current location：MLC Fine Lines at High Yields
Etch uniformity has been improved in a variety of ways:
? Copper foil thickness variation has been reduced below the customary +/-10%.
? Board center to periphery variation due to puddling has been minimized through selective timing of sprays. Vertical processing was no answer to puddling since etchant run-off creates a "vertical puddle".
? Flipping boards halfway through the etcher reduces top to bottom variation.
? The localized variation of the spray impingement due to shadowing has been addressed by the use of thin spoked wheels and staggered or oscillating thin core guides.
? Overetching of isolated lines can be remedied through selective artwork compensation.
The undesirable side-effect of horizontal (sideways) etching is often quantified by the "etch factor" (F), frequently defined as the ratio of horizontal etching (under-cut) to vertical etching. It is well known, that this ratio becomes more favorable with the degree of overetching. Therefore, to compare results, "F" needs to be referenced, e.g. for the degree of etching needed to match the width of the bottom of the copper trace with the resist width. In one study4 the etch undercut on one ounce (1.4 mil) copper was 0.5 mil per side on a 3 mil line, or F=0.35. The author concludes that under the best of circumstances, variation of 3 mil L/S etched lines is about +/- 8%. This does not include variation introduced by phototools and development since the study deliberately excludes such contributions by normalizing data vs. developed resist width.
Industry representatives composing IPC's National Technology Roadmap felt that today's mainstream production achieves etch factors of 0.3, and 0.25 for leading edge technology. However the degree of over-etching is not clearly defined. Variations in etchant chemistry have not significantly improved etch factors. The preferred foil for vertical etching (1.1.1- crystal structure) cannot be manufactured reproducibly.
Two other studies on etch uniformity are worth mentioning:
? A joint study in 90's between a dry film and an equipment supplier5 concluded: for 4 mil lines and 5 mil spaces, after improvements in the etcher design, etched line widths could be maintained within about 10%. Before etcher optimization more than 50% variability was attributed to etching and not to all other upsteam steps. Etcher variability was cut in half after design changes.
? Another study1 tried to define the most cost effective etcher controls to achieve 3+/0.5 mil etched lines for a 3 sigma population. The study identified the most important process variables, measured etched line width responses to variable setpoints, and defined control requirements to maintain 3+/-0.5 mil. Free acid content and etch time were most critical. Control feasibility and cost were assessed and found to be viable. Since the study changed "one variable at a time", control limits were cut in half to allow for two variables simultaneously drifting to the extremes of the allowed ranges. If three variables approached outer limits, process logic forced a shut down. The study concludes that +/-0.5 mil control of a 2 mil L/S pattern is feasible.
Variability in etched line width is not dependent on the final line width but on process variability creating the space. Thus, for narrower lines, % variability of the etched line increases.