Jack

You will not see a comparative on these techniques because the articles are
driven by marketing.  So here's my semi-unbiased overview-

1) Laser drilling is preferred at present over photoimageable dielectrics
for a number of reasons: a) dielectric properties (adhesion, dielectric
constant, shrinkage, etc.) are comprised in making it photoimageable, b)
yield (different via sizes have different responses to the photo process),
c) dielectric application (thickness uniformity, etc.), and d) other issues
I can't think of off the top of my head.  The concept of a mass-via
production process is very appealing, but it is not yet mature and may not
ever be, although I personally like the quality of the work and people
trying it including McDermid, Shipley, etc..

2) Which laser (UV vs. CO2)?
a) The UV laser system (synonomis with ESI) is a finely focused (about
20-30um) beam that is moved around by a set of two mirrors (x and y) in
synchronization with the laser pulse output (approximately 4000 pulses/sec).
It requires a lot of
precision accuracy (i.e., computer control) and mechanical control (lots of
movement).  The rate of via formation is directly related to the drill time
per via.  Since the mechanical movement is limited, the laser can only move
so fast.  So, you have a fundamental limit on throughput.  ESI has been good
at getting the mechanical issues optimal, but it won't go much faster from
my viewpoint.  A higher powered laser will help, but still it can only go so
fast because the laser has to be kept at a power below that which will
damage the underlayer copper.  So, there is only so much throughput ESI can
squeeze out of it.  They are near their max at this point without going to a
multiheaded system (which is not yet offered nor is there reason to believe
it will be).

Because the laser is drilling like a fine pick with the laser ablating at
about 50 different spots to get round via, the throughput depends on the
thickness and type of each layer in a board.  If the top layer is copper (on
top of dielectric-on-copper), it will be slower than a bare dielectric on
copper.  The slowest througput dielectric is FR4 and the fastest being
Speedboard, RCC, thin polyimides, etc. with Thermount type being between the
two.  Not all FR4's can be drilled, and typically 4-5 mil thicknesses are
the most you can hope to drill through.  Also, the thinness of the
underlayer can affect throughput because one has to lower the laser power
not to damage the underlayer.

The via size capability is dependent on the laser and the board material.
Typical safe via sizes in bare resin-based dielectrics is 50 um and up,
although slightly larger than 1 mil can be achieved with effort.  Again,
throughput depends on via size (approximately 1/area scaling for vias above
3 mil).  Bare dielectrics can get up to 100+ vias/s for small (3mil and
under) vias, and the copper-on-top is half that speed.  FR4 is closer to
15-30, and cannot be done bare dielectric style.

All said, the ESI 5000 series is an amazing machine and is fun to watch in
action.  It has a history of upkeep problems in terms of software and
hardware (mechanical and laser).  I compare it to a Jaguar automobile (of
the 1950's genre)- when it runs, its great...but you need a full time
mechanic.

b) The CO2 laser system is a large (~200um) beam that uses 1-5 pulses of CO2
laser light.  Think of these pulses as atom bombs, because they blast the
wazoo out of dielectrics.  What limits the damage to the board is that CO2
light (10 um) is it reflects off of copper.  So, some guy (I think Larry
Burgess...I can look it up if needed) out of Tektronix came up with the idea
of using the top layer of copper to mask the laser, and fabricate laser vias
using CO2.  This is in the mid-1980's and well before ESI had any UV laser
system, and around the time IBM was learning how to ablate polyimide.


Until shorter pulse length (<100us) lasers were used, the via quality looked
like junk.  Nowadays, many lasers produce good vias.  But they have a
problem of leaving residue at the bottom because of a fundamental physics
problem of light interfering with itself on the order of a 1/4 wavelength
(~2.5 um in this case).  The problem is not as bad at it is feared because
it also creates a mini-plasma that can etch out
the remaining resin using it's self-induced heating.  The Lumonics laser (a
TEA-laser) produces extremely clean vias, but has the disadvantage of only
pulsing at ~300 Hz.  This means the throughput is limited to 300 vias/sec.
Lumonics has a two headed machine, so this can get up to 600 vias/s in ideal
cases.  However, there is hole-to-hole move time and some dielectrics
require 2-5 pulses per via...  The math is simple...much slower.  Other
systems from Hitachi, Sumitomo, etc. use a different, RF CO2 laser (I think
Sumitomo also sells Lumonics laser type systems also) that puts out ~50us
pulses at rates up to 10kHz.  The limitation is beam movement, which is not
ideal because CO2 needs special optics (non-quartz) and beam divergence.
The throughputs are in the 200-250 via/s range.

The throughput using CO2's is also dependent on dielectric material.
However, it is less sensitive to variation compare to the UV laser.  For
this reason, the process window is very wide (except with the Lumonics which
is trying to optimize each pulse because of it's low pulse rate).  For that
reason, it's a great process.  In addition, the software/hardware is very
simple and breaks down infrequently.

Its limitation is that you have to rely on copper etch to open the vias.
This is unreliable below 4 mils and has variation in actual via size
openings depending on board location, via density, etc.  It means better
etch process, etc.

3) Lastly:
The general CO2 is in large scale production for larger vias (125-300um)
where it can be used easily.  For smaller vias, the ESI or Lumonics systems
are mostly used because of the confidence in via bottom cleanliness.  But
these tend to be more development/high cost boards (e.g., cell phones) where
the cost can be justified.  Systems that need full optimization (small
process window) also have personality (i.e., no two systems are alike).
This leads to difficulty in board yield on a work floor.

Notice I did not include plasma etch in the discussion.  I wish someone
would correct me when I say that it is not a viable process at this time.



Hope this is more than you want.  Feel free to ask more.

Carey

-----Original Message-----
From: Jack Petyak <[log in to unmask]>
To: [log in to unmask] <[log in to unmask]>
Date: Wednesday, January 12, 2000 4:40 PM
Subject: [TN] Microvia drilling


>Does anyone have any suggestions as to what they
>prefer in creating micro vias? Laser drilling (YAG or
>CO2) or Photo Dielectrics? Pros and cons? Any websites
>or companies to find out more info would be
>appreciated.
>
>Jack Petyak
>EP Technologies
>__________________________________________________
>Do You Yahoo!?
>Talk to your friends online with Yahoo! Messenger.
>http://im.yahoo.com
>
>##############################################################
>TechNet Mail List provided as a free service by IPC using LISTSERV 1.8c
>##############################################################
>To subscribe/unsubscribe, send a message to [log in to unmask] with following
text in
>the body:
>To subscribe:   SUBSCRIBE TECHNET <your full name>
>To unsubscribe:   SIGNOFF TECHNET
>##############################################################
>Please visit IPC web site (http://www.ipc.org/html/forum.htm) for
additional
>information.
>If you need assistance - contact Gayatri Sardeshpande at [log in to unmask] or
>847-509-9700 ext.5365
>##############################################################
>

##############################################################
TechNet Mail List provided as a free service by IPC using LISTSERV 1.8c
##############################################################
To subscribe/unsubscribe, send a message to [log in to unmask] with following text in
the body:
To subscribe:   SUBSCRIBE TECHNET <your full name>
To unsubscribe:   SIGNOFF TECHNET
##############################################################
Please visit IPC web site (http://www.ipc.org/html/forum.htm) for additional
information.
If you need assistance - contact Gayatri Sardeshpande at [log in to unmask] or
847-509-9700 ext.5365
##############################################################