That day, RGA reports spiked dramatically, and our team lead was furious
That afternoon, several machines on the production line suddenly stopped. As soon as the RGA (Residual Gas Analyzer) report came out, it was alarming—the values for several critical gases were completely off the charts. The boss rushed in, looking furious, and asked, "What's wrong with this batch of material? Didn't the supplier say it was fine?" Our team lead was completely overwhelmed, because this batch of goods had been in the factory for over a week, and now problems were emerging, which meant several previous batches of products might also have potential risks. At this point, everyone's attention turned to Quality Assurance (QA), demanding to know what they had actually done during the raw material incoming inspection.
Where's the problem? Relying solely on the COA is certainly insufficient!
To be frank, often we rely too much on the supplier's COA (Certificate of Analysis). COAs are important, of course, but they merely represent what the supplier "tells you" about their material's qualification. When our own materials arrive, especially chemicals and gases, how do you know they truly meet your process requirements? Many contaminants, even at ppm (parts per million) levels, can lead to severe yield losses. For example, metal ions or specific organic compounds might not be detectable with general FTIR or GC-MS, not to mention that some brands of equipment may not even have these analytical methods established. At this point, more precise and sensitive analytical techniques are needed to help you guard against these invisible devils.
How to do it in practice? ICP-MS and XRF are essential tools!
The two most commonly used are ICP-MS (Inductively Coupled Plasma Mass Spectrometry) and XRF (X-ray Fluorescence).
- ICP-MS: The powerful aspect of ICP-MS is its extremely high sensitivity, capable of detecting trace metal impurities at ppt (parts per trillion) levels. For example, if the copper (Cu) ion content in your etchant exceeds 50 ppt, it could affect your etching rate or line width uniformity. We would first set a strict specification with the supplier, such as "Cu < 20 ppt". For each incoming batch, we would take a sample and send it for ICP-MS testing. If the measurement comes back as 35 ppt, even if the supplier's COA states it's qualified, we would still have to reject the shipment. ICP-MS can also be used for post-process residue analysis to check if your cleaning was effective.
- XRF: Compared to ICP-MS, which measures liquids or powders, XRF is more commonly used to measure the elemental composition of solid surfaces or thin films. Its advantage is being "non-destructive," allowing direct analysis of your wafers and packaging materials, and it's fast. For instance, if, after wire bonding, XRF detects a tin content exceeding 0.5% in the gold ball, the solder joint strength might be compromised, leading to decreased reliability. We once had a case where the supplier changed a batch of solder balls for BGA (Ball Grid Array) packaging, and XRF analysis showed abnormally high lead content, which almost led to the scrapping of an entire batch of goods. Therefore, the key is that both analytical methods have their applicable scenarios, and using them correctly will maximize their benefits.
The most common pitfall: Establishing the correct comparison baseline is crucial!
The most absurd thing I've encountered is directly comparing data from ICP-MS measurements with data from XRF. This is like comparing apples and oranges! ICP-MS measures "total content," analyzing all elements after decomposing your sample, whereas XRF measures "surface elements" and has depth limitations. Furthermore, is your "standard reference material" sufficiently similar to your analyte? If your standard reference material's matrix differs too much from the actual sample, the measured result might only be a reference value. A calculated Cpk (Process Capability Index) of 1.08 might seem to just meet the target, but actually the DPMO (Defects Per Million Opportunities) is still 6210, which means the defect rate is still alarmingly high. Frankly, many times the QA department, to save costs, uses outdated or un-updated standard reference materials, or, for convenience, directly uses the supplier's standard reference materials. These practices are all very dangerous.
One thing you can do today
Go back and check the calibration records for your in-house ICP-MS and XRF standard reference materials, and ensure they are within their validity period.