With next weeks launch of Intel's 45nm family of microprocessors they will have official removed 100 percent of the lead (Pb) across its entire 45nm series of microprocessors. This is a huge milestone and it has taken Intel six years to strip the lead from the manufacturing process on their microprocessors! Read on to see what Intel did to remove the last 5% of lead from their processors!
"Now that the flip-chip bump solder points that used to contain lead have been replaced with a copper column and lead-free solder the processors are more environmentally friendly, but this does come at a cost. The melting point of lead-tin is 183C, whereas the melting point of lead-free solders is greater than 220C. This means increased stress on the flip-chip package and more power consumption since the convection ovens need their baking temperatures increased. To manage the increased stress and survive the higher process temperatures, Intel reengineered many parts of the processor..."
Not to turn this into a political debate, but...
It's funny to me that all electronics need to be lead free but the U.S military is firing depleted uranium rounds all over the Middle East. I guess you have to start somewhere though.
A step in the right direction, but chip making involves so many toxic chemicals though. When a wafer goes into the "bath" to take off some stuff, it's certainly not green ;)
I just registered to this forum to comment on this article.
I feel the author made a lot of comments that are not backed up by science or test data.
Lead free solder is much less reliable than Tin-Lead solder. Studies has been done in the past 20 years to find an alloy that has similar visco-elastic properties as eutectic Tin-Lead alloy. There is none. In fact, from the cross-section of Intel's lead-free solder bump, it shows that Intel intentionally increased the copper bump thickness (UBM layer) so they can use less Lead free solder.
I personally have dealt with a lot of cell phone designs that went to ROHS and lead free a couple years back. Mechanical shock and vibration testing have shown that lead free solder is not up par (not equivalent to) tin-lead solder but the cost of using lead-free solder is much higher due to the control of Ag content in the alloy. The board design of lead free parts had to deviate from IPC recommendation to improve the shock and vibe performance.
Another critical issue about lead free solder is Sn leeching. I guess this is another reason why Intel went into a thick copper bump. The lead free solder is high Sn which went biased in current will create a brittle inter metallic compound at the interface. I do not know of any rigorous life studies of this phenomenon; unless Intel has their own data (I assume they do). But if Intel has their own data in this area and decide not to share with IEEE, then I also assume it is a negative.
Some may comment that temperature fatigue performance of lead free solder is stronger than tin-lead solder. But this is irrelevant because the thermal fatigue life of tin-lead solder is already sufficient. Additionally, failures due to mechanical shock is more common.
Those of us who is in this industry knows that the RoHS requirement is more politics than anything else; it goes much further than just lead-free. It is a step backwards in terms of cost vs. reliability. You're paying more to buy a less reliable products.
There are other methods of controlling lead contamination or lead consumption.
EDIT:
Just to add another note. RoHS has a special examption for flip chip interconnect to use lead. So design houses can use eutectic Sn/Pb solder as the flip chip interconnect and still be RoHS compliant.
The cross-section of the lead free bump shown in the article already shows some Sn intermetallic growth on the copper pad.
A lot of them Middle Eastern Countries take out and sell the led from old computer parts that can't be used anymore. But the bad thing about that is that a lot of their people get really sick and usually die from led poisoning. Most of these victims are just kids.
Pro: Intel saving some people from led poisoning
Con: No business for people so they die from starvation.
Lolento wrote:Those of us who is in this industry knows that the RoHS requirement is more politics than anything else; it goes much further than just lead-free. It is a step backwards in terms of cost vs. reliability. You're paying more to buy a less reliable products.
So true! Great first post and thanks for joining to share your thoughts. You said the cross-section of the lead free bump shown in the article already shows some Sn inter-metallic growth on the copper pad. Is that what the slightly darker layer is on both top and the bottom? Here is a higher resolution picture of a lead-free bump for you to look at. I'd be interested in knowing if that is tin, why it is doing it.
Attachments
high-k metal gate processors with a copper column “bump” and a tin/silver/copper solder alloy shown here
Has the Clean Air Act done more to fight crime than any other policy in American history? That is the claim of a new environmental theory of criminal behavior.
In the early 1990s, a surge in the number of teenagers threatened a crime wave of unprecedented proportions. But to the surprise of some experts, crime fell steadily instead. Many explanations have been offered in hindsight, including economic growth, the expansion of police forces, the rise of prison populations and the end of the crack epidemic. But no one knows exactly why crime declined so steeply.
It claimsthat the rise and fall of lead-exposure rates seemed to match the arc of violent crime, but with a 20-year lag... Interesting stuff
Yes, the thin halo between the copper and the solder should be SnCu intermetallic which is very brittle.
It shows this construction is less reliable.
In the leadframe type packages (very old technology), where pure Sn is plated on top of pure Cu, the thickness of the SnCu intermetallic is controlled by using an annealing process to create SnCu3 intermetallic which is stable when current passes. Otherwise, SnCu6 is formed when current passes through the copper-Sn interface and over time it will consume the interconnect. None-the-less, any SnCu intermetallic is brittle in the scale of a flip chip interconnect.
