The growing use of lead-free soldering in electronics manufacturing has introduced new types of defects that require proper identification and troubleshooting. One specific type of defect tha
t has become more prevalent on lead-free ball grid arrays (BGAs) is called “head-on-pillow”. Head-on-pillow is a defect where both the paste deposit and the solder bump reach a full state of melt but fail to coalesce. It is important to differentiate head-on-pillow from a defect caused simply by insufficient reflow temperature, which is characterized by distinct solder spheres from the paste that have not been properly melted on the pad and BGA solder bump. With head-on-pillow the soldering temperature is sufficient to fully melt the solder bump and paste deposit, but an impediment to the formation of a proper solder joint exists. One or both ends of the failed interconnect will show evidence of displacement while it was melted, forming a shape that resembles that of a pillow with an indentation of one’s head.
A “dye and pry” test is used to identify the presence of head-on-pillow. The dye applied during this test can penetrate the gap formed between the two ends of the poorly connected solder joint and can be identified when the component is removed from the PCB. An example of the results of a dye and pry test on an assembly with head-on-pillow defects is shown in Figure 3-1.
Another method to identify head-on-pillow defects is through visual analysis. Some head-on-pillow defects are readily apparent using relatively
low-magnification optical inspection. If those defects are located on the outer rows of the assembly, an endoscopic inspection system may be sufficient to definitively identify head-on-pillow as the cause of a failure. If the defect is present at an interior location on the interconnect array, a cross-section may be necessary to optically verify the presence of head-on-pillow. Once the solder joint is exposed, low magnification may be sufficient to identify head-on-pillow depending on the severity of the defect. Figure 3-2 shows an example of a head-on-pillow defect that was verified optically without the aid of mounting and polishing.
Some examples of head-on-pillow may not be severe enough to allow identification with low magnification optical inspection. These may still be verified optically, but require polishing and high magnification optical inspection. Figure 3-3 shows an example of head-on-pillow that required high magnification to be clearly identified.
Verification of the presence of head-on-pillow leads to an investigation into the cause. The ultimate root cause of head-on-pillow is a barrier that prevents the coalescence of the solder bump and the solder paste deposit. That barrier may be contamination on the surface of the solder bump that the flux in the solder paste is not able to remove. Testing to ensure that the components are not entering the facility with the suspected contamination present is an important first step. The recommended tests for contamination of this type is an extraction based (non-destructive) method suited for detection of non-ionic contaminants, such as
contamination is found on raw materials in as-received condition, a careful assessment of the factory’s handling practices is required. It is not difficult for contamination from factory personnel, whether in the form of body oils, hand lotions, or machine lubricant to be transferred to a raw component when improperly handled. Tasks such as removing or replacing BGAs on a matrix tray or loading and unloading of trays from placement equipment can be avenues of contamination if personnel are not trained and disciplined in the proper handling of electronic components.
More commonly with lead-free soldering, an occurrence of head-on-pillow is related directly to the reflow process. During a reflow soldering process, the flux present from the solder paste is required to clean the initial oxidation from the parts to be soldered, as well as protect the materials against continued oxidation
during the reflow process. This oxidation prevents proper homogeneity between the solder ball and the solder paste deposit and will result in head-on-pillow.
One possible cause of excess oxidation is insufficient solder paste (and thus flux) volume present. In this case, the solder paste printing process may need to be optimized to ensure that sufficient paste is present. Another potential cause of flux failure during reflow is poor material control. Solder paste materials need to be stored and handled properly to ensure that they function as expected. Careful observation of shelf life control, storage conditions, factory environmental conditions, and stencil life performance are important to ensure that the flux constituent in the solder paste is able to perform all of its functions in the reflow process.
The reflow profile can be a cause of head-on-pillow defects by exposing the flux to excessive temperatures for durations beyond what it is designed to withstand. This causes the flux to exhaust its oxidation cleaning abilities well before the completion of the reflow cycle, regardless of the amount of flux present, and even if the raw component and the solder paste have been properly handled for their entire life. The obvious change that can be made to the process to prevent this occurrence is to reduce the profile length and/or modify the profile temperatures to prevent the premature exhaustion of the flux.
A case of head-on-pillow was recently encountered during a project undertaken at the EMPF. The project involved the attachment of a large BGA component
(approximately 2 in2 with over 2000 solder balls on a nearly full array) to a large PCBA using the Metcal/OKI APR-5000 hot air rework equipment. The BGA incorporated a heat spreader as a die cover and utilized lead-free
tin-silver-copper solder balls. A no-clean solder paste was applied to the PCBA prior to installation of the BGA and was to be reflowed along with the BGA during the attachment process.
Due to the sensitive nature of the PCBA, and the warpage that had been previously observed when attempting an attachment with a reflow cycle of standard length, an extended profile was designed to minimize thermal gradients across the PCBA. This profile extended for over 1000 seconds from ambient temperature to peak. Although excessive thermal gradients across the PCB were no longer present on the longer profile, an unintended consequence was a large number of head-on-pillow defects on the assembly due to the extended duration of the profile and the exhaustion of the flux during the reflow cycle. This was verified by optical inspection after performing a dye and pry operation, followed up with a cross-section; the results of this analysis by the customer are shown in Figures 3-1 and 3-2.
Since reducing the length of the profile would increase the risk of PCBA warpage, another strategy had to be employed to prevent this defect. The use of an inert atmosphere would prevent the formation of an oxide on the solder bumps after the flux was exhausted during the reflow cycle and allow the solder to coalesce. For
verification, another attachment was performed after outfitting the rework machine with a nitrogen source. No changes were made to the temperature settings of the rework equipment. A dye and pry analysis of the soldered part showed no examples of head-on-pillow and demonstrated that the use of nitrogen successfully mitigated head-on-pillow defects on this extended length profile.