Soldering SMT Components

You talk a fair bit about all the references you've read, but only cite a trade journal. You also airily dismiss as 'nonsense' the work of a lot of very smart people, with lots and lots of failure analyses of cracked intermetallic layers. This is entirely unpersuasive. The claim that 'no tin whisker has ever been observed growing from lead-free solder' takes you right to the edge of the troll category, and maybe a bit over.

The main issues with lead-free reliability are:

(1) brittle intermetallic compounds: lead-free solders eat most pad metallurgies.

(2) Higher reflow temperatures, leading to much more stress due to thermal mismatch, especially with the simultaneous move to organic chip carriers. (Can you say, "field failures"?)

(3) Higher modulus and lower plasticity in the solder joints, leading to less stress relaxation, aggravating (1) and (2).

(4) Need for different solidus temperatures to allow multilevel packaging, e.g. microbumps from Si to Si carrier, then BGA from Si carrier to module.

(5) Both plating and injection moulding solder are much harder with Pb-free. If you're only using paste, it isn't such an issue.

Tin whiskers are common in softer lead-free solders, which are otherwise helpful in reducing (1)-(5).

Our applications require, e.g. 5000-pin backplane connectors carrying 5 Gb/s differential signals (Ventura J1), and similar numbers of microjoins.

Most of these problems become quadratically worse with pin count, so some folks haven't run into them yet, but they will.

Cheers,

Phil Hobbs

Trade journals can be useful source of information IMO, as long as you consider the biases of the authors. The trade journals have covered both sides of the story and often include useful references. I am not being paid to research this subject for you, so you only get the reference within arms reach right now.

And I should have said that no tin whisker has ever been observed on any lead-free solder currently in production for PCB fabrication. The "whiskers" reported on the softer lead-free solders are a pale imitation of those grown from tin plating, mostly looking short and fat, more like grain growth than whiskers, and produced under thermal cycling no package could survive. No one has demonstrated that these are a legitimate reliability concern with actual failure statistics.

Cracked intermetallic layers are a solved problem now (at the PCB level anyhow), those early failure analysis are no longer relevant.

It seems I am talking about current PCB fabrication methods and you are discussing package assembly. If my points seem one sided, the other side has already been pretty well covered here :-).

What do you mean by "eat" - dissolve? Yes, any solder suitable for use with ENIG completely dissolves all of the plating and forms a solid solder to copper metalurgical bond. There are lead-free solders available today which perform well on all common pad metallurgies, such as Amtech NC-560-LF.

In general it is no longer difficult to find a solder which will work well with whatever your pad finish is, although only OSP and ENIG are widely used.

The dissolution of copper traces is only an issue for rework, it does limit the number of times you can replace a package on a PCB to a lower number than in the tin-lead days, but since most production boards never see any rework and one or two rework cycles are no problem, this is not much of an issue.

A valid concern, and a reason not to use SAC but a more complex proprietary solder with processing temp only slightly higher than tin-lead for Hi-rel apps. The organic chip carriers now have a higher silica content and lower CTE than they did in the tin-lead days, done to accomadate the SAC process, and you can process cooler that that, so it is not clear that you are any worse off now.

Depends on the specific solder used, and means only that you need to consider the solder properties when designing the joint. Did you look at the solder stress/strain curves I referenced?

Different solidus temps are available.

I only use paste :-). Plating is indeed a serious and not yet fully resolved problem, related to ROHS but not to lead-free solder. Injection moulding solder sounds liks a packaging issue.

So far no one has contradicted my assertion that there have been zero failures from tin whiskers growing from any lead-free solder now in production use. You complain about my references, where is your evidence that tin whiskers on lead-free solder are so common? Where are the failures from tin whiskers in lead-free solder? Just because someone can produce something that looks a bit like a whisker in the lab does not make tin whiskers in lead-free solder common.

These density issues would be difficult with tin-lead, why do you think they will be more difficult with lead-free when you have a much wider range of properties to choose from?

I do not buy that PCB fab is significantly more difficult or expensive with lead-free for any outfit big enough to own and operate modern automated assembly equipment. I expect better reliability from my new lead-free Intel workstation boards, CPUs, Adaptec raid controllers, etc., than their tin-lead predecessors. Of course this stuff was all made in Asia where they do not seem to have so much trouble adopting new technology :-).

Cheers, Glen

You're generalizing from the easiest cases--solder paste and PCBs with ordinary SMT components. For situations where those will work, you're probably fine. But your blanket statements wouldn't persuade anyone facing the actual problems--"different solidus temperatures are available", for instance, and "organic packages have a higher silica content". Plastic, whether filled or not, has several times higher CTE than alumina chip carriers, and the CTE is nonlinear due to the glass transition. It's cheaper, and its lower dielectric constant makes signals propagate faster, but if you can't make 7000-pin packages out of it reliably, you can't build advanced servers on it.

I didn't claim that whiskers were common nowadays. I claimed that the metal systems that avoid them either (a) contain lead, or (b) suffer from problems (1)-(5). If your chips and modules are small enough that you don't have to calculate whether the CTE mismatch is going to rip the solder balls off the corners of the module, you don't understand the dimensions of the problem. Since your systems are simple enough that you can get away with paste, I presume that that's the case.

And solder *still* doesn't leach out of landfills in remotely dangerous quantities. Lead-free solder has been big diversion of time, effort, and money for zilch verifiable health and environmental benefit. What a stupid waste--typical ideologically driven Eurocrap.

Best of luck.

Cheers,

Phil Hobbs

And ain't it so that they conveniently exempted car batteries? Nobody can tell me that 100% of them properly enter the recycling process. So if, say, 5% of them don't then that's a whole lotta lead.

About a month ago when I was in Europe I saw that the hardware stores still sold those big super flashlights with lead acid gel cells in them. Guess where they end up once they croak?

You have nailed the main reliability issue with solder attach of large packages exactly here; thermal expansion mismatch causing fatigue failure of the balls at the corners of large BGAs. This is exactly where lead-free outshines tin-lead by the largest margin in every published test I have ever seen, because the fatigue properties of lead free are so much better than tin-lead. Even the lowest cost SAC alloy outperforms tin-lead here. Processing temperature is not the issue, the issue is thermal cycles in service, and all of your complaints about lead-free are completely irrelevant to this issue. The result of switching to lead-free for large BGA attach are clear: a statistically significant reduction in field failures.

The existence of a thousand ways to use lead-free solder incorrectly so that reliability is reduced does not change the fact that when used correctly reliability is increased significantly.

Irrelevant. Adapt to situations you cannot change, or lose market share. I have made my choice :-).

Likewise.

Yep. Haven't seen any in stores i habitually shop in for a while. Gonna have to find some.

problems

station.

devices

They

$500.

Wellers.

I know that feeling too.

have

of

That is a bit of a ride. Do they do full custom boots?

have

of

Don't know. You could try these guys but I don't know them (might be the same store but I always drive up and don't remember phone # etc.):

Speaking of which -- I have a Hako 936 that goes from ~300C to ~ 900C. What is a reasonable starting temperature for soldering small parts?

Thanks

ms

I'd experiment first, and see what's the MINIMUM temp you can use, that will still melt the solder you will be using.

I found that a little incredible, so I tried a web search. Hits are saying more like 400-900 degrees F. That's about 755 K, which is close to the minimum to visibly glow at all in a dark room. The portion of the iron heated most by the heating element probably gets somewhat hotter and may glow to an extent that is fairly easily visible in a dark room.

As for what temperature to use: I would use the lowest that works well and does not require cooking the part too long to solder it. Many component datasheets say 260 degree C max soldering temperature, so for small parts the tip of the iron does not need to get much hotter than that. My limited experience with adjustable controlled temperature soldering irons suggests about 230-240 degrees C (about 445-465 degrees F) for most small parts. You should experiment and use the lowest temperature that works well and achieves reasonably fast soldering.

- Don Klipstein ( snipped-for-privacy@misty.com)

IPC "and other standards" recommend 40 C above the solder melting point according to:

Glen

Thank you.

Robert H.

We have 5 Metcals at work and paid less than $500 for the lot. Yea eBay. Cheapest Metcal power supply I picked up for $10 and since it was local in LA, there was no shipping as I just stopped on the way home. It was broken but the repair required an IRF130 power mosfet (?) which we had in stock. It's the Metcal I use on my bench. BTW for SMT 'lytics and through hole components I use an STTC-126 tip. We picked those up on eBay for $100 for a box of 10.

GG