Dye Penetrant Testing for BGA Joint Cracks and Voids

Dye penetrant testing is a destructive method in which a coloured fluid is drawn into every crack in a solder joint before the package is separated from the board. When the joint is opened, the stained area is the crack, and the unstained area is metal that was still bonded.

It answers a question that X-ray cannot answer. A two dimensional image shows that a joint is present and roughly what shape it has, but it does not show whether a crack surrounds the whole joint or how far it has travelled. The dyed fracture surface does.

What the Method Reveals

The result is a map of the failed area. A crack that has propagated around the perimeter of a BGA ball appears as a ring of dye, and the ratio of stained to bright metal indicates how much of the joint was still carrying load.

This matters for failure analysis, because the location of the crack identifies the mechanism. Separation near the package side points to package warpage, separation near the board side points to board or pad movement, and a crack through the bulk of the ball points to mechanical damage rather than thermal fatigue.

Preparing the Assembly for Test

Preparation starts with cleaning. Flux residue, conformal coating and moisture all prevent the dye from reaching the fracture, so the assembly is cleaned and dried before the test, and the cleaning method should not be one that could itself open or close the crack.

<img src="https://www.gopcba.com/wp-content/uploads/2020/12/project_image_07.jpg" alt="BGA solder balls stained with dye penetrant after a pry test” />

The board is then marked so that each ball can be identified after separation. Photographing the package outline and the ball grid orientation, and cutting the board so that a known corner can be recognised, prevents the confusion that ruins otherwise good work when twenty joints have to be matched back to their positions.

Penetration: Vacuum, Dwell and Dye Choice

The dye is applied to the joint area and the assembly is placed under vacuum, usually for 30 to 60 seconds, so that air is removed from the crack. The vacuum is then released, and atmospheric pressure drives the dye into the space that the air has vacated.

A dwell period follows, commonly two to five minutes depending on the dye and the gap, after which the assembly is cured or dried so that the dye stays in place. Each of these three numbers should be recorded with the result, because penetration depth depends on them and a comparison between two tests is meaningless without them.

Cleaning and the Separation Step

Excess dye has to be removed from the surface before the joint is opened. Dye left on the board or on the package will contaminate the fracture surface and produce a false indication of failure that did not exist.

Cleaning is done with a solvent that dissolves the dye but does not act on the solder or the laminate, and the board is dried again before separation. A common mistake is to clean aggressively enough to remove dye from the mouth of the crack as well, which makes a real failure look like a partial one.

The Pry Operation

Separation is done by prying the package away from the board, and the direction of the pry controls which side of the joint is exposed. A controlled force applied slowly gives a clean separation, while a sudden load can bend the package and create new fractures that were not present before.

Tooling should be chosen so that the load is applied evenly along one edge of the package. Heated pry methods are used where the alloy is ductile, but heat also relaxes residual stress and can close a crack that the dye has not yet entered, so the choice of method should be fixed for a given investigation. Packages that cannot be pried without distortion are removed with the techniques used for BGA reballing, where the ball is removed rather than the package.

<img src="https://www.gopcba.com/wp-content/uploads/2021/03/ptt_shop.jpg" alt="Cross section of a BGA solder joint beside a dyed fracture surface” />

Reading the Fracture Surface

Examination is carried out at 20 to 50 times magnification, with the light set to show the difference in texture between dyed and clean metal. The dyed region is usually dull and coloured, while the bonded metal has the bright, torn appearance of a ductile fracture.

Each joint is then classified: fully dyed, partly dyed, or clean. The classification is what makes the test useful in a comparison, because the numbers can be summarised across a package and compared with a control sample from a known good assembly.

Interpreting Dye Coverage

A joint with a complete ring of dye had no load bearing connection left. A joint with a crescent of dye had a partial crack, and the width of the crescent indicates how much of the interface has separated. A completely clean joint did not fail during its service life, and if many joints are clean the failure is localised rather than systemic.

The pattern across the package matters as much as the individual joints. Dye confined to the corner balls is a signature of package warpage, while dye spread evenly across the grid points to a global cause such as thermal cycling or an incorrect reflow profile.

Limits and False Conclusions

The method cannot detect a crack that does not communicate with the outside of the joint, and it cannot be applied to joints that are underneath a heat sink or a staked component without removing them first. It is also destructive, so the assembly cannot be returned to test afterwards.

A second limitation is that the penetration step depends on the crack being open. A crack held closed by residual stress, or one that has been healed by a subsequent reflow, will take little or no dye and be reported as healthy. For that reason the method is always used alongside a non destructive technique, normally the X-ray and AOI comparison that identifies which joints deserve a closer look.

Comparing With X-Ray and Cross Sectioning

X-ray is fast and non destructive and suits screening, but it shows a projection rather than the interface. Cross sectioning shows the internal structure of a few joints in great detail, including voiding and intermetallic layers, but it destroys the joints it opens.

Dye penetrant sits between them. It is destructive and laborious, but it produces a picture of the whole interface of many joints at once, and it is the only one of the three that distinguishes a crack open to the surface from an internal void that never reached it.

Sample Selection and Records

Because the test is destructive, sample selection carries most of the value. Joints at the corners and the centre of the package, and joints that were flagged by X-ray, give more information than a random selection, and a control package that has never been stressed should be included in the same run.

The record should contain the dye type, the vacuum and dwell times, the pry direction, the magnification used and the classification of each joint. Without the process parameters the test is not repeatable, and without the joint classification it produces photographs rather than data. The wider context for failures found this way is covered in the material on solder defects and board failures.

FAQ

Is dye penetrant testing destructive? Yes. The package is separated from the board so that the joint interfaces can be examined, so the assembly cannot be used after the test. This is why the method is applied to samples and to failures rather than to production.

How long should the dye be left under vacuum? A vacuum of about 30 to 60 seconds is typical, followed by a dwell of two to five minutes under atmospheric pressure. The exact values depend on the dye and the joint geometry and should be fixed for a given investigation.

Can dye penetrant testing find voids? It can find cracks and open voids that reach the outside of the joint, but a closed void inside the ball will not take dye and will not be visible. A cross section is needed for those.

}