Clive Ashmore

When print results are out of spec, board thickness could be to blame.

By now, you know my mantra: proper inputs result in proper outputs. Nothing could be more accurate in relation to the printing process, especially today’s printing process where smaller dimensions are the norm and finer solder paste deposition the challenge. Though mainstream production hasn’t yet fully embraced the metric 03015 or metric 0201 in high volume, it’s only a matter of time. Of course, our team is constantly evaluating printing techniques for next-generation devices, and a recent analysis on this subject only served to underscore my input/output contention.

Larger required solder paste deposit volumes with more normalized area ratios (in the standard 0.66 range) lead to a generally more forgiving solder process. Accuracy and precision are still important, to be sure, but a little wiggle room remains. On the other hand, area ratios in the range of 0.5 down to 0.4 leave little room for error. This was made quite clear during a recent evaluation our company conducted. In an effort to better understand how far we could push the print process for metric 03015s and metric 0201s, 20 sample boards that contained 100 of these devices were printed. The apertures sizes ranged from 160µm to 200µm, and the stencil was 80µm thick, yielding an area ratio of 0.5. A Type 5 solder paste was used to adhere to the five particle rule (see my August 2015 column). As is standard practice, each board was numbered to ensure they were processed in the same order for each run, and the activated squeegee and stencil were also labeled.

The experiment was progressing as planned and producing some insightful, stable data. But when the 15th board was printed, the process dropped like a rock.
Until that print, the results were quite good actually. Transfer efficiency was at 78%, with a standard deviation of 8%, very robust for dimensions this small. The board in question, however, had material transfer efficiency between 10% and 20%, with a standard deviation of about 40% – a completely unacceptable result and clearly an anomaly. After the 15th board, the process recovered to normal levels. A second run of the same board sequence produced similar results, with the one outlier board again showing poor transfer efficiency performance.

My first instinct was this had to be an alignment issue, and perhaps the inspection system wasn’t seeing the paste because of poor alignment. Strike one: alignment was perfect. The next step was to make sure the integrity of the squeegees blades hadn’t been compromised in any way. Strike two: squeegee blades had no issues. Still convinced this had to be a fiducial issue of some sort, I put the board under a coordinate measuring machine (CMM) and measured the fiducial distance to the edge of the board, rechecked the dimensions of the board and scanned its topography. Strike three: it all checked out fine. Not one to go down in defeat, I measured one final dimension: the board thickness. Finally, I had my answer.

The boards arrived in a large shipment from our supplier and were supposed to have been 1mm-thick substrates. The board in question measured 0.8mm thick. In the micron world, that’s a 200µm difference and has a significant impact when printing very small deposits. With an advanced print platform set to run an extremely tight process with area ratios of 0.5, the 200µm difference in thickness resulted in noncontact printing. With noncontact printing, when the board rises to the stencil, there is a gap. During the moment of the squeegee filling the apertures with paste, there is contact because the stencil has some malleability and the squeegee is pushing down on it with contact being made. As soon as the blade traverses and moves in the print direction, however, the stencil pops back off the board, with some material remaining in the apertures. It’s almost a tearing action, as opposed to a clean, one motion release. In this evaluation with metric 03015s and metric 0201s, that 200µm interspace created a dramatic difference in printing results.

While to mainstream manufacturers printing this small and precise still seems like a bit of a technology outpost, it is assuredly where the industry is headed. The importance of measuring and understanding the variability potential for all print process components is absolutely critical.

Clive Ashmore is global applied process engineering manager at ASM Assembly Systems, Printing Solutions Division (asmpt.com); clive.ashmore@asmpt.com. His column appears bimonthly.

Submit to FacebookSubmit to Google PlusSubmit to TwitterSubmit to LinkedInPrint Article