Are we interested in controlling the process, or testing it?

Process residues are a part of every process, and many assemblers use no-clean assembly. So why would anyone worry about process residues? Aren’t they benign substances that can be left on the assembly surface?

If there is concern, doesn’t the traditional cleanliness testing system’s measure of cleanliness provide enough information to control the process? It tests the total board surface, but if flux is applied to the assembly’s bottom with a spray fluxer, and the fluxer is uniform in spray pattern, wouldn’t the flux be evenly distributed?

When investigating a failure analysis, the review of production floor data shows low readings (averages of five readings) and a low range of data (Figure 1). According to the specified upper control limit of 10 µg/in² of NaCl equivalents, all samples tested passed without concern. Twenty-four weeks into the new process, however, field failures caused by electrochemical migration and leakage began showing up at a 23 to 30% return rate.


A thorough investigation was launched and failure analysis showed high levels of WOA on the capacitors, but only in specific areas of the assembly. Upon review of the process monitoring data, no change was seen in assembly cleanliness when the process change (new selective pallet design) was incorporated on Day 8, which correlates to the production of the failed units.

The dendrite (Figure 2) shows metal migration below the conformal coating on the boards produced on Day 8 using a no-clean water-based flux. This trapped flux from the selective pallet covered approximately one-third of the bottom side of the assembly, but no increase was seen on the ROSE monitoring results.


Often, the general average of conductive residues is not enough to help us understand the pockets of contamination that affect product reliability. In this case, the pocket is a thick layer of no-clean water-based flux on a capacitor from a selective pallet wave solder process left in an area that did not get enough heat to activate the flux. Adding a silicone coating would create conditions for electrochemical migration to occur when the unit is powered.

Visually, there is no visible residue on the surface of the component, yet partially heat-activating the flux leaves a layer of corrosive flux below the layer of insulative flux. The coating adds an additional layer of protection, but unfortunately the partially activated flux absorbs any fluid from the coating. Enough moisture gets trapped below the coating to add the third element to the corrosion cell.

Remember, the three requirements of a corrosion cell: 1) voltage differential (>1.5V), 2) fluid media, and 3) corrosive residues. In this case, below the coated component that showed a good general cleanliness as measured by ROSE, a dendrite can form in a short time after the unit is installed and powered.

When attempting to control an assembly process, it is important to understand the amount and nature of the residues in critical areas, rather than taking the general average of the entire assembly or using the average of these averages.

Are we more interested in controlling the process? Or just testing it?

Terry Munson is with Foresite Inc. (residues.com); tm_foresite@residues.com. His column appears monthly.