PCB Failures. Cracked Vias and Layer Separation

Manufacturing Manufacturing process
, , ,
1

Hello Everyone,

My company has had trouble with a series of PCB failures. These failures include cracked vias and laminate separation. The via failures apply to both microvias and standard via barrels, both through-hole and buried. There appears to be a lack of adhesion between layers. The majority of the failures appear to occur directly under BGA parts, primarily Xilinx Artix 7 FPGA. In other words, the bottom-most layers of the board do not show notable failures since no BGA parts are assembled on the bottom layer.

Here are some basic characteristics of the boards:

  • Laminate/prepreg: FR408HR
  • Number of Layers: 10
  • Number of Laminations: 3
  • Leaded Solder
  • Approximate Board Area: 2.00" x 2.50"
  • Through-hole vias are filled with Taiyo after plating
  • Microvias are fully copper-filled

We have seen the same failures on boards from two different fabrication and assembly plants. However, we haven’t seen any indication that the PCB layout design itself is a problem.

Here are some potential sources of failure we are pointing at:

  • Lack of temperature control during the fabrication process
  • Lack of temperature control during the assembly process
  • Aged or improperly stored FR408HR material being used for fabrication
  • Lack of proper planarization or cleaning prior to adhering layers

Images of the PCB stackup and cross-section photos are included to illustrate the problems.

I appreciate you taking a look at my post. I would love to hear your feedback.

Thanks!
Ian

(Since this is my first post, apparently I can only post one photo. Sorry for the montage!)

montage
montage1920×871 251 KB

1 Like
2

Hi Ian, sorry I’m just seeing this. Our engineering team will reply soon.

3

Hi Ian,

Here are some additional things to consider along with what was already listed:

  • Flat/spread glass should be used if it isn’t already.

  • The shape of the laser vias is not great and this can lead to chemical entrapment & poor plating. Has a direct effect on bond strength between the plating and capture pad where most of the problems are occur.

  • Might consider a material like I-Speed instead of FR408HR which has energy absorption issues and doesn’t laser drill all that well.

  • Planarization can also induce a lot of stress and crack stacked laser vias.

  • Hard to see what’s going on with the filled through hole but it’s not cap plated and could be an out-gassing issue from the fill material – possibly under-cured.

Just some basic observations based on limited information but we can have a more in-depth discussion if you want.

4

@lucy.sierracircuits & @steve.carney ,

Thank you for your responses! This is very valuable feedback. We already have another fabrication job underway using Isola 370HR. This was chosen over I-Speed since there is only one differential signal that goes up to 1 Gbps on this particular board . We have used Megtron 6 in other high speed designs, and have never seen these kind of board failures.

I’ll follow up when we get the new boards in house and tested.

Thanks!!!
Ian

1 Like
5

Good luck!