Solving the Mystery of Failed Components
December 31, 1969 |Estimated reading time: 10 minutes
Two keys to success in today’s electronics industry are being data-informed and customer-driven. This article examines one failure analysis/product analysis lab’s mission to explain the unexplained - providing decision-making data in reference to component and product failures.
By Jamie Clawson, Dave Mixon, and Gary Runyon
Being data-informed and customer-driven are keys to success in the rapidly evolving electronics industry. One failure analysis/product analysis lab* aims to explain the unexplained and provide decision-making data in reference to component and product failures. The lab performs scientific root-cause investigations for clients looking to better understand their manufacturing or field issues.
Much like crime-scene investigators, lab personnel use a variety of tools and techniques to determine the root cause of component and assembly failures. A range of factors can cause failures throughout component or assembly life cycles, such as defects related to manufacturing-process control, electrostatic discharge (ESD) protection practices, or handling and storage practices. In addition to physical testing and examination, analysts often must study a component’s history, or perform destructive testing in their search for the truth.
While the lab’s client base includes internal customers, such as its EMS operations and their customers, some of the more unique and challenging requests come from standalone customers (many of whom are not in the electronics industry), who use specific analysis services in product development or component testing. The value of these investigative processes is a rapid path to corrective action. Initial engineering assumptions related to root cause(s) of failure are often proven wrong as tests progress. Likewise, in new product introductions (NPIs), a combination of failure analysis and environmental tests can validate assumptions about the robustness of a design, or indicate required engineering changes. This article outlines some representative issues that lab personnel have investigated, and their methods for performing associated analyses.
Is It Real or Counterfeit?
Failures are not always driven by component- or assembly-level manufacturing defects. Component integrity is becoming a key issue. The following scenarios illustrate challenges faced by manufacturers, OEMs, and distributors when determining if an older component is what it is represented to be.
In the first example, a high failure rate occurred during production testing. Subsequent troubleshooting indicated that the problem was related to a particular integrated circuit (IC). The lab’s initial focus was to determine if these ICs had been damaged during the manufacturing process or during handling. The analysis involved the following steps:
• Perform an X-ray of the IC to determine if an internal defect could be located;
• Enlarge the X-ray and evaluate it for visual defects in the IC;
• Remove the top of the IC case and inspect the part with the metallurgy microscope.
A quick examination of initial X-rays indicated that the ICs had missing and extra bond wires, and that these components were not from an approved lot from an IC manufacturer. As a result, new components were purchased (Figure 1).
Figure 1. Results of X-ray analysis of a counterfeit component.
The second example was even more pronounced. Lab personnel were asked to evaluate some gray market ICs to determine if they were fit for use. An X-ray inspection of internal construction determined that some of the components did not have bond wires or die circuitry, indicating that the components did not come from an accepted lot from the IC manufacturer (Figure 2). New components were purchased from a direct channel.
Figure 2. X-ray of counterfeit components with missing circuitry.
The third example had a different result. As with the second example, a client planned to purchase older gray market components, and wanted die verification to help determine IC integrity. In this case, the client requesting the service was an independent distributor doing due diligence prior to purchasing these parts. In the analysis of three sample components, initial visual observation showed IC laser etching on the top housing with a manufacturer’s logo. The part number showed all three components to be identical RAM devices. An X-ray inspection was performed on all three ICs, and no anomalies were detected. Next, red fuming nitric acid was used for component de-capsulation. Each die was viewed with a metallurgical microscope at 140×. Although the parts were listed as a certain manufacturer’s ICs, the manufacturer’s logo and part number could not be found on the die. Instead, a second manufacturer’s logo, 1995 date code, a different part number, and a copyright were etched on the die. “Made in U.S.A.” and an American flag were also etched on the die (Figure 3). Data sheets on the parts from both manufacturers were then compared.
Figure 3. Die data codes, logos, and special markings can be important to determine authenticity.
The conclusion was that the parts contained a die fabricated in 1995 by the second manufacturer. Data sheets showed this part to be identical to the first manufacturer’s part sent for die verification. Although the internal die logo does not match the external logo, it is not uncommon for companies to sell components to a third party for resale. The distributor was able to sell the parts to their end customer with an accurate assessment of authenticity.
Is Process Control the Culprit?
The fourth example involves analysis of a recurring manufacturing defect. In the sample analyzed, in-circuit test (ICT) data indicated that there were several apparent opens in BGA solder joints. The assembly had failed final test in the initial manufacturing process and had been sent to a separate facility for further testing and confirmation using X-ray solder inspection.** The unit passed the inspection, but failed a subsequent ICT.
An initial microscopy revealed plating defects of vias and pads of the printed wiring board (PWB) assembly. The target BGA and a comparison BGA were cut from the board assembly using an abrasive disc. The outside rows of solder balls, when observed with the microscope, displayed some unusual appearances (Figure 4). The two BGAs were mounted in slow-set epoxy to allow solder ball micro-sectioning. This revealed that in the target BGA, six of eight balls were fractured at the bottom of the ball in the solder joint, above the nickel layer of the PWB pad. Several solder voids were also found, and the tin/lead solder displayed a granular appearance. In the comparison BGA, one ball was fractured between two layers of nickel on the PWB pad.
Figure 4. The outside row of solder balls display some unusual appearances.
Further analysis showed vias in the PWB to be incompletely plated with nickel or gold, allowing exposed copper in the vias. Scanning electron microscope (SEM) and energy dispersive spectrometry (EDS) analysis showed a defined phosphorus peak in one layer of nickel, which has been known to decrease over-plating ability. Solder mask was also noted to flow over one edge of each pad underneath target BGA balls, leading to the root failure of mis-shapen balls and incomplete soldering to one side of each pad.
Full analysis revealed three problems. The PWB had some discrepancies with plating. The double-nickel layers, particularly the top one with a rough surface, appeared to be assisting the failures. The PWB also had solder mask mis-alignment. Finally, insufficient temperature/time of the devices above proper reflow temperature occurred. The recommended corrective action was to work with the PWB manufacturer on plating and solder mask issues, and improve the manufacturing solder reflow profile (Figure 5).
Figure 5. Vias cross-sectioned during analysis were also noted to have insufficient gold and nickel plating over the copper in many locations.
null
Design Qualification
Another area of the lab’s work involves assisting with the design-qualification process. In one example, a customer was performing a qualification for their printed circuit board assembly (PCBA) and wanted solder-quality verification of nine BGAs on an assembly. The lab’s job was to determine the solderability of the BGAs to the assembly, including ball wetting to PWB pads, shape of the balls after wetting, granularity of the solder, and size of any solder voids in the balls.
For the analysis, a diagonal row of balls was selected on each BGA. Using an X-ray of each BGA on the assembly, cutting location was determined by selecting key landmarks on the assembly itself, which guided the analyst to the desired diagonal row of balls. Each BGA was marked accordingly, and BGAs were removed from the assembly and set in an epoxy coupon. Each BGA was then cut along the mark using a precision saw with a diamond blade. The BGAs were X-rayed a second time to determine the location of the cut edge relative to the selected row. They were then polished to the approximate center of the 0.35-mm-diameter balls of the selected row using a polishing wheel, and the location was verified using a metallurgy microscope. Micro-sectioned BGA balls were digitally photographed individually and sent to the customer electronically. The end result of the analysis was that the solder was determined acceptable to IPC-A-610, current revision, and the customer was able to move from prototype build to production.
In the next qualification project, the lab’s job was to determine that the BGA was soldered correctly to the PWB and would remain on the assembly under severe vibration and impact-shock conditions. The analysis involved a shear-strength test and a solderability evaluation.
Initial microscopic examination of six BGAs on six boards of the array failed to show any external defects or anomalies. A force gauge was used for the push test, using a double-chisel fixture; results were consistent within the group of BGAs being pushed off - demonstrating 33.6, 32.8, and 34.2 lb., respectively.
Optical microscopic examination of BGAs and PWBs showed nearly equal distribution of the ball breaking at PWB pads, PWB pads being lifted from the laminate, and balls breaking at the BGA pad (Figure 6).
Figure 6.Stereo micrograph showing PWB of unit after shear testing showing a mixture of balls remaining on the board, pads lifted, and balls lifted from pads.
For the solderability evaluation, sections with BGAs from remaining PWBs on the array were extracted using an abrasive disc. These were set in slow-set epoxy and microsectioned during analysis. The BGAs were microsectioned along selected horizontal, vertical, and diagonal rows, and inspected using a metallurgy microscope. All microsections showed good soldering results of solder balls, with acceptable voids and good wetting. Shear-strength analysis revealed equally distributed breakage of BGAs from PWB pads during shear testing and balls shaped in the expected forms on the microsections. No defects, failures, or anomalies were determined, other than the voids that did not violate specifications for excess size.
Manufacturing Process Development, RoHS after Lead
In this investigation, the lab examined two scenarios: a leaded and a lead-free process. The investigation was conducted to determine if any traces of lead could be detected in the lead-free process. In the second scenario, a tin/lead solder process was run after the lead-free process and analysts looked for traces of silver left over.
Element spectrometry analysis was performed on 12 PWBs where lead-free solder was applied to one pad on each PWB. Using the X-ray fluorescence (XRF) spectrometry, an analysis was performed to check for tin (Sn), silver (Ag), copper (Cu), nickel (Ni), and lead (Pb). Each spectrum was taken for a 10-second duration. EDS analysis was also performed on the same boards. SEM/EDS analyses were performed on the lead-free roll solder with the following results: 1.82% Cu; 2.33% Ag; and 93.43% Sn. No lead peak was visible in the EDS spectra. XRF was performed on raw solder, resulting in 94.5% Sn, 4.6% Ag, 0.4% Cu, and 0% Ni (Table 1).
In the second portion of the process study, elemental analysis was performed on twelve PWBs where tin/lead solder was applied to one pad on each PWB. Using XRF spectrometry, an analysis checked for elements of Sn, Ag, Cu, Ni, and Pb. Results of the XRF spectrum were taken for 10 seconds. Results show traces of lead were below 0.10% acceptable levels for RoHS compliance. No traces of silver were detected using XRF measurements. The conclusion was that the convection soldering process could be changed between lead and lead-free by adjusting the heating profile without exceeding RoHS limits of acceptable lead content.
Conclusion
As electronics manufacturing continues to increase in complexity and sophistication, providing root-cause failure-analysis data in a timely manner to ease decision making or validate design assumptions will continue to increase in importance. Each of the preceding examples involved a combination of engineering expertise, a formal analysis process, and sophisticated testing and documentation capabilities. In some cases, clients involved in time-sensitive evaluations received real-time results as tests were completed. In all cases, product analysis helped project personnel obtain the data they needed to make timely and appropriate decisions related to corrective actions or progression to the next phase of product development.
* EPIC Technologies’ Johnson City TN Failure Analysis/Product Analysis Lab.
** Hewlett Packard 5DX Series II X-ray.
Jamie Clawson is the senior analyst for EPIC Technologies. Dave Mixon is the product analysis lab manager for EPIC Technologies. Gary Runyon is the senior analyst - Product Analysis Lab for EPIC Technologies. For more information, contact Dave Mixon via e-mail: Dave.Mixon@epictech.com.