Is printing truly responsible for most end-of-line defects?

Do solder paste printing defects affect end-of-the-line defects? After more than 25 years of SMT process development, especially the solder paste printing process, there is still no consensus among process engineers on this point. Opposing statements such as “50 to 80% of the end-of-the-line defect can be directly contributed to print defect” and “print defects can’t be tied to any end-of-the-line defect reliably” are prevalent. Are these just gut feelings or do valid statistical data support these statements?

Opinions on solder paste printing defects and their effects on end-of-line defects depend on how defects are monitored and measured. My good friend and former colleague Joe Belmonte, now with Bose, says the notion that two-thirds of the defects contribute to the printing process “is driven by how operations count opportunities for defects. Most operations count each component lead as a defect opportunity and the component itself as one defect opportunity. For example: A 208-pin QFP has 209 defect opportunities, one for each of the 208 leads, plus one for the component. The one for the component covers for misplacement, reversed polarity, bent lead, etc. Using this approach, the vast majority of defect opportunities (well over 80%) are related to the printing and soldering processes (reflow, wave, selective soldering, hand soldering, etc.).”

In addition, the method used to identify defects can play a significant role in the numbers game. This is evident in a 2002 Nokia and Agilent study.¹ The study showed 2000 pads with less than 65% of the nominal paste volume were not detected as defects by the SPI, whereas 46 of those 2000 pads were detected as defects by AXI testing after reflow. This shows how the inspection method can contribute to the defect rate.

In today’s SMT line, process detection ranges from sophisticated solder paste inspection to no inspection or monitoring. On one extreme, manufacturers inspect 100% and rework the boards, even if only one print deposit fails to meet specifications. The other extreme is no inspection at all.

At this point in SMT history, one would expect statistically valid data exist that clearly define the relationship between printing process defects to defects that occur at the end-of-line, post reflow. In other words, if a solder short (bridge) is detected by the SPI, one should be able to conclude that an actual bridge will be detected at that location through various sophisticated inspection systems. Looking at it in a different way, printing rework can be greatly reduced if it can be determined that a certain percentage of alignment offset is forgiving because of solder pull-back. This will be highly beneficial to manufacturers in increasing first-pass yield.

A limited number of studies have been conducted to provide these data. However, industry still has not universally accepted the conclusions from these studies. Each organization is still on its own with regard to solder paste printing process inspection.

Adding to the complexity of the inspection method, the type of components under consideration also makes a significant difference in the numbers game.

Some components are more forgiving than others when it comes to solder paste volume variation and stencil alignment issues. This was shown in a recent study published by Speedline. It was conducted under a controlled environment to induce defects, both in paste volume and X-Y alignment offset. The volume defects were introduced by over- and underprinting the pads in a systematic way for various types of components (passives ranging from 0201 to 1206, BGAs and QFPs). The X-Y offset defect was introduced for passive components only by designing the stencil to print up to 50% off the pad.

The result was somewhat surprising, yet comforting. The alignment offset study (for passives only) showed up to 97% solder pull-back, without any tombstoning and skewed components. The result was verified through SPI, optical imaging and AXI. The paste volume variation was equally dramatic, based on the component type. Passive components showed no visual (both optical inspection and AXI verification) defects, such as bridging, tombstoning or missing components. The result for QFP was quite different. Solder paste volume over 125% of nominal value consistently showed both wet paste bridging and bridging after reflow at the same location. This indicates that when it comes to excessive paste, QFPs are less forgiving in nature than passives. It also indicates SPI can be used to predict end-of-the-line defects for QFPs.

The question remains, Are we there yet? The answer is no! The aforementioned study is highly encouraging in taking a step toward answering the debate over the percentage of SMT line defects that stem from print defects. But we are not there yet. More cooperative studies among printer and paste manufacturers and OEMs are necessary to fully understand and quantify the correlation between print defect and end-of-the-line defects.

Reference

1. S. Oresjo and V. Chatrath, “Paste Inspection Study,” IPC Apex Proceedings, February 2002.

Rita Mohanty, Ph.D., is director advanced development at Speedline Technologies (speedlinetech.com); rmohanty@speedlinetech.com.