Design For The Envirorirnerit Project
Releases Direct IMetallization
Performance Resiolts
By BiH Birch, Susan Wlansilla, Katlty Hart and Deborah Bogef
he U.S. Environmental Protection Agency
(EPA) Design for the Environment (DfE)
Program has been working closely with the
Institute for Interconnecting and Packaging
Electronic Circuits (IPC), its member compa-
nies and other interested parties to conduct a
comparative study of the performance of
alternatives to the electroless copper process
for making PWB holes conductive. PWB pap-
els designed to represent industry "middle-ot-
the-road" technology' were manufactured at
one facility, run dirough individual "making
holes conductive- (MHC) lines at 25 facih-
ties, then electroplated at one facility. The
panels were electrically pre-screened, fol-
lowed by electrical stress testing and ffltfchan-
ieal testing, in order to identify variability m
performance of the MHC process. The test
methods used to evaluate performance were
intended to indicate characteristics of a tech-
nology's performance', not to define parame-
ters of performance or to substitute for thor-
ough .on-site testing. This study was.ihtended
to be a "snapshot" of the technologies. This
' study was carried out as part of. the U.S. EPA-
IPC DfE PWB Project and was, conducted with
extensive input and participation from PWB
manufacturers, their suppliers, and PWB test-
ing laboratories, i
The studv is now complete and the results
suggest that-direct metallization (DM).techr
nologies perform at least as well as electroless
copper if operated according to specifica-
tions.
The electroless copper through-hole
process is die technology most commonly
152' CIJCIITBEE J»««»IT
-------
.v ;. ' :.. ,.,.: i'UB r.i-H
. "IlilIU ".'.'' IJIl'M '.) df( U'»K tie
riu-. r«-iinoli>irv upicallv einplovs
tormaldehuir as a copper-reducing
agent, requires large amounts of water
and energy and is a significant source of
hazardous waste. There are alternative
technologies available, most of which
eliminate the use of formaldehyde,
reduce water and energy use and gener-
ate less waste.
The potential environmental and cost
advantages of the alternatives are begin-
ning to become -apparent and have gen-
erated strong interest on the part of
industry. However, minimal product per-
formance information is publicly avail-
able on these technologies. The project
was conducted to address this data gap.-
The technologies tested include electro-
less copper, carbon, graphite, palladium,
non-formaldehyde electroless and con-
ductive polymer.
The resulting information will be
included in a document titled Cleaner
Technologies Substitutes Assessment
(CTSA), to be completed by early 1997.
The CTSA will allow business decision-
makers to evaluate the alternatives in
terms of human health, environment,
cost and performance.
One hundred 8 layer test panels mea-
suring 24" x 18", laminated to .062"
thickness, were produced from B and C
stage FR4 materials. Each test panel con-
tained 54 test coupons consisting of 27
Interconnect Stress Test (1ST) coupons
(used for electrical testing) and 27
microsection coupons, 1ST coupons
measure 6.5" x 3/4" and contain 700
interconnecting vias on a 7 row by 100
via 0.050" grid. This coupon contains two
independent circuits: a post circuit
which contains 200 interconnects, used
to measure post interconnect resistance
degradation, and a PTH circuit which
contains 500 interconnects, used to mea-
sure PTH (barrel) interconnect resis-
tance degradation: 1ST coupons had
either .013" or .018" holes (finished).
The microsection coupon measures 2"
x 2" and contains 100 interconnected
vias on a 10 row by 10 via 0.100" grid. It
has internal pads, at the second an4 sev-
enth layer and a daisy chain'interconnect
between the wo surfaces of the coupon
through the via. Microsection coupons
had either 0.013", 0.018" or 0.036" holes
(finished).
For the purposes of this evaluation,
the MHC process was defined as every-
thing from the end of the desmear step
through 0.1 mil of copper flash plating.
156 CUCIITHE JIII1IT 1117
The potential
environmental and
cost advantages of
the alternatives
are beginning to
become apparent
In order to minimize differences in per-
formance due to processes outside this
defined MHC function, the panels used
for testing were all manufactured and
drilled at one facility.
After drilling, three panels were sealed
in a plastic bag with desiccant and
shipped to each*of 25 test sites to be
processed through the site's MHC line:.
All bags containing panels remained
sealed until the day of processing.
The facilities that were used as test
sites were identified by suppliers' of the
technologies being tested and included
production facilities, testing' facilities
(beta sites), and supplier testing facilities.
An on-site observer was present at each
site to confirm that all processing was
completed according to instructions and
to record observations. Each test site's
process was completed within one day; it
took slightly longer than one month to
complete all MHC processing.
After the MHC processing, the panels
were put into sealed bags with desiccant
and shipped to a single facility, where
they remained until all the panels were
collected. At this facility the panels were
electroplated with 1.0 mil of copper fol-
lowed by a tin-lead etch, resist, etched,
stripped of tin-lead, solder mask coated
and finished with hot air solder leveling
(HASL).
After HASL, the microsection coupons
were routed out of the panels and sent to
Robisan Laboratory Inc. for mechanical
testing. The 1ST coupons were left in
panel format. The panels containing the
coupons were passed twice -through an
IR reflow to simulate assembly stress. The
panels with the 1ST coupons were then
sent. to Digital Equipment of Canada
DEC i for electrical pre^creehing ancL
electrical testing.
Electrical testing was completed using
the 1ST technology. 1ST is an accelerated
stress test method used for evaluating the
failure modes of PWB interconnect
which uses DC current to create the
required temperatures within the inter-
connect. The three principal types of
information generated from the 1ST are
initial resistance variability, cycles to fail-
ure (barrel integrity, and post separa-
tion/degradation (post interconnect).
Electrical pre-screening, used to deter-
mine initial resistance variability, was
completed on the two circuits in the 1ST
coupons. The resistance value for the
first internal circuit" (PTH circuit) for
'each coupon was determined; it gives an
indication of the resistance variability
(plating thickness) between coupons
and between panels. ,
The cycles to failure indicate how
much stress the individual coupons can
withstand before failing to function
(measuring barrel integrity). 1ST
coupons contain a second internal cir-
cuit (post circuit), used to monitor the
resistance degradation of the post inter-
connect The level of electrical degrada-
tion in conjunction with the number of
cycles completed is used to determine
the presence and level of post separa-
tion. The relative performance of the
internal circuits indicates which of the
two internal circuits, the Post circuit or
the PTH circuit, has the dominant fail-
ure mechanism.
. For mechanical testing, the coupons
were sent to Robisan Laboratory Inc.
Mechanical testing'evaluated metallurgi-
cal microsections of plated through
holes in the as-produced condition and
after thermal stress. One test coupon of
each hole size, from each panel, was sec-
tioned. The coupons were microsec-
tipned to produce the most inner-layer
circuitry connections in, the microsec-
tions. .
Microsections were stressed per IFC-
TM-650, method 2.6.8. The plated
through holes were evaluated for com-
pliance to the requirements found in
IPC-RB-276. Microsections were exam-
ined after final polish, prior to metallur-
gical microetch and after microetch.
Coupon selection was four .013" 1ST
coupons from each of the 3 test panels
from each test site. Test sites 3 and 4 had
six coupons selected from the two avail-
able panels. Three coupons from within
six inches of die 1ST coupons selected
were microsectioned from the same
panels. In some cases, the desired micro-
section coupons exhibited misregistra-
-------
- : ir-r ,.t>( att'ci
i- '.He .panel as
Testing Results .
This performance demonstration was
designed as a snapshot. Because the .test .
sites were not chosen randomly; the sam-
ples inav ncit be representative of all PWB
manufacturing facilities . in the U.S.
(although there is no specific reason to
believe that they are not representative).
Also, the'number of-test sites for. each
metallization- type ranged from ten to
one. Due to the smaller number of test
sites for some metallization, results for
these metallizations could more easily be
due to chance than the results from met-
allizations with more test sites.- Statistical
relevance cannot be determined.
Bectrtoal prerscreening
- " Sevens-four, of seventy-five test panels
from twenty- five test, facilities were
returned. One of them had missing
inner-layers.
From the 73 remaining panels, .1971
coupons received, two resistance irieaf
surements using a 4rwire resistance
meter. The total number of holes tested.
was 1.4 million. One percent (19) of the
coupons were found to be defective
because of opens and shorts and unac-
ceptable for 1ST testing.
, At inspection of die defective coupons,
the opens were found to be caused by
voiding, usually within a single via, Shorfe
were caused by misregistration.. The type
of metallization did not contribute tonhe
. shorts.- ' . .UT .":.;....
To permit resistance: distributipiKinleSr
surements, the ' layers/panefe .iveice
processed in the same orientation
throughout manufacturing. The jesis-.
tance distribution for the post cireuifcwas
determined and proved very- consistent.
This result confirms that inner-layer
. printing and,etching did not contribute
to overall resistance variability.
TaWe 1. Defective Coupons Found at Prescrsemng.
Site Metallization Opens Shorts
1 Electroiess 4
3" Hectroiess^ V" J ^2
11~ _ Graphite ~~2
1T "Graphite" 5
14 Palladium 1
i- __Palladium 2
20"" Paiiadium" "" 2 "."
Percentage of Panels
Exhibiting Defect
Percentage of Panels MMing Drfec*
"Tier Technology (avg. of all test sites)
Drill Smr. ResRec. Post Sep. Drill Smr. ResRec. Post Sep.
The resistance-! distribution far, -the
PTH circuit was aBo- dejtennineS and
showed die electFohtfis'jcbppfer plating:
- increased m thickness fromi^ to both .
torn of each panel. Copper thickness
variability-W3S-.calculated to be .Q0&3^
thicker at the? bottom Compared? toVishe;
top of eachj>anet Resistance \ariaBiEfe'
was also found from right to left ore the
panels indicating inconsistent drill regis-
tration or outer-layer etching to be the
most probable cause.
Copper plating distribution at each
site was good. Plating cells and
rack/panel locations did not create large
variability that could affect the results of
each test site. Because resistance (plating
diickness) distribution was also consis-
tent between test sites (Std, Dev. for
mean resistance was 10.6 for all test
sites), relative comparisons between met-
allization sites can be made.
It was determined during correlation
that the variations in hole wall plating
thickness indicated at electrical pre-
screening was due to the flash plate pro-
vided by each test site and not the elec-
trolytic plating.
The only defects reported from the.
micrbsection evaluation werp voids in
hole watt copper, "drill smear, resin reces-
sion and inner-layer separation. Average
hole wall diickness was also reported for
each panel. Defects present, but not
'included as part of this report because
they were not believed to be a function of
the metallization technology, are regis-
. tration, inner-layer foil cracks and cracks
in flash plating at the knees of the holes.
The inner-layer foil cracks appear to be
the result of the drilling operation and
not a result of z-axis expansion or defec-
live'foil. None of the cracks in the flash
plating extended into the electrolytic
plate in the "as received" coupons or
after thermal stress. Therefore, the
integrity of die hole wall was not affected
by diese small cracks.
There were no plating voids noted on
any of the coupons evaluated. ,The elec-
trolytic copper plating was continuous
and'very even, with no indication of any
voids.
Nailheading was present on all panels,
so it was determined that comparisons
were possible since all test sites had
158' CIHCUITREE JAHiMY 1117
-------
', , » .,,' ti'.ii sr.tMi uiujlu
':r r\, t sN.'.i-. . > :,li,t:i'«l 'u rai'h site's
':vii::ui" pio
-------
Neither :i»[i-ii»un,i'i,; :io.8^
Je& Micro-
Site# sectioH
Paote
1ST affected
ElSctoolessCSr
Note: Y or N (yes or no) denote
whether post separation was detected on
any coupon or panel from each test site.
The "panels, affected" column refers to
how many of the panels within each test
site exhibited post separation.
Test site 17 was the only site for which
post separation was found in the micro-
section, but not on 1ST.
This study was. a snapshot based on
products built with B and C stage FR4
materials and a specific board construc-
tion. The data can not necessarily be
extrapolated to other board materials or
constructions. However, the results indi-
cate that each MHC technology has the
capability to- achieve comparable (or
superior) levels of performance to elec-.
troless copper.
Product performance for this study
was divided into two functions: PTH
cycles to failure and the integrity of the
bond between the internal lands (post),
and the PTH. The PTH cycles to failure
observed in this study is a function of
both electrolytic plating and MHC
process. -
162- CllCHUm - J1IIJIT 1117 .
-------
[pcomirig Seminars
i;lis;ill. . ,....
,mmsu% It 'A.I- .ipp..u-nt ilut.ill MHC
tftinitil'lint's. :iu liuiui',' rlec troll's-; n>p-
|HM'. art- Mis.opubk- to thi-. 'MH' "' luil"
tiro,
The results of this stucK further sug-
gest tlut posi separation ma\ occur 111
ditterem decrees. The level of post sep-
aration nun plav a role in determining-
product performance; however, the
determination of levels of post separa-
tion remain to be discussed and con-
firmed by the PWB industry.
Fine line evaluations in microsec-
tions have always been a point of con-
tention within the industry. Current
microsection specifications state that
anv indication of separation between
the hole wall plating and the inner-
layer is sufficient grounds to reject the
product. An indication of post separa-
tion would be a black line on the
microsection thicker than normally
appears with electroless. copper tech-
nology (normally 0.02-0.04 mils).
Separation may also be determined
bv a variation in the thickness of the
line across the inner-layer connection,
especialK on electroless deposits that
are very thin. The rationale for these
rejection criteria is that product with
post separation degrades with time and
temperature cvcling.
With traditional electroless copper
products where post separation is pre-
sent, it can usually be determined-
where the separation occurs: between
the electroless and foil, within the elec-
troless or between the electroless and
the electrolytic plating. This detenni-
' nation Often helps in troubleshooting
the plating process., . .
PWB Project participants, includ-
ing IPC, industry, EPA, Microelec-
tronics and Computer Technology
Corporation, University of
Tennessee, and Silicon Valley
Toxics Coalition representatives, will
to present the results of the MHC
CTSA. Information on the health
and environmental risk, perfor-
mance and cost of the technologies
evaluated in the CTSA will be pre-
-.ented, along with guidance for
installing and implementing the
alternative technologies. The pre-
liminary schedule for .these one-day
seminars is as follows: ^January,
Orlando, FL, AESF/EPA Pollution
Prevention Conference; 10 March,
San Jose, CA, IPC Expo; 5 May,
Chicago, IL. Chicagoland Circuit
Board Association: <)-12June.
Boston. MA, Nepcon East: 15 May,
Midwest Circuits Association; (5
-.October, Washington, J)C. IPC . . .
TechWorks: and 13 November.
Phoenix, A/,, Ari/oiia Circuits
Association meeting.
In this study, some of the direct met-
allizations resulted in no line at all after
microetching the microsections. This
posed a problem in interpreung the
results. If traditional criteria were used,-
accurate evaluations of "no lines" prod-
uct would not be possible. So, for. this
studv the criteria we used on "no line
products was: if tke sections exhibited
anv line of "-demarcation after
microetch, the product was considered
to have inner-layer separation.
This issue is significant to die PWB
industry because there remains a ques-
tion about the relationship between
the appearance of a line on the micro-
section to the performance of a board.
Traditionally (with electroless copper
products), the appearance of a line
thicker than normal electroless line is
considered to be post separation, and
For IVLore Infbrxriatiori
,= _,rrh.rornDl«e ' lei: 202-260-1023; fax: 202-2604)178:
To obtain a copy of the complete
technical paper for this study, includ-
ing appendices, please contact Star
Siimmerfield at IPC (Phone: 817-509-
0700, ext, 347). For more information
about the DfE Program or the DfE
PWB Projecj^to obtain copies of docu-
ments produced by the Project, or to
he added to the project mailing list,
contact EPA's Pollution Prevention
Clearinghouse (PPIC), U.S. Environ-
mental Protec-tion Agency, 401 "M" S
e-mail: PPIC@epamail.epa.gov. You
may also obtain additional project
information by visiting thefPC
Websiie(hup:/7wwv.pic.oig/htlelistypes.
htmftlesign).
For additional background informa-
tion on the DIE performjfe demon-
stration, please refer to tff|Sept.ember
1995'issue of CircuiTm. tfore back-
ground on the DfE Program and the
DfE PWB Project can be found in the
HieilUil Iiuitu-uuii iig^"^-;, - -- -- ; - ._
S.W., (3404), Washington, DC, 20460; i February 1995 tssue of CmmTm
the board is scrapped. However, there-
are no criteria for how to evaluate no
line" products. In addition, there are
. no official means of determining when
"a little separation" is significant- to the
performance of the board.
1ST is not a subjective test and is not
dependent upon the presence or
.absence of a line in a microsection
after microetch. The test provides a rel-
ative number of 1ST cycles necessary to
cause a significant rise in resistance in
the post interconnect. This number of
cycles may be'used to predict intercon-
nect performance. Tests such as this,
when correlated with microsecuons,
can be useful' in determining how «>
interpret "no line" product characteris-
tics. In addition, 1ST may be able to
determine levels of post separauon.
Future industry studies must deter-
mine the relevance of these curves to
performance, based on number of
cycles needed to raise the resistance as
well as the amount of change in resis-
tance. Definitions for -marginal" and
"gross" separations mav be tied to life-
cycle testing and subsequently related
to class of boards produced. C
Bill Birch is President of PWB
Interconnect Solutions Inc. (Nepean,
Ontario) Deborah Boger was an
Environmental Protection Specialist with the
EPA She currently attends Harvard
University, Kennedy School of Government
Kathy Hart is an Environmental
Protection Specialist tilth EPA's Design for
the Environment Staff (Washington. DC).
Susan Mansilla is Technical Director of
Robisan Laboratory Inc. {Indianapolis,
IN).
' 164- CIRCIIUEE JAKIUT 1917
------- |