Is There a Thermal Cycle Limit?

Phil
Welcome to Board Talk. This is Phil Zarrow and Jim Hall, the Assembly Brothers also known as ITM Consulting. We're here to answer your process questions with regard to SMT materials, components, equipment, and materials. We have a theory question today.

Jim
The question is from P.W. and the wording is, "How many thermal cycles can we expose a Class 3 printed wiring assembly to before it no longer maintains a Class 3 rating?" They're talking about IPC Class 3 as spelled out in the holy Bible of our industry, IPC 610D.

This is the highest rating for critical electronics. The question continues, "In general, how many cycles should we allow at a given location on a circuit board, such as a BGA site, before we should consider the assembly compromised and thus be unreliable at any class rating?"

Now these are really good questions.

First off, we're assuming when we're talking about thermal cycles, soldering cycles, be it reflow or wave or hand soldering.

Phil
Let's say if we were putting down an SOIC, that's one thermal cycle. And let's say we have to repair it, that's another thermal cycle to remove it and another thermal cycle to replace it.

If it's on the A side of the board, it would have gone through one thermal cycle to place it on the board, and a second thermal cycle while the other side of the board gets reflowed.

Jim
So the answer is that any given site can potentially see a lot of thermal cycles and what happens? Well, one thing that can happen, particularly if you're dealing with anything but an ENIG surface, the copper tin inner metallic will continue to grow and get thicker every time you heat that joint.

We all know that the copper tin inner metallic is strong but brittle, so you want a thin layer, not a thick layer. So as you continue to heat this through repair and multiple reflow and wave soldering cycles, you stand a chance of making that inner metallic too thick and reducing the reliability of the solder joint.

This is greatly reduced if you're using an ENIG finish, electroless nickel with immersion gold, because you're not soldering to copper, you're soldering to nickel and the nickel tin inner metallic doesn't tend to form nearly as quickly.

The other issue is the circuit board. What about the materials underneath the pads? What about the adhesive material that's bonding the copper pads down to the PCB and what about the PCB materials, the laminate materials, and the vias in the same area?

Phil
The thermal cycling will affect the Z axis, as well as the XY axis.

Jim
And of course the key given here is a BGA site because when you get into BGA repair, you have one or two reflow cycles up front. Then you've got to take the BGA off and then you've got to get in and scrub the excess solder off and then you've got to put stuff down and then you have to reflow it again.

So there's a lot of concern about damage to the BGA pads in the BGA areas because when removing them, you tend to have to heat them quite intensely to make sure you melt all the joints before you pull the BGA off. The BGA repair sequence raises the bar, raises the concern.

As Phil said, unfortunately, we don't have any numbers. IPC doesn't seem to publish it but people have come up with their own numbers, although I can't recall off the top of my head the methodology they went through.

You can compromise a circuit board, a local area of a circuit board, by applying too many heat cycles and the rework cycles are particularly a concern, especially with area rays where you've got to get in and really heat them to get the parts off and then scrub them to get the excess solder off the pads before you do the repair.

P.W., you're on the right track. I'm sorry we don't have any hard numbers for you but I would say think about it and certainly think about how many BGA repairs you want to allow on a given site.

If you're talking about BGA repair, think about minimizing the intensity and duration of every one of those heating cycles because none of that heat is doing you any good. That's the bottom line with any of these materials. Extra heat cycles do you no good.

Phil
That about attempts to answer that question and we thank you for tuning in to Board Talk. And on that note, I will say a fond farewell on behalf of Jim Hall and myself, Phil Zarrow, and whatever you do, wherever you're doing it ...

Jim
Don't solder like my brother.

Phil
Please don't solder like my brother.

Another point to be considered is the thermal stress toward the laminate, especially in case of halogen free FR4. Here the usage of Aluminum Hydroxide (ATH) as a flame retardant filer is very common. Unfortunately it decomposes at temperatures starting all ready above 200C to release water vapor. Too many thermal cycles above 200C will cause internal or full delamination. While the full delamination is visible and scrap boards can be sorted out, internal delamination can lead to creation of voids where the water can accumulate and strongly accelerate CAF problems.

At my company we qualify PCB suppliers based in part on HATS or IST via reliability testing. For the test, we specify 6 preconditioning cycles, which simulates the manufacturing process for a complex board (2 reflow cycles, 1 wave solder, 1 BGA rework - it adds up fast).

Then we limit our assemblers to 6 thermal cycles at any one site. Thus we maintain the validity of the reliability qualification.

There IS a limit, but there are so many variables that its hard to define. Too long at the Td and the resin degrades. Too long at Tg and delam occurs.

Too much Z axis movement and the interconnects separate from the barrel and the pads rise breaking the "knee" of the hole wall/pad interface. The only way to understand the limits for your own process is to perform DOE's and destructive analysis as part of the validation of the reflow and rework processes.

Too often CCA assemblers are enthusiastic about throwing the PWB supplier under the bus and unwilling to examine their own processes. I have been on both sides of the business.

"Board Talk" is very informative. Keep the articles coming.

Thermal Cycles - tough question to answer - as you stated there is no set number of cycles, it will be dependent on temperature and the component, board material etc..

Keep up the program very informative.

Is there a thermal cycle limit ... no.