[SI-LIST] Re: Skin Effect, Au, and Cu

Dear Eric,

As an academic question, where would be a good place to look for
permeability vs. frequency data for immersion nickel and a whole host of
other materials?=20

Best Regards
=20
David Greig
______________________________
GigaDyne Ltd
Buchan House
Carnegie Campus
Dunfermline KY11 8PL
United Kingdom
t: +44 (0)1383 624 975
______________________________

-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] =
On
Behalf Of eric bogatin
Sent: 23 February 2005 13:34
To: si-list@xxxxxxxxxxxxx
Cc: eric@xxxxxxxxxxxxxxx; susan@xxxxxxxxxxxxxxx
Subject: [SI-LIST] Skin Effect, Au, and Cu

Hi folks-
=20

Thought I would chime in on this topic as I have a section related to it =
in
my book (available with a click through from my web site). I also have a
number of my past PCD&M columns on this topic of current distribution in
conductors, both due to skin depth and in a differential pair, that are
available for download from my website, www.bethesignal.com
<http://www.bethesignal.com/> =20

=20

Here=92s an important rule of thumb to keep in mind: @ 1 GHz, the skin =
depth
for copper is about 2 microns. It will get thicker if the resistivity
increases, and thinner if the permeability increases. For gold, it is =
about
2.5 microns, and for nickel, depending on the permeability, thinner than =
2
microns.

=20

The thickness of =BD oz copper is 17 microns, for typical electroless =
nickel
which is about 100 microinches, it is about 2.5 microns and for =
electroless
gold, typically 40 microinches, about 1 micron. In a typical surface =
trace
metal stack, there is a layer of about 17 microns thick copper, with =
about
2.5 microns of nickel on top and 1 micron of gold on top of that.

=20

The gold is there to provide an oxide free surface so the board can be
compatible with either a =93dry=94 electronic connection, such as a =
surface
mount connector or edge card connector, be wire bondable, or solderable =
with
wide process windows. If the adhesion between copper and gold was =
better,
and there was no long term reliability problem with copper diffusing =
into
the copper, there would be no need for the nickel. The nickel acts as a =
glue
layer and a diffusion barrier layer for the copper to prevent it from
migrating into the gold and oxidizing on the surface.

=20

If the gold is thin, or wears off, the nickel will be exposed and it =
will
oxidize and solder will not wet it. Typically, when solder reflows on =
the
gold pad, the gold is dissolved away in the tin of the solder and the =
solder
joint is actually to the oxide free nickel layer below. This is why the =
gold
needs to be thin- thick enough to prevent the nickel from oxidizing, but
thin enough to dissolve quickly and not have so high a gold =
concentration in
the solder to cause it to be brittle. Many a failed BGA solder joint is =
due
to too thin a gold layer or a porous gold layer that allowed the =
underlying
nickel to oxidize and created basically a cold solder joint which was =
easily
cracked into an intermittent open.

=20

When everything is working, the copper is a great conductor, the nickel =
is a
great diffusion barrier and the gold is a great oxidizing barrier. Each
piece has to work in concert, playing their parts. This is why I have so
much respect for the fab industry and good chemical engineers. They have =
to
work hard to make all the chemistries play nice together.

=20

Given this stack up on the surface, think about the current path through =
the
conductors. Remember, the key feature that influences the path the =
current
takes is the path of lowest loop impedance between the signal and return
currents. To first order, this translates to currents within each =
conductor
that want to go to the outside to minimize the internal partial self
inductance and the currents between the signal and return paths will =
want to
get as close together as possible, to maximize their partial mutual
inductance.

=20

While the currents in the microstrip signal conductor will want to move =
to
the outside surface, they are not going to flow very much through the =
nickel
layer- this will have too high an impedance They will be either in the =
gold
layer above, or the copper layer below. Above about 10 MHz, all the =
currents
in a copper trace are skin depth dominated. At 1 GHz, where concern for
series resistance might begin, currents in the signal path are in the =
outer
2 microns of the trace. Most of them are on the bottom surface of the
copper, adjacent to the return plane. The rest will be in the gold. Keep =
in
mind that at 5 GHz, the skin depth of gold is 1 micron. Virtually all =
the
currents above this frequency will be in the gold. The nickel will play =
very
little role in the current distribution of surface traces.

=20

Keep in mind also that in FR4 interconnects, for traces wider than 8 =
mils,
series resistance is always a lower loss mechanism that the dielectric =
loss.
It is in very low loss substrates, such as Rogers Duroid materials, and
narrow lines, where the series resistance plays an important role,
especially at higher frequencies, where the surface roughness might be
importance.

=20

The only way to really know if the series resistance is an issue is to =
put
in the numbers. There are a variety of simple approximations that can =
help
you analyze your problems (many of them, in my book- shameless plug!). =
The
value of approximations is just as much in identifying what is important =
and
may be a first order problem, as in identifying what is not a problem =
and
should not distract you from worrying about the big factors that will =
make a
significant difference in your design. After all, almost every design is
custom. Each engineer has to be responsible for analyzing the problems =
in
their design and not rely on the interpretation of generalities. This is =
the
origin of most of the myths in signal integrity.

=20

Finally, if you identify that series resistance, for example, is =
important,
and the performance of your design will depend on the current =
distribution
and role of the different metal stack ups for your specific design, it =
will
be of value to run a few simulations using one of a number of 2D field
solvers that solve for the frequency dependent current distributions. An
example is the Ansoft 2D tool, part of their Q3D product.

=20

Hope this helps.

=20

--eric=20

=20

**************************************

Dr. Eric Bogatin

www.BeTheSignal.com

Signal Integrity on-demand training

26235 w 110th terr

Olathe, KS 66061

v: 913-393-1305

f: 913-393-0929

c:913-424-4333

e:eric@xxxxxxxxxxxxxxx

=20

Signal Integrity-Simplified

Prentice Hall, 2004

****************************************

=20


------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field

List FAQ wiki page is located at:
                http://si-list.org/wiki/wiki.pl?Si-List_FAQ

List technical documents are available at:
                http://www.si-list.org

List archives are viewable at:    =20
                http://www.freelists.org/archives/si-list
or at our remote archives:
                http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
                http://www.qsl.net/wb6tpu
 =20

--
Virus scanned by Lumison.


------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field

List FAQ wiki page is located at:
                http://si-list.org/wiki/wiki.pl?Si-List_FAQ

List technical documents are available at:
                http://www.si-list.org

List archives are viewable at:     
                http://www.freelists.org/archives/si-list
or at our remote archives:
                http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
                http://www.qsl.net/wb6tpu

Other related posts: