Gold embrittlement of solder joints still appears as a problem from time to time even though the issue has been understood for many decades. We diagnose this problem by using SEM/EDS to estimate the amount of gold in the solder joint. If the gold level exceeds about 3 wt% in the bulk solder joint then the joint is considered to be embrittled.

We also diagnose gold embrittlement by using BSE SEM images to estimate the area fraction of AuSn4 intermetallic compound in the solder joint.

There is a relationship between the wt % of gold in the bulk solder joint and the area fraction of AuSn4 intermetallic compound in the solder microstructure.

For eutectic tin-lead solder the relationship is as shown in the above figure, suggesting that 3 wt% Au corresponds with ~ 12% area fraction of AuSn4 intermetallic compound in the solder microstructure. AuSn4 is a relatively hard/brittle phase in a softer Sn & Pb matrix, so when the AuSn4 exceeds ~ 12% the integrity of the solder joint begins to suffer.

See also GOLD EMBRITTLEMENT OF SOLDER JOINTS.

Check out SEM Lab, Inc.  to learn more.

This is a microsection of an MLCC as mounted on a PCBA. There is a fracture in the top of the MLCC, which is unusual.

The fracture was clearly associated with some type of damage introduced while the ceramic dielectric was in the “green” state, i.e. prior to firing during fabrication of the multilayered ceramic structure. This type of damage cannot have occurred after firing when the ceramic is no longer deformable as it was as green tape.

It is interesting that this defect escaped any visual and electrical inspections that might have been performed at the factory.

Check out SEM Lab, Inc.  to learn more.

The image below is a “stud diode” that had failed shorted.

The diode was chemically decapsulated and the device die was examined in the SEM. A breakdown site was found near one corner of the die.

Below is a higher magnification image of the breakdown site…

… which shows a striking resemblance to the “Face on Mars”.

Conclusion – the short was caused by an alien intelligence far greater than ours.

Check out SEM Lab, Inc.  to learn more.

Sometimes we’re lucky and stumble into something like this ESD damage in a Zener diode. The diode was short circuited, but it was micro sectioned in order to determine the die thickness (not something we would normally do on a shorted diode).

A damage site was found directly under the ball bond. The damage was determined to be a gold spike (as in ESD) that punched the junction shorting the diode.

Check out SEM Lab, Inc.  to learn more.

Activated solder flux residues left on a PCBA can cause severe corrosion problems. The leads on the component shown below are severely corroded.

The elemental spectrum suggests that the needle like corrosion product is lead chloride & lead bromide.

Conclusion – avoid inadequate cleaning of PCBAs.

Check out SEM Lab, Inc.  to learn more.

Corrosion due to residual solder flux and cleaning process chemistries left on PCBAs can be a real problem for reliability. The corrosion can cause both open circuits as it eats through metal runs and short circuits due to electro-chemical migration.

This (above) is the elemental spectrum of the end of a through-hole connector solder joint. The chlorine in the spectrum is likely due to chloride activator from the solder flux.

This (above) is the elemental spectrum of some corrosion product associated with the through-hole connector solder joint, which appear to be Pb and/or PbO crystals growing out of the surface of the solder joint.

These are the Pb and/or PbO crystals growing out of the surface of the solder joint.

Check out SEM Lab, Inc.  to learn more.

Electrical continuity tests suggested that the gate was shorted to the drain and the source was intact. SEM analysis revealed a suspected damage site at the end of a gate finger (see below).

An elemental dot map at the breakdown site showed some slight disturbance of the aluminum metallization at the site.

Voltage contrast imaging confirmed that the suspect damage site was indeed the location of the short.

These results suggested that the device failed due to a voltage transient on the gate signal. The small dimensions of the damage site suggest that the transient was very fast (rough estimate 1E-9 to 1E-6 seconds). The damage could potentially be related to an ESD event. It could also potentially be caused by accumulated damage at the tips of the gate contacts due to excessive turn-on or turn-off dV/dt.

Check out SEM Lab, Inc.  to learn more.

Here is another example of electro-chemical migration that resulted in shorted signals on a printed circuit board assembly.

These ECM residues can often be hard to see under an optical microscope because there are optically transparent metal oxides present with very small metallic dendrites.

Backscattered electron SEM images show contrast related to atomic number, so for example lead appears very bright (Pb z=82) and carbon appears dark (C z=6). The dendritic structure is a clue that this was ECM related.

The elemental map below shows that the dendrites are mostly lead but there is also some tin. Both lead and tin are likely to be involved for Sn-Pb solder joint assemblies.

Check out SEM Lab, Inc.  to learn more.

ECM (electro-chemical migration) can occur when moisture, voltage bias, and ionic contamination come together, such as was the case for the PWB battery contacts shown in the BSE SEM image below.

Elemental analysis suggested that the ECM residues were primarily copper, but also zinc (from brass?) and nickel (from ENIG finish or nickel under-plate on the connector contact?).

The feature that identifies this as ECM versus plain chemical corrosion is the dendritic structure of the residue as shown in the next image.

Historical data from this laboratory shows that the most significant factor is often moisture, followed by voltage (or local electric field), and finally ionic contamination. The lesson seems to be “keep it dry”.

Check out SEM Lab, Inc.  to learn more.