Working for manufacturer engineering couple years. Sharing the manufacturing experience, skill, DFx, design, and information. Be care of this blog is personal share and content may not be 100% correctly.
Does the Gold Thickness in ENIG PCBs Affect Component Detachment?
Do you know how the thickness of the gold layer in ENIG surface-finished PCBs impacts soldering quality? What role does the gold layer play in PCBA soldering? How could the detachment of electronic components from a PCB be related to the gold thickness in ENIG? Should the gold layer be as thick as possible?
For a while, the SMT contractor managed by Workingbear got into a dispute with the PCB manufacturer. Both parties were debating the gold thickness specifications for ENIG (Electroless Nickel Immersion Gold) surface-finished PCBs. The issue arose because the EMS (Electronic Manufacturing Services) factory recently used a batch of ENIG PCBs, and during the final assembly after surface mount and board-level testing, components were found detaching from the PCB. Initially, the SMT factory strongly believed that the detachment was caused by the “black pad” issue commonly associated with ENIG. From the appearance, the pads where the components had detached showed the characteristic black pad color. Most of the pads had come off the board entirely, sticking to the component leads. The theory was that the solder fracture might have occurred at the Electroless Nickel (EN) layer or the phosphorus-rich (P-rich) layer.
Since our company outsources the entire production to contract manufacturers, they are responsible for quality control. However, outsourcing can sometimes lead to confusion, especially when it comes to assigning blame and handling compensation. In this case, we witnessed several rounds of back-and-forth between the SMT contractor and the PCB manufacturer. The SMT side insisted it was a black pad issue, citing cross-section analyses and EDX/SEM results showing slightly elevated phosphorus (P) levels. Meanwhile, the PCB manufacturer countered with their own analysis, which showed phosphorus levels within normal limits. The SMT team claimed the gold layer was too thin, measuring less than 1.0µ”, while the PCB manufacturer argued that the gold thickness doesn’t significantly impact soldering strength. Despite all the back-and-forth, no one took the time to conduct a proper analysis to determine exactly where the components had delaminated. Was it poor IMC formation? Inadequate soldering heat? Or was it oxidation of the EN layer that weakened the bond?
Eventually, the situation became so problematic that we couldn’t ship our products. In the end, we had to step in, bring both sides to the table, and hold a joint meeting to resolve the issue.
First, we need to assess the current situation and determine whether component detachment is occurring solely during the final assembly (Box Build) stage. This stage involves the actual insertion and removal of I/O components, while earlier SMT production and ICT testing showed no issues. After inspecting both faulty and non-faulty PCBAs, we found that good boards could withstand 6–8 Kg-f of pushing force without component detachment, whereas faulty boards experienced detachment with less than 2 Kg-f of force.
As a short-term solution, we sorted good and faulty boards by applying push force tests. However, any components that underwent this force test needed to be re-soldered by hand to ensure that microcracks didn’t form at the solder joints. For finished products that had already gone through final assembly, we faced a bigger challenge. We decided to perform 100% insertion/removal tests on the most recent batch, followed by AQL 0.4 testing to check component push force. For other batches, we conducted 100% insertion/removal tests and randomly tested 2 units per pallet for push force. This was quite a big undertaking!
To figure out the root cause of component detachment, we examined the break points, which usually provided clues to the issue:
If the component leads had no solder wetting on them, it was likely due to oxidation of the leads or issues with the solder paste.
If no IMC (Intermetallic Compound) had formed, it suggested that the reflow heat was insufficient.
If the break occurred at the IMC layer, we needed to evaluate whether the IMC formed properly. If the design was correct but IMC formation was poor, reflow temperature might have been too low.
If the break occurred between the IMC and the nickel layer, checking the phosphorus-rich layer is key. Elemental analysis should confirm whether the phosphorus content is too high. If the phosphorus-rich layer is too thick, it can affect long-term reliability and weaken the structure. Additionally, oxidation of the nickel layer could also lead to weakened solder strength.
The following images show the problematic PCBA. We took cross-sections from the pads where components had fallen off, as well as from pads where components remained intact. Additionally, we took a cross-section from a previously produced PCBA that had no issues, focusing on the pad where a component had now fallen off.
Directions
Picture of cross-section
This image is a cross-section of the problematic PCBA at the pad where the component fell off.
You can clearly see that the IMC didn’t fully form. There are still faint traces of round AuSn and AuSn2 that didn’t have a chance to dissipate (though I didn’t do an elemental analysis, so I can’t say for sure). (For more details: The Role of Immersion Gold and Electroplated Gold in PCB Soldering)
This image is a cross-section of the same problematic PCBA, but at a pad where no component fell off.
We found that the IMC on the pad grew normally, and the gold layer had fully blended into the solder.
This Image is a PCBA from a previous batch that had no issues.
Checked the same position where a component had fallen off. The IMC growth there was normal as well.
After several days of discussion and investigation, the real issue began to surface. It turned out the components were detaching between the IMC and the nickel layer, with the IMC growth proving to be insufficient. Both sides detected the presence of oxygen (O) in the nickel layer. One side insisted that nickel erosion, often referred to as “black pad,” was the culprit, while the other side attributed the problem to nickel oxidation. Although we suspected the PCB manufacturer wasn’t revealing the full picture, they did acknowledge that there were issues with their process. They discovered irregularities in the control of one of their gold baths and accepted full responsibility for the losses, so we decided not to pursue the matter further.
As for nickel erosion and nickel oxidation, they seem to be inversely related to the thickness control of the gold layer. It seems Workingbear still has a bit more to learn!
Be Noticed: Workingbear’s opinion here is just for reference. If any PCB experts are reading this, feel free to share your insights. According to IPC-4552 standards, the recommended thickness for the immersion gold layer is generally between 2µ” and 5µ”, with the electroless nickel layer ranging from 3µm (118µ”) to 6µm (236µ”). However, the gold layer should be as thin as possible to avoid issues like gold embrittlement and reverse corrosion since gold is a non-reactive element during soldering. But if the gold layer is too thin, it won’t fully cover the nickel layer, which can lead to oxidation and soldering problems if the board sits in storage for too long. So, the main purpose of the gold is to prevent oxidation. As for the role of nickel, you can refer to this article: ” The Purpose of Nickel (Ni) Plating on Components or Circuit Boards in the Electronics Industry“
Due to the recent spike in gold prices, the thickness of the gold plating on our ENIG-treated PCBs has been reduced from the original 2.0µ” to just over 1.2µ”. That’s pretty thin, and since PCBs are sometimes stored for three to six months—or even over a year—we’re a bit concerned. Honestly, we’re still keeping a close eye on whether such thin gold plating will cause any issues. However, since management has already agreed with the supplier and made this decision, all we can do is wait and see.
The faulty PCB in this case had been stored for about three months. The gold layer on the defective boards measured around 1.0µ” or even thinner. According to the final 8D report from the PCB manufacturer, the issue was that they used a 2mm x 2mm square as the measurement standard for controlling gold thickness. But the problematic pads were actually much larger than that, meaning the gold thickness on those pads wasn’t properly controlled. This resulted in some boards having insufficient gold thickness, leading to nickel layer oxidation and, ultimately, weakened soldering strength. That’s part of the PCB supplier’s response, though I still have some lingering doubts.
Leave a Reply