Docstoc

LEAD FREE SOLDERS IN ELECTRONICS

Document Sample
LEAD FREE SOLDERS IN ELECTRONICS Powered By Docstoc
					                             LEAD FREE SOLDERS IN ELECTRONICS

                                                     Angela Grusd
                                                     Heraeus Inc.
                                                West Conshohocken, PA


ABSTRACT                                                      It is widely known that lead is related to certain health risks.
Lead-bearing solders are used extensively in the              If lead particles are inhaled or ingested, they accumulate in
electronics industry. In recent years, efforts to develop     the human body causing damage to the blood and central
alternatives to lead-based solders have increased             nervous systems. Lead poisoning can be detected by blood
significantly. As researchers began to focus on Pb-free       analysis and the limits are defined by national governments.
solders they recognised their potential in high               The standard upper value of untainted human beings should
temperature applications such as automotive where Sn/Pb       not exceed 130 µg/l. The upper blood lead limit of people
solders do not meet the demanding requirements. In            who are exposed to lead at work is, according to the German
particular, the Sn-Ag-X lead free solders offer superior      Biological Place of Work Tolerance values (TRGS 903, June
creep resistance at room temperature and 100°C as             1994), 300 µg/l blood for women and 700 µg/l for men. In the
compared to Sn-40Pb. Results of this work will be             USA, these critical values have been lowered from 250 to 100
presented as well as factors to consider when developing      µg/l.
and implementing lead free alloys such as
manufacturability, availability, and cost. One of the most    As a result, some industries have already eliminated lead and
promising replacement alloys is Sn/4Ag/0.5Cu. This            have found suitable alternatives, for example plumbing
alloy will be discussed in detail.                            solders, tinned cans, lead-free gasoline for vehicles, and lead-
                                                              free cut crystal glass. The majority of lead consumption is
INTRODUCTION                                                  automobile batteries and ammunition. The lead consumption
Lead free solders are currently in production in some         of the electronics industry is relatively small and, according
facilities and some of the “green” companies have             to different sources, lies between 0.5 - 7%.
proposed timelines for their implementation in the next
year. The development of lead free alternatives was           When choosing alternative metals, consideration must also be
initially driven by impending U.S. legislation and            given to their health risks as well. Recent studies in the USA
Environmental Protection Agency regulations restricting       and in Europe came to the following conclusions concerning
lead usage in the electronics industry due to the toxicity    the toxicology of lead and some alternative metals:[1]
of lead. More recently, the European Commission has           • Cd is extremely toxic and should not be used (high risk).
proposed legislation aimed at abolishing lead in End of           Many companies such as Ford Motor Company have
Life Vehicles by 2002. The European Commission has                banned Cd-containing materials.
made a second proposal against lead in End of Life            • Pb was also identified as highly toxic (high risk - it is
Electrical and Electronic Waste, however no firm date             considered harmful to the reproductive system).
has been set.                                                 • Sb is very toxic and should not be considered a major
                                                                  alloying element (medium risk - in Europe this material is
Japan has similar proposals surrounding lead usage. In            considered potentially carcinogenic).
December 1997, the Japan Environmental Agency                 • Ag and Cu are used in lead-free alloys in small
proposed legislation on the disposal of lead scrap. Lead-         concentrations - in Europe these materials are seen as low
containing scrap, such as electronics, must be disposed of        risk.
in sealed landfill sites to prevent the leaching of lead. A   • Sn and Zn are essential elements in the human diet, yet
second Japanese proposal was initiated by the Ministry of         may be toxic if exposures are sufficiently high (low risk).
International Trade and Industry (MITI) Japan                 • Bi is a relatively benign metal with a history of medicinal
Automobile Industries Association. This called for a              uses (low risk).
50% voluntary reduction of lead in vehicles (excluding
batteries) by 2001 and to one-third by 2003. Several          • Greater Thermal Stress of Components
major Japanese Consumer Electronic Manufacturers have         In the automotive industry, more and more circuits are being
publicly announced accelerated plans to eliminate lead        placed in the engine compartment in order to reduce the
solder completely by 2001. Lead free alternatives are         quantity of cables and, therefore, reduce cost. These under
being considered for several reasons:                         the hood conditions often see temperatures in excess of
                                                              150°C.     A leading automobile manufacturer has even
•   Health Concerns                                           measured temperatures as high as 170°C on a hybrid circuit.
The high thermal stresses imposed on the solder joints at      compared to 63Sn) and higher reflow temperature
these temperatures has led automotive manufacturers to         environments accelerate the diffusion rates for copper base
research Pb-free alternatives with high thermal fatigue        metal in Sn. As its corresponding composition is reached, the
resistance, because they observed that Sn-37Pb has poor        brittle Cu6Sn5 intermetallic compound is nucleated and begins
thermal fatigue properties even at 125°C. Higher               to grow. To slow the diffusion rate and thereby decrease the
temperatures dramatically reduce the strength of the           growth kinetics, alternative surface finishes such as
solder joint during thermal cycling, due to greater plastic    immersion gold (Au over Ni over Cu) may be used. The 2 µm
deformation of the solder as well as diffusion,                Ni in the immersion gold coating serves as an effective
recrystallization and grain growth inside the solder.          diffusion barrier, limiting the Cu from diffusing into the
                                                               solder and forming the brittle Cu6Sn5 intermetallic compound.
For conventional alloys such as Sn62/Pb36/Ag2 (melting         Other surface finishes such as immersion silver (Ag over Cu)
point 179°C) and Sn63/Pb37 (melting point 183°C),              and immersion palladium (Pd over Cu) do not contain a Ni
there is a major concern for the mechanical and                barrier layer. Their effect on the growth kinetics of the
microstructural stability and, therefore, the reliability of   intermetallic compound layers is under investigation.
the solder joint at an operating temperature of 150°C,
because it is approaching the melting point of the alloy.      2.       Sn/Ag/Cu
                                                                   1) 95.5Sn/4.0Ag/0.5Cu                    217-219°C
• Use of Temperature Sensitive Components and                      2) 95.5Sn/3.8Ag/0.7Cu                    217-219°C
    Substrates                                                     3) 95.0Sn/4.0Ag/1.0Cu                    217-219°C
Some industries have driven down costs by replacing                4) 93.6Sn/4.7Ag/1.7Cu                    216-218°C
higher cost plastic components with less expensive             Because the mechanical stability of the joint is degraded when
plastic ones.    These components, however, cannot             the melting point is approached, elevated temperature cycling
withstand the standard reflow temperatures of 210-             produces more damage for Sn/Pb solder (m.p. 183°C) as
230°C. Therefore low melting temperature solder alloys         compared to higher melting point solders. The melting
are used in this case. This is especially apparent with        temperatures of Sn/Ag/Cu solders make them ideal in high
consumer electronics, as the operation temperatures are        operating temperatures up to 175°C. As for wetting,
from 0°C to +60°C. This lower temperature range                Sn/Ag/Cu solders do not wet Cu as well as Sn/Pb using
corresponds to much less thermal stress on the solder          commercial fluxes. However, good fillet formation can be
joints as compared to those temperatures found in under-       easily achieved provided the fluxes are suitable for higher
the-hood applications which typically reach 150°C or           temperature use. Soldering in nitrogen atmosphere also
greater. An example of a solder for these lower                improves wettability using no-clean fluxes. The copper
temperature applications is Sn/Bi eutectic.                    dissolution test provides a relative measurement of the
                                                               solder’s tendency to dissolve Cu from the base metal and
LEAD FREE ALTERNATIVE SOLDER ALLOYS                            form the Cu6Sn5 intermetallic compound. For alloys 1-3, the
1.       Sn/Ag (96.5Sn/3.5Ag:221°C)                            rate of copper dissolution is slower than the Sn/Ag alloy yet
This alloy exhibits adequate wetting behavior and              faster than the Sn/Pb eutectic. For alloy 4, the high level of
strength and is used in electronics as well as plumbing.       Cu in the alloy prevented the dissolution of the copper wire
Several sources have also reported good thermal fatigue        (See Dissolution section).
properties as compared to Sn/Pb. Thermal fatigue
damage in solders is accelerated at elevated temperatures.     3.       Sn/Cu (99.3Sn/0.7Cu:227 °C)
In the Sn/Pb system, the relatively high solid solubilities    This alloy might be also suitable for high temperature
of Pb in Sn and vice versa, especially at elevated             applications required by the automotive industry. It is a
temperatures, lead to microstructural instability due to       candidate especially for companies looking for lead and silver
coarsening       mechanisms.        These    regions     of    free alloys. Preliminary testing conducted on this alloy has
inhomogeneous microstructural coarsening are known to          shown a significant improvement in creep/fatigue data over
be crack initiation sites. It is well-documented that these    standard Sn-Pb alloys.
types of microstructures in Sn/Pb alloys fail by the
formation of a coarsened band in which a fatigue crack         4.       Sn/Ag/Cu/Sb      (96.2Sn/2.5Ag/0.8Cu/0.5Sb
grows. By comparison, the Sn/Ag system, has limited                     (known as Castin):217-220 °C)
solid solubility of Ag in Sn, making it more resistant to      This alloy has similar mechanical properties to the Sn/Ag/Cu
coarsening. As a result, Sn/3.5Ag eutectic forms a more        alloy.
stable, uniform microstructure that is more reliable.
                                                               5.       Sn/Ag/Bi            (91.8Sn/3.4Ag/4.8Bi:200-216°C)
Although the Sn/3.5Ag alloy itself exhibits good               In general, bismuth is added to Sn/Ag/X solder alloys in
microstructural stability, when soldered to copper base        order to depress the melting point. Another benefit of Bi
metal, the combination of a higher Sn content (96.5Sn          addition is greater joint strength as indicated by ring and plug
testing. This particular alloy was developed by Sandia
National Labs. Sandia’s internal studies have found no        8.        Sn/Sb (95Sn/5Sb:232-240°C)
electrical failures on surface mount devices following        The 95Sn-5Sb solder is a solid solution of antimony in a tin
10,000 thermal cycles using 68 I/O PLCCs, 24 I/O              matrix. The relatively high melting point of this alloy makes
SOICs, and 1206 chip capacitors on standard FR-4              it suitable for high temperature applications. The antimony
PCBs. The boards were cycled 0 to 100°C at a ramp rate        imparts strength and hardness. In comparing the yield
of 10°C/ minute. No cracks or deformation were observed       strengths of several solder alloys, the strength of 95Sn/5Sb
on boards cross-sectioned after 5000 thermal cycles.          was 37.2 N/mm2 and was nearly the same as 96.5Sn/3.5Ag
Cross-sectional data on 10,000 cycles is being collected.     (37.7 N/mm2 ).[2] The high strength of this alloy causes the
These results are in good agreement with data collected       lowest energy crack path to be at the solder/intermetallic
by the National Center for Manufacturing Sciences             interface in the case of thinner intermetallics. As the
(NCMS) Lead Free Solder Project, which reported very          intermetallic thickens, the crack path is through the
good thermal fatigue resistance on OSP printed circuit        intermetallic layer. Formation of the intermetallic compound
boards (Organic Solderability Preservative that protects      SbSn is possible at these levels of Sb. This phase has a cubic
copper pads and through-holes). The NCMS High                 structure with a high hardness. The wetting behavior was
Temperature Fatigue Resistance Project is currently           measured on a wetting balance in air using a standard RMA
evaluating this solder at temperatures up to 160 and          flux. The wetting force at 2 seconds for 95Sn/5Sb on a Cu
175°C. In combination with Pb from the PCB or                 wire is significantly less than Sn/37Pb and Sn/3.5Ag. In
component metallisation, a Sn/Bi/Pb ternary compound          addition to marginal wetting performance, the toxicity of Sb
is formed with a melting point of only 96°C. As the trend     has also raised concerns. As with bismuth, antimony is also a
toward eliminating lead continues, this alloy may become      by-product in the production of lead.
more attractive.
                                                              9.       In/Sn (52In/48Sn:118°C)
6.       Sn/Ag/Bi/Cu        (90Sn/2.0Ag/7.5Bi/0.5Cu           The melting point of this alloy makes it suitable to low
         (138):198-212°C)                                     temperature applications. With regard to indium, it displays
Although the addition of Bi to the Sn/Ag/X system             good oxidation resistance, but is susceptible to corrosion in a
imparts greater strength and improved wetting, too much       humid environment. It is also a very soft metal and has a
bismuth (greater than 5%) leads to the presence of a          tendency to cold weld. In addition, the 52In/48Sn alloy
small DSC peak near 138°C, corresponding to the binary        displays rather poor high temperature fatigue behavior, due to
Sn/Bi eutectic at 138°C or the ternary Sn/Ag/Bi eutectic      its low melting point. The high indium content limits the
at 136.5°C. For this alloy with 7.5 weight percent Bi, this   widespread use of this alloy due to cost and availability
corresponds to approximately 1% of the total melting.         constraints.
This small amount of eutectic melting has an uncertain
effect on joint reliability as the temperature approaches     10.       Sn/Zn (91Sn/9Zn:199°C)
138°C. This combined with the aforementioned concern          The presence of zinc in solder alloys leads to oxidation and
of forming a SnPbBi compound at 96°C, makes this alloy        corrosion. Samples of bulk alloys that were steam aged for 8
an unlikely candidate for a Pb-free solder.                   hours exhibited severe corrosion as evidenced by a purplish
                                                              color. In powder form, it reacts rapidly with acids and alkalis
7.       Sn/B (42Sn/58Bi:138°C)                               and forms a gas. Zinc-containing solder alloys have been
The low melting point of this alloy makes it suitable for     known to react with the flux medium in as little as a day,
soldering     temperature-sensitive  components      and      resulting in a paste that is “hard as a rock”. Thus, its
substrates. If these contain Pb, the SnPbBi ternary           compatibility with fluxes and its storage stability is
eutectic compound may form at 96°C, which in turn             questionable. The reflowed solder joints do not wet as well as
adversely affects the thermal fatigue properties. The         other lead-free alloys. When wave soldered, this alloy tends
NCMS Lead Free Solder Project recently reported the           to produce excessive dross. Therefore, manufacturability of
results of thermal cycle testing at 0/ 100°C and -55/         this alloy and zinc-bearing alloys in general is a concern.
+125°C for over 5000 cycles on OSP boards. The result
was that the Sn/Bi outperformed the Sn/Pb at both             11.       Au/Sn (80Sn/20Au:280°C)
temperature excursions. It was thought that the closeness     Au/Sn eutectic solder is a very strong, rigid solder due to the
of 125°C to the binary Sn/Bi eutectic at 138°C would          formation of brittle intermetallic compounds. Problems of
cause this alloy to be a poor performer. Two possible         cracked dies have been seen using Au/Sn eutectic solders in
explanations for this unexpected result were presented.       die attach applications. It is not known if the cracks occur
The Sn/Bi alloy may be annealing at 125°C, relaxing the       from processing or during thermal cycling. The high cost of
stresses produced during thermal cycling. A second            this alloy restricts its use in many applications where cost is a
explanation was the alloy may be undergoing                   factor.
recrystallization.
DISSOLUTION KINETICS OF COPPER                                                                       positioned so it made contact with the copper wire. The time
In the electronics industry, copper is commonly used as a                                            it took to break the copper wire (i.e. until the copper dissolved
basis material for                                                                                   in the solder) was measured and recorded as the dissolution
• conductor traces and solder pads on the PCB                                                        time.
• lead frames of SO, QFP, PLCC, and other
    components.                                                                                      The following alloy compositions were tested:
Alloys with a high tin content and a higher melting point                                            • 60Sn/40Pb,
have a greater tendency to dissolve copper. If a larger                                              • 95Sn/4Ag/1Cu,
quantity of copper is dissolved from the base metal into                                             • Castin 96.2Sn/2.5Ag/0.5Sb/0.8Cu,
the solder material, there is excessive formation of the                                             • 95.5Sn/4Ag/0.5Cu,
Cu6Sn5 intermetallic phase. Solder joint reliability can                                             • 88Sn/3Ag/8.5Bi/0.5Cu,
be adversely affected by the brittle nature of this                                                  • 88.42Sn/3.07Ag/8.51Bi,
intermetallic compound, in particular the mechanical                                                 • 99.3Sn/0.7Cu,
properties of the solder joint, especially under high                                                • 96.5Sn/3.5Ag.
impact conditions.[3]                                                                                60Sn/40Pb had the slowest rate of dissolution of the copper
                                                                                                     wire as expected due to its lower Sn content. For the high tin
The extent of copper dissolution in various alloys can be                                            solders, the graph shows that the addition of 0.5% copper to
evaluated by means of a simple test. A 50 gram weight                                                the solder alloy can decrease the dissolution rate dramatically.
was attached to a 125 µm diameter copper wire. A small                                               In the case of the Sn/Ag/Bi alloy, the effect of adding 0.5%Cu
quantity of A611 liquid flux was brushed on the copper                                               was to increase the dissolution time from 1.5 minutes for
wire. Then the alloy was placed on the tip of a soldering                                            Sn/Ag/Bi to 3 minutes for Sn/Ag/Bi/0.5Cu.
iron (pre-heated to 280°C), and the soldering iron was




                         10
                                  8.76
Dissolution Time (min)




                          8

                          6
                                                  4.1
                          4                               3.42
                                                                               3
                          2
                                                                                             1.8       1.8          1.56     1.25

                          0
                                0P
                                  b           u
                                                          sti
                                                             n             u             u
                                                                                                     -C
                                                                                                       u           Bi        -A
                                                                                                                               g
                                            1C                           -C            5C                        g-
                              /4          g-            Ca             Bi            0.            Sn          -A          Sn
                            Sn          -A                           g-            g-                        Sn
                          60          Sn                           -A            -A
                                                                 Sn            Sn
Figure 1. Dissolution Kinetics of Copper in Several Solder Alloys.

THE EFFECT OF ISOTHERMAL AGING                                                                       gram batch of each paste was mixed in a small production-
It is important to study intermetallic growth formation                                              scale Ross mixer at 89.5% metal loading/ 10.5% flux by
because in solder joints with coarsened Cu-Sn                                                        weight. Heraeus V365 no-clean/ halide-free flux was used.
intermetallics, fracture is brittle and occurs through the
intermetallic layer. An aging study was performed on                                                 The test pieces were 2” x 2” copper coupons cut from 0.021
96.5Sn/ 3.5Ag and 95.5Sn/ 4Ag/ 0.5Cu solder alloys on                                                inch thick, commercial grade alloy 110 copper foil. They
copper substrates.      The intermetallic layer growth                                               were then pressed flat and cleaned in acetone. The solder
characteristics of the two alloys were compared in order                                             paste was screen printed through an 8 mil thick, stainless
to determine the effect of copper addition to Sn-Ag based                                            steel, laser cut stencil on a DEK 247 printer with all printing
alloys.                                                                                              parameters kept constant. Therefore, the solder volume is
                                                                                                     presumably constant and was not considered a factor in this
Experimental Procedure                                                                               study. Six coupons of each alloy were printed. The printing
Two solder pastes were made, 96.5Sn/ 3.5Ag and                                                       characteristics of both pastes were very good.
95.5Sn/ 4Ag/ 0.5Cu. The pastes were made with -325/
+500 mesh electronic grade (Type 3) powder. A 1000
The test coupons were reflowed at Heraeus in a nitrogen       hours). Following aging, each sample was sectioned across 3
convection reflow oven using a standard profile for the       joints for metallographic examination. For each sample, the
pastes. The test pieces were then placed in a Lindberg/       average thickness of the resulting interfacial compound was
Blue M air convection oven held at 150°C. The samples         reported.
were aged for periods of 2, 4, 11, 20, and 41 days (984




Table I. Intermetallic Thicknesses for Solder Alloys Aged at 150°C.
            Time (hours)                          Intermetallic Thickness (µm)* (Cu3Sn + Cu6Sn5)
                                      96.5Sn/ 3.5Ag                           95.5Sn/ 4Ag/ 0.5Cu
                  0                   0.25 + 2                                0.25 + 2
                 48                   0.5 + 3.25                              0.5 + 2.5
                 96                   0.75 + 2.5                              0.75 + 2.5
                 264                  1 + 2.5                                 1+3
                 480                  1.5 + 3                                 1.5 + 4.5
                 984                  2.5 + 4                                 2.5 + **
*The standard deviation for the measurements is on the order of 0.5 µm.
**A nonuniform morphology of the Cu6Sn5 layer precluded a characteristic thickness measurement.
                                                               Recent work indicates that similar failure mechanisms are
Results and Discussion                                         involved in thermal fatigue in shear and unidirectional creep
It is widely known that copper is soluble in molten Sn-        in shear. Also, since the temperatures during thermal fatigue
Ag-X solders. The dissolution of copper results in the         represent high solder homologous temperatures, creep
formation of ε-phase Cu3Sn and η-phase Cu6Sn5. Due to          deformation is involved. Creep deformation is the time-
the concentration gradient, the Cu-rich Cu3Sn phase            dependent plastic flow of a material under constant load at
forms adjacent to the copper substrate. Cu3Sn has a more       elevated temperature. As the homologous temperature (the
planar structure. The more Sn-rich Cu6Sn5 phase forms          ratio of the test temperature to the melting temperature on an
adjacent to the Sn-based solder and has a scallop-edge         absolute temperature scale) increases, the ease with which
appearance. The reason why Cu6Sn5 has a scallop-edge           plastic flow occurs also increases. Creep is significant at a
appearance may be due to the fact that Cu6Sn5 dissolves        homologous temperature greater than 0.5. Therefore, creep
faster along the grain boundary. Between the Cu6Sn5            deformation occurs in solders even at room temperature.
grains, there are molten solder channels extending all the     Every high temperature excursion results in a straining of the
way to the Cu3Sn/Cu interface. Since the Cu3Sn                 solder joint as the constraining materials expand different
compound layer is so thin, these channels serve as fast        amounts. By understanding the mechanisms that lead to
diffusion and dissolution paths of Cu in the solder to feed    fatigue failures, researchers can use the appropriate
the interfacial reaction.[4] This interfacial layer grows      metallurgical strategy to slow down or stop these mechanisms
during solid-state aging as the tin and copper diffuse to      and thus develop an improved, more fatigue resistant solder
the interface and react.                                       alloy.[6]

The growth kinetics of the intermetallic compounds was        The elevated operating temperature and operative strain rates
found to be similar as expected due to the similar Sn         imply that creep is the major deformation mode during low
contents and reflow temperatures of the two alloys. The       cycle fatigue. Also, the observation that solder joint fatigue
microstructural features of the Sn-Ag-X alloys are also       failures and creep failures appear the result of similar
similar. The matrix is polygranular Sn with a grain size      metallurgical mechanisms indicates that both techniques can
in the as-solidified condition of approximately 1 µm.[5]      be used to study the fatigue failure mechanism and relative
Five phases can be seen in the SEM micrographs given          solder alloy fatigue resistance. As such, it becomes important
in Figures 6 to 17: Sn, Ag3Sn, Cu6Sn5, Cu3Sn, and Cu.         to understand how the solder microstructure accommodates
                                                              the applied strain. For the Sn-Ag-X solders, the strain
CREEP DEFORMATION                                             accommodation occurs through the tin matrix at individual
                                                              Sn-Sn grain boundaries.[5]
                                                                3 different loads (4, 8, and 16 MPa) at both temperatures.
Creep testing was performed on samples of the same              Results for time to failure at 100, 500, and 1000 hours were
dimension and preparation method as that used for               recorded. Loads were applied which were expected to give
standard tensile testing of solder alloys. All creep testing    lifetimes in the region of those times but extrapolation was
was performed at International Tin Research Institute           carried out to estimate values for the times required. The
(ITRI). To generate the creep-rupture data, the solder          results at 25°C are shown in Figure 2. To interpret the data,
alloys were cast into dumbbell-shaped test samples              compare the times to rupture for a similar applied stress on
having 20 mm gauge length and 2 mm diameter. They               the two alloys. For example, for an applied stress of 4 MPa,
were cast at a temperature of 50 degrees above the              Sn/40Pb failed         after    265    hours,    whereas     the
liquidus into a heated steel mold. The mold was then            95.5Sn/4Ag/0.5Cu alloy took 3000 hours for failure to occur.
water cooled. Samples were then subjected to the                Figure 3 presents the data collected at ITRI for several
standard aging procedure of 24 hours at 125°C, in               candidate lead free alloys compared to Sn/40Pb. The
addition to at least 24 hours at room temperature for the       95.5Sn/4Ag/0.5Cu alloy performs the best at room
benefit of stabilizing the microstructure as much as            temperature compared to Sn/3.5Ag eutectic, Sn/0.7Cu
possible.                                                       eutectic, and Sn/40Pb. As expected, the Sn/Ag/Cu and
                                                                Sn/Ag alloys behave similarly due to their similar
Samples were held isothermally at both room                     microstructural development. The graphic representation of
temperature and 100°C. A weight was hung from the               the Sn/Cu data greatly differs from that of the other three,
sample during the test representing an applied stress, and      perhaps indicative of a different failure mechanism. The
the time to rupture was recorded. Samples and test              results of creep testing at 100°C are presented in Figure 4. At
method conformed to the British Standard BS3500: part           100°C, the Sn/Ag and Sn/Ag/Cu curves appear switched from
3: 1969 “Method for Creep and Rupture Testing of                the 25°C results with the best performer now being the Sn/Ag
Metals.” Time to rupture was determined by measuring            eutectic alloy. Figure 5 shows the creep-rupture data for the
electrical resistance across the sample; after fracture the     Heraeus Sn/4Ag/0.5Cu alloy tested at both room temperature
resistance became infinite and timing stopped. All              and 100°C. As expected, higher temperatures allow materials
samples were tested in duplicate. Data was collected for        to creep at a faster rate, thereby reducing the time to failure.




                                        Creep Data at 25°C
                             40
            Applied Stress




                             30
                (MPa)




                             20
                             10
                              0
                                  0.1     1       10           100      1000       10000
                                              Time to Rupture
                        Sn-40Pb                                   Heraeus Sn-Ag-Cu
                        Log. (Sn-40Pb)                            Log. (Heraeus Sn-Ag-Cu)
  Figure 2. Creep-Rupture Data for Heraeus Sn-4Ag-0.5Cu and Sn-40Pb at 25°C.
                                      Creep Results at 25°C

                           35
    Applied Stress (MPa)



                           30
                                                                                  Sn-40Pb
                           25
                                                                                  Sn-Ag-Cu
                           20

                           15                                                     Sn-3.5Ag

                           10                                                     Sn-0.7Cu
                            5

                            0
                                0.1    1        10         100   1000   10000

                                      Time to Failure (hours)

Figure 3. Creep-Rupture Data for Several Candidate Lead Free Alloys Compared to Sn-40Pb at 25°C.



                                      Creep Results at 100°C

                           35
   Applied Stress (MPa)




                           30
                                                                                   Sn-40Pb
                           25
                                                                                   Sn-Ag-Cu
                           20

                           15                                                      Sn-3.5Ag

                           10                                                      Sn-0.7Cu
                            5

                            0
                                1          10        100         1000     10000
                                           Time to Failure (hours)
Figure 4. Creep-Rupture Data for Several Candidate Lead Free Alloys Compared to Sn-40Pb at 100°C.
                                            Creep Results of Heraeus Sn/Ag/Cu Alloy
            Applied Stress (MPa)   35
                                   30
                                   25
                                   20
                                   15
                                   10
                                    5
                                    0
                                        1          10         100              1000               10000
                                                        Time to Rupture
                                                          25°C      100°C
Figure 5. Creep-Rupture Data for the Heraeus Sn/4Ag/0.5Cu Alloy Tested at 25°C and 100°C.

CONCLUDING REMARKS                                            scanning electron microscopy of the samples used to study
Recent work with candidate lead free alloys indicate a        aging was performed by F.W. Gayle and L. Smith at the
significant improvement in reliability over Sn/Pb.            National Institute of Standards and Technology and is greatly
Figures 2-4 clearly demonstrate a superior creep              appreciated. Also, thanks to Dr. M.R. Notis of Lehigh
resistance over Sn/Pb for all lead free alloys tested         University and A. Z. Miric of Heraeus Germany for their
including Sn/Ag eutectic, Sn/Cu eutectic, and                 useful discussions.
Sn/4Ag/0.5Cu at both room temperature and 100°C.
Although the Sn/Cu eutectic outperformed Sn/40Pb, it          REFERENCES
did not perform as well as the Sn-Ag-X alloys.                1. National Center for Manufacturing Sciences, Lead-Free
                                                                   Solder Project Final Report, August 1997.
An aging study of both the 95.5Sn/4Ag/0.5Cu and               2. P.T. Vianco, K.L. Erickson, and P.L. Hopkins, “Solid
96.5Sn/3.5Ag solder alloys was performed in order to               State Intermetallic Compound Growth Between Copper
evaluate the growth kinetics of the intermetallic layers           and High Temperature, Tin-Rich Solders-Part I:
following extended heat treatment. An understanding of             Experimental Analysis,” Sandia National Labs (Contract
the microstructural evolution that occurs at the                   Number DE-AC04-94AL85000), 1994.
solder/copper interface at elevated temperatures is helpful   3.   G. Humpston and D.M. Jacobson, Principles of Soldering
to understand the failure mechanisms that dominate at              and Brazing, ASM International, 1993.
elevated temperatures. Creep occurs when materials            4.   H.K. Kim and K.N. Tu, “Kinetic Analysis of the
under constant stress, below the tensile stress, slowly            Soldering Reaction Between Eutectic SnPb Alloy and Cu
deform and finally fracture. The creep rate is dependent           Accompanied by Ripening,” Physical Review B, Vol. 53,
on alloy composition and microstructure and is strongly            No. 23, June 15, 1996, p. 16028.
temperature dependent. Because Sn/Ag and Sn/Ag/Cu             5.   D.R. Frear, “The Mechanical Behavior of Interconnect
have similar microstructures, they behave similarly                Materials for Electronic Packaging,” J.Metals, (May
during isothermal aging and creep testing.                         1996), pp. 49-53.
                                                              6.   J.W. Morris, Jr. and D. Tribula, “Creep in Shear of
ACKNOWLEDGEMENTS                                                   Experimental Solder Joints,” Journal of Electronic
The author is grateful to acknowledge the support of the
                                                                   Packaging, June 1990, Vol. 112, pp. 87-93.
International Tin Research Institute for mechanical
testing of the alloys. The metallographic preparation and

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:23
posted:10/31/2011
language:English
pages:8