Empfasis - Corrosion in Electric Hardware

An EMPF customer was experiencing corrosion of a 100k ohm resistor on a CRT socket. The initial resolution by the customer was to go with a no-clean process for attachment of the arc suppressor housing. The problem persisted, however. As a result, the EMPF proposed elemental analysis of the corrosion through Scanning Electron Microscopy with Energy Dispersive Spectroscopy (SEM/EDS). The results of the SEM/EDS analysis of a stock resistor and a corroded resistor indicated:

The ends of the resistor were composed of tin (Sn) over copper (Cu) over iron (Fe).
The wire leads coming out of the resistor were tin (Sn) over copper (Cu).
The internal body of the resistor was composed of silicon (Si) and aluminum (Al).
The potting compound was silicone based.
The appearance of the corrosion suggested iron oxide (reddish/ brown) and copper oxide or copper hydroxide (green/blue) and was confirmed to be iron (Fe) and copper (Cu) (Figure 2).
The internal body of the resistor was composed of silicon (Si) and aluminum (Al) and the casing was composed of was silicon (Si).

The corrosion was concluded to be the result of dissimilar metals (iron and copper) in the presence of moisture. The presence of visually detectable flux residue was not necessary to cause this type of corrosion from occurring. The moisture most likely penetrated the resistor’s silicone potting compound and the corrosion began at the junction where the wire contacted the end of the resistor. The electrical failures expected from this are loss in impedances at the resistor at best and shorts at the worst. EMPF’s recommendations were to take measures to assure the assemblies are kept dry and identify a component of a different composition in order to remove the potential for this failure mechanism if humidity or moisture is present during operation.

An alternate resistor was identified and replaced the steel end-capped resistor. After some period of time with the new resistor in place corrosion, failures developed once again (Figure 3). It was determined that a no-clean hand-soldering process was implemented. The bare board had a HASL finish which was applied with a water soluble flux. The resistor according to the customer was Corrosion in Electronic Hardware (continued from page 5) operating at 9.5KV and +/- 15μAmps. In addition, the assembly was not conformally coated. As in the previous work, the assemblies were visually examined and elemental analysis of the corrosion was performed through SEM/EDS. Along with these steps, the ionic cleanliness of the raw materials, inprocess and finished assemblies was determined.
The results of these findings were:

Corrosion was observed on field returns at the resistor and the through-hole leads of the housing.
Inventories of the video amplifiers displayed white residue at the solder joints on the resistor and at the through-hole connections of the CRT socket.
The blue-green color of the corrosion indicated it was mostly copper oxidation which was confirmed through SEM/EDS analysis (Figure 5).
The resistor body was composed of oxygen, silicon and magnesium while the lead was tin (Sn) plated copper (Cu). There was no indication of a mismatch in materials (i.e. galvanic corrosion).
Ion chromatography (IC) analysis of a full assembly (assembly 1) indicated the presence of chloride, bromide, nitrate, sulfate and an unknown peak at 2.4 minutes and 4.5 minutes with the bromide levels and the peak at 4.5 minutes being significant (Table 1).
IC testing of a bare board indicated the presence of chloride, bromide, and a peak at 1.1 minutes. Bromide levels were above ACI’s recommended limit of 5 μg/inch2 for bare boards (Table 1).
IC testing of the resistor and a suppressor housing unit indicated trace amounts of chloride, nitrate and sulfate along with the peak at 1.1 minutes observed on the bare board (Table 1).
The arc suppressor housing unit was removed from assembly 2 and both parts analyzed separately. Prior to analysis residue was noted on the PCB surface underneath the location of the CRT socket. IC testing indicated the presence of chloride, bromide, nitrate, sulfate and the 1.1 minute peak on both the board and arc suppressor housing. The peak at 4.5 minutes observed earlier was observed only on the arc suppressor housing unit (Table 1). Bromide levels on the board were above ACI’s recommended limit of 15 μg/inch2 for assemblies.
Analysis of the no-clean flux used to attach the suppressor and resistor displayed chloride, nitrate and a peak at 6.5 minutes. This peak did not correlate with peaks observed after secondary soldering of the suppressor or resistor.
Three assemblies prior to attachment of the arc suppressor housing unit and resistor were IC tested. These results indicated the presence of chloride, bromide, nitrate, sulfate and the 1.1 minute peak (Table 1). Bromide levels were excessive however.

EMPF has generic guidelines for assemblies that are not application specific. The level of ionic residue that is acceptable for this application is unknown. Given the application and enduse environment, it was determined that a removable water soluble flux chemistry was necessary. As a result Highly Accelerated Stress Testing (HAST) was performed to asses the potential for such a process to provide adequate long term reliability. The matrix of samples involved in the HAST exposure were subjected to 168 hours of unbiased Corrosion in Electronic Hardware (continued from page 6) HAST at 130°C and 83%RH at 1.27 atms vapor pressure based upon MIL-PRF-38535E and JESD22-A110-B. One test board treated with the water soluble process was analyzed by ion chromatography (Assembly 6) and displayed acceptable levels of chloride, bromide, nitrate and sulfate. The peak at 4.5 minutes observed earlier was also observed on this board but relative levels of this peak indicate it was not excessive (Table 1).

The HAST exposure did not initiate corrosion on the ACI-recommended water soluble processed assemblies.

Conclusions/Recommendations:

Choice of components was initially identified as the cause of the failures. The complexity of problem was not initially clear and subsequent analysis identified ionic residue contamination from the secondary noclean hand soldering process as a related cause given the application and end-use environment.

Bromide levels were high for the bare board which would be expected for a HASL finish as a highly active flux is utilized during the hot air solder leveling (HASL) process. However, corrosion was only observed at the leads of the CRT and resistor suggesting the bare board residue was not the key factor in these failures.

A number of unknown peaks were observed from analysis of the assemblies. In some instances these peaks were significant relative to the overall level of ionic residue observed. The major unknown peak at 4.5 or 4.8 minutes was not observed on the in-process assemblies prior to attachment of the CRT and resistor. The presence of this peak after secondary hand soldering indicates it was part of this processing step, most likely a weak organic acid from the no-clean flux.

A no-clean flux is designed to burn off during soldering removingsurface oxides and provide a clean solderable surface with no residue or some benign residue left behind. No-clean flux chemistries historically do not solder as well as activated fluxes. As a result, hand soldering habits are to utilize more flux to get the necessary soldering conditions. This practice and the non-uniform heating of hand soldering generated the residue underneath the CRT. The presence of significant flux residue in a higher voltage application and uncontrolled end-use environment provided a situation for corrosion to occur.

Because of the application and possible end-use environments, ACI recommended utilizing a water soluble flux chemistry that can be cleaned for attachment of the CRT housing and resistor. Limited testing has demonstrated that a water soluble flux if removed properly will not induce corrosion. Application of a conformal coating was also suggested to further extend the usable lifetime of this assembly and would be able to be qualified by ACI if such an additional processing step is a prudent option from a cost/benefit standpoint.

EMPF was further tasked to provide support by doing the CRT attachment through the same water soluble hand soldering step identified during HAST exposure. The cleanliness of the assemblies is currently being determined and to date assemblies have been well below the 10.06 μg NaCl equivalent/inch2 limit prescribed in J-STD-001. This data will be supplied to the customer and used as a baseline for further technology transfer to allow for on-site processing of assemblies by the customer or contract manufacturer if so identified.