..........
Lead (Pb)-Free Packaging Strategy
2000−
−−
−2003
JUNE 2000 (DECEMBER 2001 Updat e) JAMES H AYWARD
NOTE ON USAGE
It is a peculiarity of the English language that the words “lead” (as in package lead) and “lead” (as in
the metallic element) are homographic; that is, they are spelled the same, although they have
different pronunciations. In this document the chemical symbol “Pb” will be used whenever there is
a reference to the metallic element. The word “lead” will be restricted to its meaning as in package
lead unless a contrary usage is obvious.
- 1 -
Lead (Pb)-Free Packaging S trategy 2000−2003
Executive Summary
The us e of lead (Pb) in elec tronic prod ucts is a c oncern to s ome envir onmental groups. Im positio n of a ban on this
use has been threatened in the past, but such a ban has never been implemented due to serious technical
concerns. W hile it is not clear when, or even if, outright bans on the use of Pb may come into effect, a number of
large electronic manuf acturers believe that a de facto ban may occur in some markets in the near future. Many of
these manufacturers are our customers, so we must understand the issues and problems of eliminating Pb from
our products, and we must develop plans to address potential customer requirement that we do so.
The use of Pb has been banned from plumbing, paint, and other comm on applications for many years in the USA
and Europe. The most significant new regulatory activity is proposed directives on Waste in Electrical and
Electronic Equipment (WEEE) and Reduction of Hazardous Substrates (ROHS) now before the Parliament of the
European Union (EU). The draft ROHS directive states that: “Member States shall ensure that the use of Pb,
mercury, cadmium, hexavalent chromium, PBBs and PBDEs is substituted by 1 January 2006.” The proposed
directive has elicited strong opposition from industry organizations. A few other countries have enacted, or are
proposing to enact, legislation regarding disposal of electronic equipment, but none of these proposals explicitly
bans the use of any materials. There is at present no legislation anywhere that explicitly bans the use of Pb.
Many electronic manufacturers view the elimination of Pb from their products as an advantage to product
marketing. However, under a legislative regulation “Pb-free” would be explicitly defined, for market-driven action
“Pb-free” is a much more elastic term. A number of “Pb-free” products have been introduced by, for example,
Panasonic, Toshiba, and Nortel. Other companies, including Hitachi, NEC, Matsushita, and Sony, had announced
plans f or c onvers io n to Pb- f ree as s embly during 200 0−2001 . Ma n y of our cu stomers in the t el ecomm unication s and
automotive sectors have programs for development of Pb-free processes and are querying their suppliers regarding
availability of Pb-free compatible components.
The issues for components suppliers such as AMD can be summarized as follows:
• There is no exact replacement alloy for eutectic tin/lead (Sn/Pb) solder.
• The m os t likely rep lacement alloys ha ve melting or liq uidus temperatures th at ar e 30−40° C h ig her t han e utec t ic
Sn/Pb (183°C).
• There is no single best choice of replacement alloy. Industry is slowly converging on a tin/silver/copper
(Sn/Ag/Cu) alloy. Usage of this material may be limited by patent restrictions.
• Board assembly reflow temperature profiles will increase to ≅
≅≅
≅ 260°C to accommodate the replacement alloys.
• A change in package contact metallurgy (e. g. BGA solder balls) will not necessarily be compatible with
conventi ona l Sn/ Pb ref lo w pr oc ess es .
• Existing moisture-sensitivity data is invalid with respect to the new reflow profiles.
• Existing board-level reliability data is invalid with respect to new solders alloys and reflow profiles.
• Compatibility of existing package construction and materials to higher temperature reflow is not well
understood.
• The effect on device structures of internal package stresses caused by a higher temperature reflow is not well
understood.
• Changes in package materials to accommodate a higher temperature reflow will require new package
qualifications.
A three-par t s tr ategy to addr es s the a bove iss ues is pr oposed . During P art 1 we wil l e va luate the l imits with r es pect
to higher reflow temperatures of existing package materials and construction; we will also evaluate available
options f or lead f inishes and BGA s older bal ls. Duri ng Part 2 we will c om plete dev elopm ent of new m aterials where
necessary and the processes for new lead finishes and solder ball metallurgy, and demonstrate a capability for
qualif ication of Pb-f ree pac k aged devic es. Par t 3 will c om pr ise produc tion im plem entatio n. Com pletion of Par t 3 will
be in conjunction forecasts of the actual need for shippable product.
- 2 -
Lead (Pb)-Free Packaging Strategy 2000−2003
Introduction
The electronics industry uses a wide range of materials in manufacturing processes that have been dubbed
hazardous. At various times the use of these materials has been the focus of regulatory activity, and the industry
has responde d by dev elop ing saf er m eans of us ing these mater ials or b y reducin g or elim inating th eir use. T he use
of Pb in electronic products has been of concern to some environmental groups. The threat of a Pb ban for elec-
tronics has arisen before (in the US, most recently in 1991), but the threat dissipated under pressure from the
industr y. Action curre ntly proposed by regulator y agencies is dr iven by consum er perception th at Pb in an y form is
a dangerous material.
While it is not clear when, or even if, outright bans on the use of Pb will come into effect, many electronic
manufacturers believe that a de facto ban will happen in some form within the foreseeable future. Since some of
these manufacturers are significant customers for AMD products, we must understand the rationale for, and
problems of, eliminating Pb from our products and develop plans to address any possible customer requirement
that we do so.
Legislative Incentives for Eliminating Pb in Electronics
The ha zards of Pb in the e nvironm ent and its ef fec t on hum ans are gener al ly acc epted. The us e of Pb in pl um bing,
paint, and other common applications has been banned in developed countries for many years. In the US, such
bans were established in 1991 via Congressional legislation, although the Pb accumulated in the environment by
that time continues to cause health problems. Virtually all such legislation to date in the US and elsewhere has
exempted electronic products from regulation either explicitly by name or implicitly by usage levels. There is, at
present, no le gislati on an ywhere in the world that dire ctly regu lates the use of Pb in electronic prod ucts. The lack of
regulation on Pb c on tent can pr o bab ly be attrib uted to t he lo w us ag e le ve l in el ec tronic s (< 1 per c ent). Ho wev e r , the
regulatory environment has been changing in recent years.
European Regulation
The spotlight of international environmental, health, and safety initiatives has focused sharply on electronics since
1998. Of major concern for AMD and our customers are initiatives incorporating chemical bans or restrictions and
product “take-back” requirements. The most significant such international action is the directives on Waste in
Electr ical a nd Electr onic Eq uipm ent (W EEE) and Redu ction of Hazar dous Subst anc es (RO HS) propos ed within t he
European Union (EU). In troduced in May 1998 b y Direc torate General X I, the Environm ental Director ate of the EU
Comm ission, the fourth draft of the W EEE proposal was distributed in May 2000. At the tim e of introduction to the
European P arl iament in O ctober 200 0, the or i gin al dr a f t doc ument was s plit int o t he c ur ren t WEEE an d RO H S dr aft
directives. (W EEE includes the recycling and take-back provisions and ROHS includes the m aterial bans.) Among
its other prov isio ns, the RO HS draf t propos al st ates t hat: “M em ber States shal l e nsure th at t he use of Pb, m ercur y,
cadmium, hexava lent chrom ium, PBBs a nd PBDEs is substituted by 1 Januar y 2006.” Product categories such as
household appliances, networking and communications equipment, control instruments, electronic tools and toys,
and medical equipment are covered b y the proposal; integrated circuits and other components are included in the
definition of electronic e quipment. Autom otive electronics are exem pted from the directive implicitly by omission of
this categ ory from the lis tin g of t ypes of equipment cov ered by the regulat io ns . Nu merous other s pec if ic exemptio ns
are included in one of the annexes to the document.
The WEEE and ROH S pro pos als f oll o w f rom propos als or igin ati ng in s e ver a l of th e EU member c ountries . Pr inc ip al
among these was a sweeping proposal introduced in Denmark. The Danish proposal was severely criticized on
technical grounds by the Scientific Committee on Toxicity, Ecotoxicity and the Environment (CSTEE) in a report
dated 5 May 2000. Many of these technical criticisms probably apply to the WEEE and ROHS proposal as well.
The draf t W EEE and ROH S directives h ave also e licited strong oppositio n from indus try organi zations such as the
American Electronics Association (AEA) and Electronics Industry Alliance (EIA) in the US, and the Printed Circuit
Interconnect Foundation (PCIF) and ORGALIME in Europe. Publication of a final regulation in the year 2002 is
likely although the final form is difficult to determine at this time. The draft proposals were presented to the
European Parliament in October 2000. The first reading and vote on the proposals occurred in May 2001.
- 3 -
Parliament members added numerous amendments to the documents that are now being resolved in the
Environmental Committee. The year for enactment of the provisions (either 2006 or 2007 at this time) is a major
point of discussion. A second reading and vote will likely occur in the 2001-2002 Parliamentary session. Approval of
the directive proposals by the European Parliament will then require the member states of the EU to establish
national regulations that implement the provisions of the directive in each country.
Of the substances to be banned by the ROHS directive, Pb is b y far the most significant to AMD. Pb is present in
our products in the solder balls on Ball Grid Array (BGA) and Fine-Pitch Ball Grid Array (FBGA) packages, the
external plating on leadframe packages, and the C4 bumps used for flip-chip assembly. The ROHS draft in its
present f orm will ban s uch us es , as well as the us e of Pb-bas e d s old ers f or board - lev el ass embly of c om ponents by
our customers. The proposal includes exemptions to the ban such as:
• Pb as an alloyi ng element in steel (<0.3%), aluminum (<0.4%) or copper (<4%);
• Pb in electronic ceramic parts (e.g., covers the use of Pb oxide in passive chip components);
• Pb in the glass of cathode ray tubes, light bulbs, and fluorescent tubes;
• Pb as radiation protection.
The document allows for additional exemptions in the future if the European Commission judges the use of a
material “unavoidable”. While the draft does not now include any specification of allowable threshold concentra-
tions, it d oes require that “as nec essary” the Com mission m ay establish maxim um concentrati on levels f or specif ic
components and m aterials in the future. T he initial allowable le vels are left to t he mem ber states to deter mine an d
identify in their implementation legislation. The procedures for setting the allowable limits and for defining future
exemptions are not defined.
The W EEE Directive pr oposal also inc ludes take- back provisions that would aff ect AMD customer s. Manufacturers
and im porters m ust set up s ystem s to collect and trea t waste electr onic equipm ent from holders ot her than pr ivate
households, and absorb the costs of collection, treatment, and disposal for private households. Distributors
supplying new products must offer to take back similar equipment. Manufacturers may require that such costs as
are incurred from these take-back provisions be reflected in the prices that they are willing to pay for components.
The WEEE and ROHS Directives are significant because they would affect a market that consists now of 15
member states, 4 associated states, and an unknown number of future EU members. The provisions of the
directi ve will gov ern n ot on ly pr oducts pr oduce d in the EU, but als o th ose so ld in t he EU re gard less of ori gin. Since
the Direc tives are no t yet approv ed and in force, a m ajor questi on for AM D is wh ether the pr oposed ban o n Pb will
remain as written in the final legislation. However, some EU legislation will eventually incorporate Pb bans or
restr ictions, or requir e an end-of -life solu tion (i.e., r em oval of Pb at pro duct end o f lif e). T he res ult of th e leg isla tion,
in whatever form it finally appears, will be increased pressure to remove Pb entirely from production processes.
Additional bans have been proposed at various national levels. In 1998 Denmark proposed a ban on the import,
sale, and production of most products containing greater than 50 ppm Pb; Pb used in electronic equipment was
exempted until “further notice.” In 1998 and 1999 Sweden and Norway also identified Pb as a material they will
target for further restr iction s in th e futur e. The WEEE/RO HS direc tives would s upers ede thes e nationa l bans if they
come into effect.
Other Countries
Most European countries, Japan, and Taiwan have enacted electronic take-back regulations that require
manufacturer recycling or recovery of materials at product end of life. In 1998, the government of Japan published a
Bill on Recycling of Specific Household Appliances, covering such “major” appliances as refrigerators, washing
machines, and televisions. Extension of this regulation to include all electronic appliances has been proposed.
W hile this type of “tak e-back” regulation do es not explicitly ban the use of Pb, it does m andate the recovery of Pb
contained in the specified home electronics, and is an indirect incentive to Japanese manufacturers to avoid the
use of Pb entirely. A legislative ban on the use of Pb in Japan has been discussed, but it is not clear in what form or
when this might occ ur .
US initiat ives af fecting m anufac turers of electronics have to dat e focus ed on com munit y report ing, rec ycling, waste
restrictions, and product labeling. In August 1999, the U.S. Environmental Protection Agency (EPA), under the
Emer gency Planning an d C om munit y Right to Kn ow Act, pr op os ed st ric t er c om m unity report ing r eq uirements f o r all
industries that use Pb compounds. EPA has labeled Pb a persistent, bioaccumulative and toxic (PBT) chemical
posing a high risk of danger to human health. While present usage reporting thresholds are 25,000 and 10,000
- 4 -
Panasonic MJ-30 Mini-disk Pla
y
e
r
pounds, t he pr oposed c han ge wou ld requ ire r eport ing of r eleases and of f- site tran sf er of gr eater than 10 p ounds of
Pb compounds.
Rationale for Pb Bans
The basis for the above legislative activity is to curtail the disposal of hazardous materials in landfills. This is a
particu lar pr ob lem in Euro p e and J a pan, s i nce th e availabil ity of areas f or landf il l is extremely lim ited. T he ar g ument
in the case of Pb is that elem ental or compounded P b can be leached b y p ercolatio n into lower level gr ound water
sources, and then be cons umed by humans from drinking water or by uptak e by edible plants. Although there has
been consider able stu dy of the mechanis ms for the leaching of Pb, the science is far from def initive on the subjec t.
Leaching processes have been identified, but there is uncertainty if these processes occur naturally. Regulatory
agencies generally wish to err on the side of conservatism where even a remote possibility of hazard is involved.
Market Incentives for Eliminating Pb in Electronics
Aside f r om the pr ess ur e of leg is lat ive ac ti vity, man y elec tr onic manufac turer s are vi e wing t he elim ination of P b f r om
their products as an advantage to product marketing. With concern for environmental damage due to industrial
activit y co ntinuing to rise in the general world populat ion, labeling of products as being m ore “green” is on the rise
as well. However, under a legislative regulation “Pb-free” would be explicitly defined, whereas for market-driven
action “ Pb-f r ee” is a much more elas tic term. Althoug h a degree of mythology has ar isen ar ou nd the is s ue in r egard
to activity such as that in Japan, market forces may have more long-term impact than any legislative action.
Japan
In fact, some Japanese e lectronic com panies are now marketing Pb-free products or have announced plans to do
so. T hese pro duct a nd plan an nounc ements have g eneral ly bee n prese nted in the tr ade press in o versim plified form ,
but it is clear that these companies are at least making a commitment to the appearance of being truly Pb-free.
The most frequently cited example of an actual product is a mini-disk
player prod uced by Panas onic. Althoug h the product is not co mpletely Pb-
free, it was assembled using a Sn/Ag/Bi solder alloy, and many of the
components had Pb-free lead finishes (no BGAs were used). This product
has reportedly allowed Panasonic to gain some market share for this
product line, although hard data confirming this claim is not readily
available. Other Pb-free products reportably available in Japan include a
Panasonic television, a Toshiba laptop computer, and other consumer
products.
A number of Japanese companies, including Hitachi, NEC, Toshiba,
Matsushita, and Sony, had announced plans for conversion to Pb-free
assembly during 2000−2001. Although few of these plans have been accomplished as originally announced, the
fact of their publication indicates the intent of the issuing companies. The method announced by Sony for this
conversi on is ins tructi ve in that Son y is tak ing acc ount of the f act that it wil l be ver y diff icult to con vert to Pb-f ree a ll
at once. Son y has dev eloped their own Sn /Ag /B i/Cu /germ anium alloy that they intend to use in it ially in one m odel in
each of their sales categories. However, they will be satisfied if one component type in each of those models is
completl ey Pb-fr ee. In 2000, th ey will use t he Pb-fr ee solder f or a minim um of one com ponent t ype in all domes tic
products ( i.e., those so ld in Japan) . In 2001, the y will use the Pb-free s older for a m inimum of one compone nt type
in all products sold worldwide. In briefings given to their suppliers, Sony recognized that the major obstacle to
moving beyond these immediate goals would be the ability of components from all their suppliers to withstand the
higher processing temperatures needed for their chosen solder process.
At least one Japanese semicond uctor manuf acturer, Matsus hita Electro nics Corp, announced a program to supply
Pb-free integrated circuits (ICs). The announced program was originally restricted to leaded packages, and
palladium and Sn/Bi a lloy plat ing are spec ified as lea d finishes. Al l “new pac kages to be pr oduced f rom April 1999”
will conform to Pb-free specifications. Other products will be converted as needed with “approval,” but conversion
will be c ompleted b y th e e n d of f is cal 2 000 ( i. e., c a le nd ar 20 01) . H o we ver, r es ista nc e to h igh er r ef lo w temperatures
will be treated “ separatel y with eac h package.” T he date for c om plete conversion has since been adjust ed to fiscal
2002.
- 5 -
Nortel Meridian Phone Board
Europe
There have b een few actu a l produc t a nn ouncements outs ide of J apan.
But one such product was the Nortel Meridian desk telephone. This
phone was assembled using a Sn/Cu solder paste screened on an
organic solder pr eservat ive ( OSP)-Cu printed circuit b oard (P CB). The
lead f inish on t he c omponents was e ith er tin or pa ll ad i um. This pr oduct
was not intended to be a high-reliability telephone, but it apparently
functioned well in the home/office environment for which it was
designed.
In private communications most of AMD’s major customers in Europe
(Siemens, Nortel, Ericsson, etal.) have indicated that they have
internal programs for the development of Pb-free solder processes.
The extent and status of these programs is not completely clear, but
these com pani es are clear ly dr iven b y both the W EEE/ROHS direct ive
proposals (that they expect to gain final approval at some point) and the threat of Pb-free products im ported from
Japan. There have been few, if any, public announcements regarding plans from European companies, although
general knowledge of their intentions is widely assumed. Product introductions in 2002 are very likely.
Requirements are beginning to be defined in general component specifications. A notable development is the
acceptance by one m aj or Europe an OEM of “Pb-f ree pr ocess compatible” com pon ents that ar e not them s elves Pb-
free; that is, components that can withstand the elevated process temperatures of Pb-free assembly processes.
United States
Pb-free development in the US has been sponsored primarily by automotive electronics suppliers such as Ford
Visteon and Delphi-Delco Electronics. The interest from the automotive electronics sector, in both the US and
Europe (e.g., Siemens and Bos c h), is dr iv en to a l arge ex tent b y a d esir e t o r a ise the operating temperature o f their
products. Since a weak link in the reliability chain for higher temperature operation is the eutectic Sn/Pb solder joint,
converting to a higher temperature solder material is desirable. That such materials are basically Pb-free and,
hence, environm entally friendl y is a side benefit. But validation of reliability and transfer of designs into production
in the automotive business is a time consuming process, and it is not likely that changes will come rapidly.
However, th e automotive e lectronic s uppliers will l ikely spur dev elopment activ ities by their s uppliers in order to be
ready to make the conversion to Pb-free assembly when they need it.
Other US manufacturers, such as Motorola and Lucent, have made public expressions of support for
environm entally friendl y manufacturing. And Lucent ha s used Pb-free s older process es in a few subm odules, such
as power convertors.
Industry-Wide Activity
In addition to individual development activity, many suppliers and OEMs are working with various trade organiza-
tions in all three geographic areas (US, Europe, and Japan) on Pb-free roadmaps and processes. In the US, the
IPC and National Electronic Manufactures Initiative (NEMI) have attem pted to take a leadership role by organizing
conferences and initiating consortial development projects. NEMI is including implementation of Pb-free assembly
for US m anufac turers in its bia nnual t echno log y roadm ap. The Hig h Dens it y Packagi ng Users Group (H DPU G) has
initiated a consortial project with European and US manufacturers. NEMI and HDPUG are coordinating their
respective projects to avoid unnecessary duplication of effort. In Europe, the International Tin Research Institute
(ITRI), based in England, has set up a Pb-free soldering research consortium under the name SOLDERTEC that
has 45 member companies and wishes to be the premier such consortium in Europe. The European national
governments and the EU itself have funded other research activities. The Japan Electric Industry Development
Associati on (J EID A) and t h e J apan Institute f or Inter c onnecti ng an d Pac kaging E lec tr onic Cir c u its ( JIEP) ha v e been
developing roadmaps for Pb-free manufacturing by Japanese companies. The New Energy and Industrial
Technology Development Organization (NEDO) was established in 1998 to coordinate the activities of JEIDA, the
Japan Welding Society (JWS), and the Electronic Industries Association of Japan (EIAJ); NEDO had a budget of
¥350 million for two years of operation. In 2001 JEIDA and EIAJ merged their respective organizations into the
Japan Electric Industry Technology Association (JEITA) in order to enhance their presence as a representative of
Japanese industry.
- 6 -
Table 1: NCMS SOLDER SELECTION CRITERIA
Property Definition Limits
Liquidus
Temperature Temperature at which solder alloy is completely
molten. < 225 °C
Pasty Range Temperature difference between solidus and
liquidus temperatures. Represents the temperature
ran g e wh ere the alloy is part s o lid and pa rt liquid.
< 30 °C
Wettability A wetting balance test assesses the force resulting
when a copper wire is wetted by molten solder. A
large force in dicates a good wetting, as does a short
time to attain a wetting force of zero, and a short
time to attain the two-thirds of the maximum
wetting force.
Fmax > 30 0 µN
t0 < 0.6 s
t2/3 < 1 s
Area of Coverage Assesses the coverage of the solder on Cu after a
typical DIP test. > 85%
Drossing Assesses the amount of oxide formed in air on the
surface of molten solder after a fixed time at
solder ing temperature.
Qualitative scale
Thermomechanical
Fatigu e (T MF -1) Cycles-to-failure for a given percent failed based
on a specific solder joint/board configuration, as
compared to the eutectic Sn/Pb.
> 75%
Coeff. of Thermal
Expansion (CTE) Differences in thermal expansion behavior between
alloys might create differences in thermal stresses. < 29 pp m/°C
Creep Stress required at room temperature to cause failure
in 10,000 minutes. > 500 psi
Elongation Total p ercent elongati on of material und er unia xial
tension at room tempera ture. > 10%
Alternatives to Pb in Solders
In all the discussions about elimination of Pb in electronic products, the principal question is what will be the
replacement. The usage of Sn/Pb solders dates back some 6000 years. Sn/Pb solders have been the primary
interconnection material in the electronics industry for the last 100 years. One of the great successes of the industry
has been t hat for joi ning tec hnology the sam e materia l is used in very sim ilar ways for almost all applicatio ns. T he
large body of knowledge that has been developed regarding the properties and behavior of Sn/Pb solders would
become irrelevant if the alloy were significantly changed.
Manufacturers that have begun programs to develop Pb-free assembly processes have until recently worked
independe ntly, and as a res ult a var iety of alter native s older al loys have been deve loped. Man y of thes e allo ys ar e
the subj ects of iss ued pat ents or p endin g app licat ions. W hile sort ing ou t the c laim s of inte llectua l pr opert y is not an
insurmountable obstacle to the adoption of an alloy, it will be an additional factor in the selection process.
Pb Alternativ es
The number of elements that can be substituted for Pb in solders is limited. A number of metals with desirable
properties are either in lim ited supply (germanium, Ge) or of equal or greater toxicity than Pb itself (antimony and
cadmium). The short list of metals includes the following:
• Tin (Sn): readily available, very low toxicity, easily workable, 232°C melting point;
• Silver (Ag): limited availability, expensive, low human toxicity but potentially harmful to aquatic animals and
plants, oxide is conductive, easily workable, 962°C melting point;
• Copper (Cu): very abundan t, inexpensive, ver y low toxic it y, easil y work able, 108 4° C m elting poi nt;
• Bismuth (Bi): readily available, a byproduct of Pb smelting, low toxicity, low ductility and difficult to work,
272°C melting point;
• Zinc (Zn): readily available, inexpensive, low toxicity, easily oxidized, 420°C melting point;
• Indium (In): limited availability, low or unknown toxicity, very ductile, 157°C melting point.
It is generall y assumed that tin will form the bas is metal for any solder replacement allo y since the other possibili-
ties (bismuth and indium) are not sufficiently plentiful. Tin forms eutectic alloys with all of the above metals with
melting po ints in the range of 118°C (Sn/In) to 227°C (Sn/Cu). With additions of third, fourth, or fifth metal com po-
nents to the alloys, a very wide selection of alloys with melting or liquidus points within the 118−227°C range
(whether or not of eutectic composition) is possible.
Comparativ e Stu dies
The National Center for Manu-
facturing Sciences (NCMS) in the US
and the Brite-IDEALS project in
Europe independently undertook
extensive evaluations of solder alloys
based on the above metals, with and
without add it ion al al loying m etals . The
most ex tensive com pilat ion of data o n
solder materials is that published by
the NCMS in 1998. This consortium of
11 US manufacturers and research
organizations evaluated 79 solder
alloys based on toxicology, econom-
ics, and material properties. This
evaluation resulted in a short list of
alloys that were subsequently
evaluated for manufacturability and
reliability.
Table 1 shows the primary character-
istics that NCMS used to select
candidate alloys. The consortium was
primarily looking for alloys that
- 7 -
Selected Pb-Free Alloys
Composition Melting/Liquidus
Temperature (°C) Alloy Type Companies
Evaluating or Using
Sn-37Pb 183 Eutectic (St andard)
Sn-3.5Ag 221 Eutectic Siemens
Sn-0.7Cu 227 Eutectic Nortel
Sn-58Bi 139 Eutectic Matsushita
Sn-3Ag-0.5Cu 217 Non-eutectic Siemens,Nortel
Sn-3Ag-2Bi 220 Non-eutectic Fujitsu
Sn-2.6Ag-0.8Cu-0.5Sb 211 Non-eutectic Sony (+ Ge)
Sn-3.4Ag-4.8Bi 210 Non-eutectic M atsushita
Sn-2.8Ag-20In 187 Non-eutectic
Sn-3.5Ag-0.5Cu-1Zn 221 Non-eutectic NEC
became liquid below 225°C and have small pasty ranges. (Eutectic alloys have a distinct melting temperature at
which the alloy goes directly from the solid to the liquid state. Non-eutectic alloys have a temperature at which
some phas e begi ns to m elt and anot her at which all phas es ha ve m elted; the t wo tem per atures ar e the soli dus and
liquidus tem peratures r espectiv ely. The range bet ween th e solidus and li quidus tem perature is t he “past y” range i n
which the material is a
mix tur e of s olid and l iqu id
phases.) Since the list of
candidate materials does
include non-eutectic as
well as eutectic alloys,
the size of the pasty
range is of concern. A
smaller pasty range
allows for easier control
of the solder processing
temperature. The alloys
in the final NCMS list, as
well as Sn/37Pb, are
shown in the Table 2.
Although none of the
materials on the list can truly be said to be drop-in replacements for the eutectic Sn/Pb alloy (Sn/37Pb), the
materials on the short list do have many characteristics that may be in fact better than the standard alloy. One
particular eutectic alloy that had interested several compan ies (Sn / 0.7 Cu) is pr oba bly not practic a l d ue t o the melting
point of 227°C. The listed Sn/Ag/In alloy is also probably not a viable candidate due to the limited availability and
cost of indium. Nevertheless, the alloys on this list are generally similar to those under evaluation by various
companies.
In board-level reliability tests using surface-mount and chip components, several of the listed alloys appeared to
perform better than the control Sn/Pb eutectic due apparently to higher mechanical strength and/or ductility. The
final recom m endation of NCMS was th at thr ee of the l isted al lo ys warr anted f urther s tud y. Under the NCM S cr iteria
the Sn/Bi eutectic would be acceptable for consumer electronics, and both the Sn/Ag/Bi and Sn/Ag/Cu would be
suitable for telecom munications , automotive, and aero s pace appl ic ati ons.
Individual Studies
In addit ion t o the N C M S list, other all o ys are b ei ng i nv e st igat ed or implem ented by various c ompanies in Ja p a n and
Europe. Sho wa Denko c lai m s to have s ol ved the prob lem of using zinc-c ont ain in g alloys and is prod uc in g m ater ia ls
with eutectic (Sn/9Zn, 199°C melting point) and non-eutectic (Sn/8Zn/3Bi, 197°C liquidus) compositions; NEC is
reported to have adopted the Showa Denko process. Sony developed a Sn/Ag/Cu/Bi/Ge alloy with a 220°C
liquidus. Other Japanese companies are working on variations of Sn/Bi or Sn/Ag/Bi compositions. Nortel has
work ed primar ily with the Sn/Cu eutec tic, but is eval uating variati ons of the Sn/A g/Cu class. Addit ionally, res earch-
ers at Iowa State University developed and patented a series of alloys based on Sn/Ag/Cu, and Flip-Chip Technolo-
gies (FCT ) developed and patented a Sn/Ag/In/Cu alloy (Sn/3.1Ag/10In/1Cu, 200°C liquidus) for direct chip attac h
(DCA) applications. In Europe and the US, where the production of high-reliability equipment for
telecommunications and networking constitutes a high percentage of electronic manufacturing, there is a trend
toward the use of the Sn/A g/Cu al loy possibly with addit ion al al loying metals.
Evaluation Issues
Evaluation of many of these new solder alloys is difficult because reliable information about the mechanical
properties and phase constituents of these materials is lacking. Mechanical properties of many of these materials
remain to be determined. Construction of phase diagrams for materials with the addition of third or fourth
constituents is practically impossible and can only be done via thermodynamic modeling methods; construction of
phase diagrams for binary alloys has proven difficult enough. Sn/Bi-based alloys exhibit a specific example of the
subtleties of phase structures. When these alloys are used with components that still have Pb-bearing contacts a
Sn/Pb/Bi eutectic phase (96°C melting point) can be formed. This phase would melt at any temperature above 96°C
causing a solder joint to become pasty and the mechanical strength of such solder joints could be severely
degraded. At least one automotive manufacturer has stated that they will not use any solder containing bismuth due
to this phenomenon,
Most discus sions abo ut solder r eplacem ent have foc used on the exter nal contac ts on pack ages and th e means f or
connecting packages to boards. However, Pb-bearing solders are essential materials for flip-chip assembly
- 8 -
processes used for connecting die in packages. The well-known IBM C4 process, of course, uses not only a Pb-
based solder but also a solder that contains mostly Pb. No acceptable alternatives to the Pb/3Sn eutectic (318°C
melting point) and related alloys used in this process have been identified. A good argument has been made that
C4 contributes only a very small portion of the Pb contained in present-day electronic equipment and C4
applications have been granted exemption in the current version of ROHS. Flip-chip processes that use solders
with lower Pb content (non-C4 assembly processes) would probably not be exempted from regulation. But even if
flip-chip-in-package applications are exempted from regulation, the solder used for the flip-chip connection would
have to be c ompatible with the elevated tem peratures require d by the board-l evel assem bly proc ess. Ther e will be
a problem if the melting temperatures of the flip-chip solder, or the liquidus temperatures, are below the board
assembly temperature. For either of these cases, the flip-chip solder joint may melt or become soft during board
assembly, and the reliability of that joint may be reduced.
Packaging Issues Resulting from the Elimination of Pb in Solders
The issues for packaging that derive from a change to Pb-free solder assembly can be broken into several groups:
• Metallurgical: selection of a compatible lead finish or solder ball alloy;
• Temperature sensitivity: compatibility of existing package materials and construction with elevated board
assem bly proces s tem per atur es;
• Reliability: evaluation of the solder joint reliability of the new solder/lead finish system;
• Cost: evaluation of the cost of converting manufacturing processes;
• Schedule: managing the conversion to Pb-free processes.
The metallurgical issue of compatibility with the chosen m aterial for board assembly is both an easy and a difficult
one for the component supplier. The easy part is that to a large degree customers will dictate the choice of
materials. The difficult part is that different customers may choose, for very logical reasons in each case, different
solder systems that they wish to use in their manufacturing. Selecting an alternate to Sn/37Pb solder is an engi-
neering problem f or the OEM th at does not ha ve a un ique solutio n. The com ponent suppl ier may hav e the diff icult
task of s upporting m u ltiple l ead fin ishes or d evelo ping a pack age so lution that is compatible with a vari et y of s older
systems. The choice is driven by external events that the component supplier may have little control or influence
over. Alt hough indus try org anizations throughout t he world are at tempting to f oster agreem ent to a single or lim ited
number of replacement alloys for board assembly, this effort is not yet complete.
For leaded packages the list of possible lead finishes is short. Plated tin (electrolytic or electroless), plated silver,
nickel/palladium (several variants), and nick el/gold have been suggested. Evaluation of these and other less likely
finishes have begun , but n o clear ly s uperior finish has been d em onstr ated. All ha ve sp ecif ic dra wbac ks : nickel/g old
and nickel/palladium may be too costly; tin has the problem of whisker growth; silver has the problem of
electrom igrati on; etc. An y of thes e problem s c an be overc om e, but succ essfull y dem onstrating the acc om plishm ent
may require a substantial effort.
Lead Finish Issues
During the transition to use of non-Pb solders for board assembly, leaded packages with Sn/Pb finishes may
continue to be utilized. T he solder f inish o n the package le ad is thin, and it is only inten ded to ensure wettability of
the lead dur ing assem bly. T he majorit y of the so lder in the joint c omes from solder paste s creened on th e board or
from solder plated on the boar d. So, the integrit y of th e solder jo int is inf luenced only to a sm all extent by the lead
finish. However, with BGA-type packages the solder ball contributes most of the material in the joint. If a non-Pb
solder paste with a higher melting temperature is used on the board, then the Sn/Pb solder ball on the package
ma y melt dur ing assem bly before the s older past e melts. T he lower m elting Sn/Pb s older will a lloy with th e tin-rich
solder paste f orm ing a no n-eutec tic al lo y of ind eterm inate co m position t hat will l ikel y free ze bef ore the solder pas te
would have melted. In effect, a “c old” s o lder j oint wil l b e f ormed that m a y have lo w f atigue resistanc e. T hus , the P b-
bearing solder balls on BGA packages will have to be replaced in most Pb-free applications.
For BGA-type packages not only the solder ball material but also the solder attach pad material on the package
substrate are of interest. Raw solder balls used in BGA package assembly can be manufactured of any desired
alloy. Ho wever , if the s old e r ball materia l is c han ged, t hen th e in ter f ac ial c h arac te r ist ics at the s ubs trat e ar e l i k el y to
change as well. Without a clear understanding of the metallurgy involved, predicting the results of a different
interface condition is difficult. Research organizations must be encouragement by industrial partners to obtain the
necessary metallurgical knowledge. Otherwise, the strength of the solder ball to substrate joint can only be
evaluated on an empiric al bas is.
- 9 -
Similar considerations apply also to packages assembled with flip-chip processes. Although C4 assembly is likely
to be exempted from regulation, non-C4 processes (those using low Pb content solders) will be included in the
regulation. T he solder used for the flip- chip joint needs to be com patible wit h the ele vated proc ess tem peratures of
the new b oard assem bly sold ers. Melt ing or sign ifican t softening of the f lip-chip j oint duri ng board assem bly is ver y
undesirable and may degrade the reliability of that joint. Thus, solders with melting points or solidus temperatures
below 260°C should not be used. Packages that are attached to boards only with sockets are not be effected by
board assembly processes, but may still need to avoid the use of Pb solders to ensure that the final product falls
within the scope of regulation. Any change in the flip-chip solder alloy may require a change in the barrier layer
metallization or the solder bump structure as well. The same issues apply to passive chip components used in
packages or assemblies; contact metallization and stability after high temperature reflows have to be evaluated.
Assembly Issues
A potentially more difficult problem for the component supplier comes from an increase in the board assembly
process ing tem perature tha t com es with a change to Pb- free solders . Most r eplacem ent solder a lloys ha ve m elting
or liquidus temperatures that are higher than Sn/37Pb by as much as 20°C. There are several factors that
determine the processing temperatures for Sn/Pb assembly, and they also determine the processing temperature
for an y other solder. Cur rent boar d assem bly process tem peratures are from 40°C to 50°C abo ve the m elting point
of 183°C for the eutectic Sn/Pb. This temperature difference is necessary to ensure that all components going
through a reflow furnace reach temperatures where the solder liquefies and wets the component leads. The
process temperature must account for variations in board size in a given manufacturing line, and for varying
therm al mas ses of com ponents on the b oard, in or der to satis f y throughput req uirem ents . The proc ess tem perature
must also compensate for the fact that often the composition of material at the solder joint is not exactly at the
eutectic composition. Material contributed by the board finish, solder paste, and lead finish may all deviate by ±5
percent from the eutectic composition, and such deviation increases the melting (now liquidus) temperature. In an
actual assembly the liquidus temperature may be as much as 10°C above the stated melting point of 183°C
because of the variation (intentional or not) in solder composition. For non-Pb materials the same situation will
exist; m anufac turing variat ion will res ult in com positio ns that dev iate from the specif ied com position, and t he actua l
liquefaction temperatures will deviate as well. For all these reasons, processing temperatures up to 260°C can be
expected for alloys with melting points up to 220°C. And in order to maintain assembly throughput without adding
significan tly to t he size of ref low furnaces , heat-u p and coo l-down rat es m ay be hig her than those use d at pr esent.
It must be noted that for BGA packages, the above considerations apply to the ball attachment process as well.
Temperat ure Ef fect s
Higher assembly process temperatures affect packages in several ways. If the package contains absorbed
moisture, then the elevated temperature of the reflow process and the ramp to that temperature can cause the
moisture to desorb from the package rapidly with a high risk of crack formation or fracture. This is the classic
“popcorn” effect. Current test methods to determine the susceptibility of packages to this problem specify peak
reflow temperatures of 225°C (or 240°C for thin packages). The tests are empirically derived, and there is no
method to extrapolate the results to higher temperatures. Thus, the susceptibility of packages to moisture-driven
damage at higher processing temperatures cannot be predicted.
In addition to the moisture-driven effects, higher temperatures can also degrade the package integrity by directly
eff ecting the adhesion at interfaces , and by stressing t he structure as the t emperature g oes ever far ther above the
glass trans istion tem perature ( Tg) of the m aterials. Som e of these eff ects can be m odeled b y use of finit e elem ent
analysis (FEA), but a thorough understanding of the material properties and the interfaces is necessary both to
construct adequate models and to interpret them. In the case where package functionally survives the non-Pb
assembly process (that is, the electrical performance of the device does not degrade as a result of the assembly
process), a reduction in the useful life of the component due to incipient defects initiated by the assembly process
would remain a concern. Reduction in adhesion may result in an earlier onset of catastrophic delamination;
development of harmful intermetallics at wirebond interfaces may be initiated; and harmful residual stresses may be
induced that lead to early thermo-mechanical failure. Mold compounds, die attach adhesives, underfill adhesives,
and substrate materials may all exhibit significant temperature-driven degradation; whether such degradation
occurs is ver y diff ic ult to de monstrate ex c ept b y extensiv e test in g. Cle arly, if it is nec ess ary to develop or ado pt ne w
materials to prevent such degradation, then the tim e required to im plement a conversion to Pb-free manufacturing
ma y be dramatic ally increased.
- 10 -
Reliability Effects
Changing the material used for the solder joint also changes the thermo-mechanical behavior of the joint. The
mechanical properties of the solder alloy contribute significantly to the strength of the joint. Many of the non-Pb
materials seem to be stronger or more ductile than Sn/37Pb, and the microstructure of these materials also
appears to be immune to the grain-coarsening process that Sn/37Pb exhibits. Increased strength and ductility
should result in stronger, more fatigue-resistant solder joints, and the absence of grain coarsening should result in
solder joints with greater stability. But the increase in solder joint strength m ay also result in higher leve ls of strain
at the interfaces between the solder joint and the package lead or substrate attachment pad. Thus, the quality of
this interface becomes more critical to board-level reliability. The level of residual stress generated at the
temperature at which the solder joint solidifies also affects the reliability of the solder joint. The full range of
mechanical properties with respect to temperature for most of the alternate solders has not been well
characterized; adequately modeling the solder joints using FEA techniques is not possible without this material
characterization. Therefore, the solder joint reliability must be determined experimentally.
Implementation Costs
The c ost of conver ting t o Pb-f ree m anufac turing com es from the im plem entation of new proc esses , develop m ent of
new materials, and qualification of the resulting products. The implementation of the new processes has an
expense associated with both the new process itself (equipment, materials, facilities, etc.) and running the new
process in paralle l with old er proces ses durin g any trans ition perio d. The durat ion of the transit ion period is dif ficult
to determine until clear plans for conversion on the part of customers are defined. However, it is safe to assume
that the conversion of customers will not be well synchronized. The component supplier will have to maintain the
parallel process lines f or an indefinite period, or demonstrate that the new lea d finishes are sufficientl y com patible
with conventional Sn/37Pb solder processes. And, in the absence of explicit regulation, some OEMs may never
convert to Pb- fr ee assem bly for part or all of their pr oduc t lines. In th at case, s eparate c om ponent proces s lin es will
be permanent installations. Such parallel process lines increase the cost both of conversion and of continued
component pr od uctio n.
The cost of qualifying products is potentially much more than establishing the necessary new processes. As a
minimum, families of packages will need to be requalified for moisture sensitivity under a revised version of the
standard t est method f or precond itioning. (A pro posal f or modif ying the exist ing JEDEC pr econdit ioning stan dard is
about to be submitted to the JC-14 committee.) If new materials or assembly processes are needed to make the
packaged component more resistant to temperature, then complete product qualification will also be needed. For
component s uppliers with e xtensive catalogs , the cost and t ime required to d o such qualif ications m ay exce ed any
other cost incurred.
Implementation Plan
In responding to requirements for components compatible with Pb-free assembly, two extreme positions are
possible. The first is to do nothing, wait to see what happens, and address the problem at the time customers
demand Pb-free components. The second is to rush into a program to change lead finishes and put Pb-free
components on the market as quickly as possible. Either approach is a defensible option that comes with
associated pr oblems . The danger in the first approac h is that the com pany would not be able to r espond in tim e to
customer requirements and would potentially lose business. The danger in the second approach is that the
company would ignore the reliability and manufacturing issues outlined above, commit excessive resources to the
program, and convert our manufacturing processes ahead of the time that business would naturally support. The
newly installed manufacturing capacity may stand idle for an indefinite time.
AMD mus t f ind a r easona bl e middlegro und bet ween t h es e ex tremes that all o ws u s to f ully und er stand the probl em s
and the solutions, and to map whatever manufacturing conversion is needed against the business plan. It is
ultim ately to our adv antage that th e developm ent proces s be undertak en methodic ally and ef ficiently, and th at any
conversion process be an orderly addition to current manufacturing capabilities. Additionally, our development
and/or conversion must be coordinated with our family of subcontractors. All of this implies that we begin our
development process as early as possible.
Because of the package-related issues described earlier, AMD cannot make a commitment to produce Pb-free
components without having completed certain initial evaluation projects. The evaluation projects require
cooperation between Sunnyvale and the offshore sites in Singapore, Penang, Bangkok and Su zhou. The strategic
program should consist of three phases of which the first will be an evaluation of existing packages and potential
replacement materials.
- 11 -
1A Evaluate the sensitivity of current package materials and assembly methods to the elevated
temperatur es nec essary f or the n ew reflow profiles. T his pr ogram would begin by selecti ng a repres entat ive
group of produc tion p ack ages, inc ludin g leade d and B GA types , and subjec ting t hese pack ages to our stan dard
qualif ication tes ts with th e higher temperatur e reflow profiles used f or the pre- conditionin g steps . The pack ages
would be exam ined in detail durin g the tests in order to deter mine the onset of addit ional defec ts beyond those
expected f rom the s tandard qualif ication pr oced ures. (O ur standar d qualif icatio n procedur e woul d be the co ntrol
for this evaluation.) T he results of the evaluatio n will determ ine the extent of our jeopard y with regard t o higher
process ing tem peratures, a nd be the bas is for subsequent evaluation of ne w materials and assem bly m ethods.
This evaluati on shou ld also be the b asis fo r evaluat ion of board- level r eliab ility for new lead finish es and s older
alloys.
1B Evaluate the impact of higher package and board assembly temperatures on device performance. We
need to determine the effect of exposure to higher processing temperatures on device performance. Issues
such as Flas h mem or y data reten t ion above 250ºC an d mechanica l stability of ne w fab pr oc ess es using organic
dielectrics must be included in the evaluation.
1C Evaluate lead finish options for leaded packages. We must evaluate metallurgical, process, and cost
parameters of alternative plated finishes in order to determine the optimum replacement process.
1D Evaluate solder ball options for BGA and FBGA packages. We must evaluate metallurgical, process, and
cost parameters of alternative solder ball alloys. We will need to consider possible changes in the solder pad
finish on the BGA substrate as well. This part of the evaluation includes an investigation of solder joint reliability.
The second part of the program will extend the work of phase 1 to a development phase in two parts:
2A Pb-free process compatible packages. A selection of package types will be modified though changes in
materials or construction as recommended by the work in phase 1 in order to achieve the required moisture
sensitivity level with a 260°C preconditioning profile. The modified packages will undergo a complete
qualif ication proces s to dem onstrate achievem ent of this requ irement. No cha nge will be m ade to lead fin ish or
solder ball composition.
2B Fully Pb-free packages. A selection of package types will be modified though changes in materials or
construction including lead finish and solder ball composition as recommended by the work in phase 1. The
modified packages will undergo a complete qualification process to demonstrate the required moisture
sensitivity level with a 260°C preconditioning profile. Board-level testing will be performed to demonstrate the
reliability of the new lead finish and solder ball composition.
Phase 3 of the program will be production implementation of the solutions developed in phase 2. The timing and
target devices for this phase will depend strongly on market forecasts and business plans formulated by the effected
product lines. Consultation between manufacturing and these product lines will be required to do the capacity
planning and any necessary equi pment convers ion in the ass embly sites.
During the course of this strategic program certain additional activities will take place:
1 Establish a program manager for all Pb-free activities. Since these activities cross over boundaries with
customers and with internal organi za ti ons , the es tab lis hment of a direc ting bo d y, wheth er a spec if ic ind i vid ua l or
a group, is needed to oversee the activity within MSG and coordinate with outside groups.
2 Select a set of specific customers with whom we wish to engage in cooperative development activity.
Such development will include reliability and compatibility evaluations. Candidates such as Nortel Networks,
Delco-Delphi, Siemens, and Compaq Computer would be appropriate.
3 Select specific industry consortial organizations with which we will engage in development activity. This
action serves to lend our weight to encourage the worldwide industry to converge on a single solder replace-
ment, or at least a very limited number of such alloys. The most important candidates are the High Density
Packaging Users Group (HDPUG), the International Tin Research Institute (ITRI), and the Japan Electronic
Industry Development Association (JEIDA) or EIAJ.
4 Encourage our subcontractors to participate in the above activities or incorporate them as partners
explicitly. We cannot depend on our subcontractors to develop conversion plans consonant with our needs
independently. We must help them choose materials and processes, and validate those choices.
5 Work with AMD’s product groups to perform an official survey of customers to determine both the need
and the requirements for specific AMD products in Pb-free form. The product groups will need to develop
business plans based on this information as justification for MSG to formulate subsequent conversion plans.
- 12 -
6 Dev elop an in -h ou se exp ert ise in t he metallurg y o f so lder s. U nder s ta ndi ng th e ap pl ic ab il it y of any partic ular
lead f inish or s old er bal l att ac h pr oces s r equires an u n der st and in g of the metallurg y of interfaces at the p ac k age
(for BGAs/FBGAs) and at the board.
7 Establish a website for information on our Pb-free program.
Completion of phases 1 and 2 outlined above requires in itself a significan t commitment of resources by MSG and
other divisions of the company. The goal for the first phase evaluation should be to define by Q4, 2001, the extent of
any problem s with existing pack age m aterials an d con struction an d prop ose solut ions to th ese prob lem s that will be
the basis for the phase 2 projects. This goal should be accomplished by Q4, 2001.
The phase 2 projects, as well as implementation as manufacturing processes in phase 3, will be a longer-term
program extending into 2002.
Acknowledgments
Thank s are hereby gi ven to Donna Sadowy and F umio Namba of AMD, and to B rian Swigge tt and Neil Mos kowitz
of Prismark, for their assistance in gathering the information necessary for this document. Thanks are also due to
man y of m y AMD colleagu es for as sistance duri ng vis its to custom ers througho ut the wor ld. In add ition, thank s are
due to John Hunter who encouraged this study.
