Everything starts with a question.
This one came from Phil Marcoux who mentioned in passing that a recent discussion about component exposure to x-rays prompted him to wonder whether there would be any lingering effects. Could it be possible that excessive exposure could be the root of otherwise unexplainable field failures?
“Is this an issue that has ever been brought to your attention?” Marcoux wanted to know. “If so, did it appear to be a real issue?”
First a word about who was doing the asking. Marcoux’s is a name that should be right there on Mount Assembly for his contributions to electronics manufacturing, not the least of which was introducing SMT to the US.
Marcoux’s partner in this “casual” chat was Silicon Valley engineer Tom Clifford. If Marcoux is the Father of SMT (as IPC has called him), Clifford is no worse than a first cousin on the materials side. His history dates to the St. Louis Project Gemini in the early 1960s with McDonnell Aircraft, then Monsanto, Raychem, and United Technologies in R&D and manufacturing before landing at Lockheed Martin, where he studied aerospace electronics and materials for 17 years.
The notion that exposure to radiation could pose issues is not new. While at Lockheed, Clifford investigated whether certain ICs suffered from exposure during x-ray inspection. Clifford was spurred on by the introduction of BGAs, which forced new ways of inspection, given that the leads are hidden from traditional human or machine view. His memory of the study is clear and focused: “I was spooked by our shop folks’ newfound fascination with finding voids and de-wets and solder coverage in BGA joints. Looking and looking and cranking up the power and more looking and asking Quality Control to stare at the x-ray monitor for endless minutes, and ‘What did they think?’
“The problem is that nobody on any shop floor knows what IC technology is hiding inside the little black squares on a circuit board. It could be robust; it could be very, very sensitive and is dying right before your eyes. The solder joint is fine. The IC now is mostly dead.”
When something piques the interest of a couple of gurus like Marcoux and Clifford, we lesser types are best to sit up and take notice.
My short answer to Marcoux was no and, upon asking around, maybe. But Clifford was already on the case. He performed a literature search, which turned up work by JPL, Spansion, and Creative Electron, all of which suggested IC vulnerabilities. It was about that point that Dr. Alan Rae, the former Cookson scientist, chimed in, noting that dose intensity could have a bearing on the results.
“There are lots of web references expressing concern about damage to flash memory in particular. X-rays are ionizing radiation; they can raise the energy levels of electrons in atoms (then re-emitted as heat) or in extreme cases strip electrons off atoms to break bonds.” Citing work by Nordson, Dr. Rae concluded, “Care must be taken at higher magnifications when the beam is concentrated, but in general the conclusion is that x-ray inspection does not pose a significant risk to most components.”
It’s that “most components” disclaimer that I found troubling. Furthermore, as Clifford notes, there are no known standards for the exposure time of various components to x-rays, which in practice ranges from seconds to minutes.
Dr. Gary Lum is an engineering fellow at Lockheed and a former colleague of Clifford’s. Most of Dr. Lum’s research is proprietary, but he allows that when it comes to x-rays, there are pros and cons to smaller parts.
“As technology shrinks, gate oxides are thinner. Overall hardness to TID [total ionizing dose] is better for the transistors. However, there are areas of concern. Gate oxides are thinner with less charge trapping, but the field oxide is still thick, so the isolation area is weak in TID. Discrete devices are okay, but if a designer designs in the low current 10µA range where TID effects are large, then there’s a problem. We could have micro dosing, too, if devices become very small. People may not be aware that in batch quantity, leak testing using Kr-85 with betas can degrade a batch of bipolar transistor gains.”
Clifford suggested the time might be right to develop a new list of vulnerabilities and perhaps collect anecdotal info on field failures and possible failure analysis linking that to radiation exposure. “If reports of unexplained field failures are coming in, we certainly must remind folks out there. Everybody is looking hard at BGA solder joints, using their fancy new x-ray cookers!”
I volunteered that perhaps this could be a crowdsource kind of project. I would write an editorial to raise attention, and ask for volunteers to send samples to Dr. Michael Pecht. Once Dr. Pecht was done testing, we could put something together to publish.
Pecht, the renowned reliability engineer and professor who runs the University of Maryland Center for Advanced Life Cycle Engineering (CALCE), replied with a cheerful yes. While tests he ran some 15 years ago showed no problems, he volunteered to run new ones on any samples supplied.
Three members of the CALCE team then joined Clifford on a conference call in mid May to outline a strategy. Bhanu Sood, director of CALCE’s Test Services and Failure Analysis Laboratory, synopsized the questions and outlined the current and future plans.
“Studies in literature have shown a correlation between enhanced doses of x-ray radiation and potentially irreparable damage and failure of semiconductor electronic components. Since accumulated dosage is a function of radiation exposure over time, certain radiological techniques that require prolonged exposure time will increase accumulated dosage. These include detailed x-ray inspection, computer tomography and other forms of digital radiography.
“Not all damage will be immediately noticeable. During typical practice, a typical semiconductor component radiological inspection system may not produce large enough radiation doses to cause a semiconductor device to fail, but may induce certain latent defects in the semiconductor dice. These latent defects present themselves as intermittent loss of programmed data, bit flips and leakage.”
Clifford says he has a bad feeling that radiation exposure might be similar to the ESD-damage gorilla that was lurking unknown in microelectronics some 40 years ago. Lest anyone think Clifford is getting ahead of himself, he points out that we simply don’t know the effects, if any, at this point. “The failure mechanisms – immediate, latent, cumulative, long-term, interrelated – are complex and need serious exploration. We have no solid evidence yet that field failures have occurred that were either unknowable or linked to enthusiastic x-ray inspection. The next step is to identify methods to apprise the industry of potential risks involved.” Following that, a thorough review of theoretical vulnerability of ICs, as well as a search for reports of unexplained/premature field failures, is in order.
In all likelihood, few assembly engineers are knowledgeable of all the x-ray inspection components may have endured early in the supply chain. It will take some doing to coordinate the semi fabs and appropriate OEMs and EMS companies, but there’s willingness on the test services side to perform the heavy lifting (read: testing and analysis).
So here’s where your humble editor comes in. We are looking for x-ray equipment OEMs, semiconductor manufacturers and EMS providers willing to participate in this discussion. Readers, please consider any contribution you can make, particularly in being able to submit samples along with any component history available.
It may be that radiation isn’t causing field failures, but something is, and the time is ripe to narrow the possible causes.
P.S. Phil and Tom are speaking at PCB West in September. See the course catalog in this issue.