Homebrew PCBs

Earl T. Hackett, Jr. wrote:
> This is getting a bit off topic.
> 
> The comments below are close and when I started working with photoresists that was my opinion as well.  However, when we examined the results of resolution testing, the data showed there was something other than just columination and exposure at work.  Tests at DuPont showed that oxygen migration within the photo resist from under the opaque areas of the phototool had a major effect on resolution of the spaces.  Oxygen inhibits free radical polymerization.  High intensity light generates so many free radicals in the clear area of the phototool that dissolved oxygen is consumed almost instantly and polymerization proceeds quickly.  As oxygen migrates from under the opaque areas of the photo tool those areas loose the only thing that prevents polymerization.  Stray light from poor columination or reflection from the copper surface would initiate polymerization in what should have been shadowed areas.  If you get the clear areas polymerized quickly the dissolved oxygen doesn'
t have time to migrate and it prevents polymerization in areas of lower light intensity.  In other words, high intensity will overcome the effects of poor columinization.  The reverse is not necessairly ture because of scattered reflection from the copper surface.  This effect was demonstrated in dozens of production shops.  
> 

This is the first time I've heard of this effect on dryfilm resists. The 
question now remains, what is the speed the oxygen diffuses near the 
boundaries between the exposed and unexposed areas of resist. Obviously 
with perfect collimated light, then the limiting factor of resolution 
would be due to oxygen migration effect. But as you mention this can be 
reduced by faster exposures.  Without any numbers its impossible to know 
the amount of photoresists that gets polymerized beyond the opaque mask.
Are you saying that the oxygen migration is so significant that 
fluorescent tubes place at 5 cm distance with an exposure time of 20 
seconds will produce the similar light undercut to a 5 minute exposure 
of collimated light ?
My own experience does not show this to be true at all. The oxygen 
migration effect would appear to be insignificant when producing images 
with 8 mil tracks/spaces or larger. At 2 or 1 mils then it could be a 
different story.
In any event, the only situation a collimated light source would not 
provide noticeable improvement over non-collimated light source is when 
the oxygen migration effect becomes very appreciable.