-- [ From: Doug Jeffery * EMC.Ver #2.10P ] --

Friends,

We have seen many designs that have smaller webs between SMP's than
they have circuits.  Imagin creating a .003" line in Soldermask yet the
board has .006" traces.  We have found that any Soldermask web design
that is .005" of web by design is reproducable and  can be placed
reliably, but below .005" the LPI masks do no hold on.  
The key reason is undercut. The  LPI between SMP's is thicker than
anywhere else on the board.  This requires that the exposure be set up
to accomadate the thicker material, however overexposure can cause
other feature problems.  After you have optimised the exposure you play
with the developing until you keep a good dam between pads, Bingo you
are leaving ink in the holes.   So,  you crank up the developer and get
that ink out of the holes....Catch 22...

AT .005" feature size (dams) you are able to optimize the process,
below .005" dams the undercut takes the foot of the LPI down to .003"
or .002" which makes the adhesion a problem, hense the peelers and
redoposit problems that you expressed.  

?What is the answer for .020" pitch devices that require .014" pad
widths?

?what is the answer for .006" lands on .012" centers?

I don't know but certainly we have to get to one.  We have tried double
coating, Unpigmented material, reduced pigmented material..No
significant result differences.  We try to get customers to leave use
.009" min space between SMP's to keep a dam (2/5/2 by design.  This
makes LPI exposing a tighter regitration than our outerlayer
requirements.