I met a guy on a plane last month who has a career testing electronic components for the military and aerospace and setting standards for reliability. He told me about the new European Union's Restriction of Hazardous Substances (ROHS) standard, which contains provisions outlawing lead in motherboards and a host of commercial and consumer electronic equipment. All very high sounding lead is poisonous, protecting the individual, and so on. Yet the replacement of lead with pure tin directly is not without its own problems: It leads to much higher incidences of unreliability.
In what way? how about shorting out the entire motherboard? How about frying the hardware in one fell swoop? Or maybe just those annoying little hardware bugs you can't explain.
For an interesting demonstration of tin degrading in the span of one day, take a look at this video on YouTube.
But it gets even worse, tin forms tiny whiskers that grow over time sometimes starting after a day, sometimes after many years and these eventually trigger a short circuit in electronics. This isn't a one-off. It is a well-known phenomenon that has been documented since the 1940s. A team at the University of Maryland has been studying tin whiskers and tin degradation for years.
Now this could well have happened inside your server, blade or laptop that failed unexpectedly. I had a brand new Dell laptop a couple of years back, and the motherboard shut down all of a sudden. Nobody could explain the failure other than the generality of a faulty motherboard. But for the past few years, most computer electronics have had lead-free solder. Pure tin solder has gone from nowhere to a majority share in just a few years. And unreliability might be the price we have to pay.
It might be coincidence, but blade enclosures are designed so that WHEN a failure happens, one blade is always spare as a failover (i.e., the system is now engineered for the expected unreliability). While failover itself isn't a bad idea, when did we decide to accept failure as a fact of life? If it seems distinctly un-American, it is. It's a European concept: Who cares if your cell phone dies or your laptop fries or your server shuts down it's good for the environment. Amazing what a group of suits will do when you lock them in a room and try to reach a consensus.
The guy on the plane was very worried about how this ROHS tin/lead standard had rolled across the planet and was making its way into aerospace. He told eerie tales of entire satellites failing in orbit, space shuttle components that keep failing and more. All in the name of green IT. Shocking!
(For the scarier side and a mini-briefing on tin whiskers, see five one-minute videos.)
To make matters worse, some parts of the world are adding metal bismuth into the solder mix as a solution. If you leave a tiny trace of lead in it, however, you get cracks and disintegration. And of course, bismuth is far more harmful to the environment than lead. Other attempts at a solution using a variety of tin alloys have also been subject to tin whiskers. Coatings haven't solved the issue, either. Bottom line: It's one great unsolved problem.
Now let's get some perspective on lead. The effects of lead poisoning are pretty horrific. It poisoned the Romans when they used it for plumbing. In effect, they were drinking a potent lead/water cocktail. So plumbing pipes now use copper, iron, steel and plastic. Lead-based paints were eliminated in the past couple of decades, too. Yet in my last two houses, built before the 1950s, lead-based paints were almost certainly used from top to bottom. Half the nation could well be in a similar predicament and most of Europe. If it's so appallingly harmful, where is the lead poisoning pandemic?
Lead is actually a fairly common element and is found in nature in several forms. It really is hard to avoid. So what's next lead-free mountains? The point is that lead must be eliminated sensibly and in some areas like electronics it is not always a smart move. We are talking here about a tenth of a gram of lead per chip. Maybe a gram or so per motherboard. Is it so dangerous in those quantities that it is preferable to remove it and resulting in more hardware crashing? Let's think about it in terms of carbon footprints or environmental footprints: a gram of lead on a motherboard vs. far more electronic components sent to the landfills of the world due to rampant unreliability?NASA is terrified of the tin whisker problem. Type the term into the www.nasa.gov web site and take a read. Space Shuttle, space station, Hubble and satellite troubles galore are traceable to this problem. The military is refusing to adhere to lead-free solder. Yet both NASA and the military will feel the adverse effect of it, as they can't control every single component because they have come to rely on commercial, off-the-shelf hardware so heavily. For the average IT department, though, all of the big server OEMs have moved over to lead-free solder. So get used to strange glitches, part replacement and general hardware havoc. It might be a good idea to really lean on your vendors for top-notch hardware replacement warranties before this becomes too big of a problem.