTORMING TOOLS
Still looking for possible influencing factors, the team used the fishbone diagram (also known as a cause-and-effect diagram) in Minitab Engage to brainstorm. Based on the fishbone diagram and the leader’s previous experience, the team identified two possible reasons why a malfunction in the equipment or a process issue may cause a variation of the silicon nitride film thickness:
- Frequency of which the butterfly valve, which is a valve that regulates the flow of fluid, in the coating equipment is cleaned
- Furnace temperature and position of the silicon in the furnace
First, the team investigated whether the butterfly valve in the coating equipment cleaning was an issue. The team created a two-sample t-test in Minitab to confirm whether the frequency of the butterfly valve cleaning makes a significant difference to the film thickness.
After the analysis, the team found that the p-value is less than 0.05. This means that there is a significant difference in the film thickness before and after the butterfly valve is cleaned, which means that the cleaning of the butterfly valve also has a significant effect on the film thickness. Eureka!
To test the second possible cause of varying film thickness, the project team used Minitab’s Regression Analysis to determine the temperature and position of the silicon in the furnace are significant factors.
Using Minitab’s capability analysis and control charts, the team identified the conditions needed to produce the ideal silicon nitride film thickness:
- Butterfly valves should be cleaned twice a day
- Determined the reasonable furnace temperatures. The process output showed the ideal temperature settings for the three respective furnace positions to be 500°C/932°F for the mouth of the furnace, 480°C/896°F for the middle, and 472°C/881°F for the bottom of the furnace
The team quickly found the ideal settings for the process parameters using statistical analysis. Inside Minitab’s regression model, the process parameters were adjusted, and the silicon nitride film thickness achieved a stable output. Using the adjusted parameter settings, the team implemented a verification analysis, and the process output showed a temperature settings of 500°C, 480°C, and 472°C for the three respective positions to be the most ideal. At the same time, the value of Cpk was greater than 1.67, and the process capability index was high, far exceeding the pre-set target.
Using Minitab’s powerful statistical analysis, the team was able to take very targeted and accurate measures. “In the past, we used continuous blind testing, and now we finally can quantitatively test our hypotheses. Minitab has provided us with tremendous help,” said the team manager. “During the various stages of discovery, testing the hypothesis, confirming and problem-solving, Minitab Statistical Software came through, not only shortening the time to solve problems but also improving team members’ confidence in their decisions.”
By understanding and being able to control the thickness of the silicon nitride film, the team helped improve the energy efficiency of the panels by 7%, from 81% to 88%. According to current production estimates of 6,048 single tube capacity per day, this measure has increased the companies’ revenues by CNY 650,000 (USD 101,400) annually.
Global Solar Technology and Renewable Energy Provider
- Founded in 2001
- Listed on NASDAQ since 2006
- HQ in Ontario, Canada
One of the world’s largest solar technology and renewable energy companies needed to control the thickness of the silicon nitride film to improve the solar cells’ surface energy efficiency levels after it was discovered that some of the films being produced were not meeting internal standards.
Minitab® Statistical Software
Using the powerful charting and analysis tools included in Minitab® Statistical Software and Minitab Engage™, the company was able to pinpoint and rectify the causes for the differences in film thickness.
- Improved the energy efficiency of the panels by 7%, from 81% to 88%
- Increased production boosted revenues by CNY 650,000 (USD 101,400) annually