Reading time ( words)
Initially adopted internal specifications for acceptance of printed circuit boards (PCBs) used for wire bonding was that there were no nodules or scratches allowed on the wirebond pads when inspected under 20X magnification. The nodules and scratches were not defined by measurable dimensions and were considered to be unacceptable if there was any sign of a visual blemish on wire-bondable features. Analysis of the yield at a PCB manufacturer monitored monthly for over two years indicated that the target yield could not be achieved, and the main reasons for yield loss were due to nodules and scratches on the wirebonding pads.
The PCB manufacturer attempted to eliminate nodules and scratches. First, a light-scrubbing step was added after electroless copper plating to remove any co-deposited fine particles that acted as a seed for nodules at the time of copper plating. Then, the electrolytic copper plating tank was emptied, fully cleaned, and filtered to eliminate the possibility of co-deposited particles in the electroplating process. Both actions greatly reduced the density of the nodules but did not fully eliminate them. Even though there was only one nodule on any wire-bonding pad, the board was still considered a reject. To reduce scratches on wirebonding pads, the PCB manufacturer utilized foam trays after routing the boards so that they did not make direct contact with other boards. This action significantly reduced the scratches on wire-bonding pads, even though some isolated scratches still appeared from time to time, which caused the boards to be rejected. Even with these significant improvements, the target yield remained unachievable.
Another approach was then taken to consider if wire bonding could be successfully performed over nodules and scratches and if there was a dimensional threshold where wire bonding could be successful. A gold ball bonding process called either stand-off-stitch bonding (SSB) or ball-stitch-on-ball bonding (BSOB) was used to determine the effects of nodules and scratches on wire bonds. The dimension of nodules, including height, and the size of scratches, including width, were measured before wire bonding. Wire bonding was then performed directly on various sizes of nodules and scratches on the bonding pad, and the evaluation of wire bonds was conducted using wire pull tests before and after reliability testing. Based on the results of the wire-bonding evaluation, the internal specification for nodules and scratches for wirebondable PCBs was modified to allow nodules and scratches with a certain height and a width limitation compared to initially adopted internal specifications of no nodules and no scratches. Such an approach resulted in improved yield at the PCB manufacturer.
The development of the chip-on-board (COB) type of image sensors has significantly increased in the last few years to accommodate an increasing demand for low-cost, high-performance industrial inspection cameras. Wire bonding is the most commonly used technology for making the electrical interconnection between a silicon image sensor and its substrate during image sensor fabrication. Therefore, wire bonding directly on the PCB surface is inevitable for COB-type image sensor fabrication.
IPC-A-600J section 18.104.22.168—surface plating, wire-bond pads —states, “Wire-bond pads are free of surface nodules, roughness, electrical test witness marks or scratches that exceed 0.8 mm (32 min) RMS (root-mean-square) in the pristine area in accordance with an applicable test method AABUS.” Since the tolerance of surface nodules (<0.8 mm RMS) in IPC-A-600J for wire bonding pads is not easily measurable and may not properly reflect single nodules and shallow scratches, the initially adopted specification for wire bonding pads as a user was defined to be free of surface nodules and scratches.
Figure 1: Average yield loss from multiple wire-bonding PCB products.
Therefore, a board was considered a reject if a nodule or a scratch was visible in the pristine area of the wire-bond at 20X magnification. If in any doubt, one could use 40X magnification for inspection to clarify pass/fail. However, in reality, it was difficult to achieve a wire-bonding surface in specification without nodules or scratches, resulting in low yield at the PCB manufacturer. In particular, PCB products designed with a center cut-out slot close to bonding pads were prone to significant yield loss due to nodules.
Yield review with the manufacturer and the analysis of causes for rejects using Pareto charts pointed out that two main reasons for PCB rejects were due to nodules and scratches on the wire-bonding surface. Yield had been reviewed on a monthly basis for 28 months, and the major yield losses continued to be due to nodules and scratches on the wire-bonding surface during the period. Figure 1 shows the average yield loss per month for wire-bonding boards due to nodules and scratches. The average yield losses from multiple PCB products due to nodules and scratches for wire bonding boards were in the range of 4–41% and 2–30%, respectively, during the period.
It is worth noting that the data in Figure 1 may not necessarily provide the picture of actual average yield loss due to nodules and scratches. When any defect—either nodule or scratch—was first found during an inspection of a board, the reason for rejection was recorded and further inspection of that board was not pursued since the board was already determined to be a reject. For example, the manufacturer recorded a 39% yield loss for nodules and none for scratches in Period 3. However, it may not mean that there were no scratches on those rejected boards since those boards had not been inspected further for scratches after nodules were first observed.
To read the full article, which appeared in the July 2019 issue of SMT007 Magazine, click here.