Silver Scavenging Inhibition
of Some Silver Loaded Solders
The results of experiments show that silver dissolved in a
solder alloy has an inhibiting effect on the scavenging of
silver from a silver substrate
BY R. A. BULWITH AND C. A. MACKAY
ABSTRACT. Comparisons of the rates of
dissolution of silver in some binary t i n /
lead alloys and some silver-bearing alloys
were reported. Activation energies for
each alloy were calculated, and an alloy
selection system was suggested together
with a discussion on the dissolution inhibition mechanism.
Introduction
The use of silver loaded solders to
prevent or reduce the rate of solution of
silver from substrates during the soldering
process has become widely accepted.
However, development of alloys for this
purpose have almost (Refs. 1,2) entirely
been by analogy with the effect of copper on inhibition of copper solution by
solders (Refs. 3-5). In fact, the analogy is
an extremely good one; because in both
cases —copper in tin/lead and silver in
tin/lead — the systems are eutectic. Also,
in both cases, the ternary eutectic being
three phase involves tin-rich and lead-rich
solid solutions and an intermetallic.
In copper containing alloys, the intermetallic was Cu6Sn5 while for silver-bearing alloys it was Ag3Sn. The mechanism of
solution inhibition would, therefore, presumably also be similar. Because silverloaded alloys have been used for so long
with apparent success, the analogy has
not been challenged. However, the performance of different alloys, which have
over the years been proposed and developed, have never been compared.
This paper reports the results of tests
to measure the silver dissolution rates on
sheet silver over a range of temperatures
and for a range of alloys.
ble, according to the ASTM standards.
Where no standard existed, the alloys
were made to the impurity levels as
specified by ASTM (Ref. 6).
Table 1 lists the range of solders examined in this investigation together with
their melting points. The silver strip used
was 99.99% pure Ag.
Experimental
Silver strip of 0.05 in. (1.27 mm) thickness was cut into 0.20 X 1.0 in. (5 X 25.4
mm) coupons. A 0.20 X 0.40 in. (5 X 10
Materials
mm) area was then outlined by masking
the remainder of the sample with a solder
The materials used in this investigation
resistant compound (i.e., Alpha Solder
were standard Alpha Metals production
Resist 2398).
materials manufactured, where applicaApproximately constant volume samples of the solder alloys were melted into
Based on a paper presented at the 14th
International A WS-WRC Brazing and Soldering 10 ml pyrex glass beakers which fitted
Conference held in Philadelphia, Pennsylvania, completely into a recess in a 5.25 X 1.5
in. (134 X 39 mm) steel guard ring heated
during April 26-28, 1983.
on a Thermoline temperature controllaR A. BULWITH is engaged in research and
ble hot plate. The guard ring eliminated
product development at Alpha Metals, Newdrafts from the sides of the beaker and,
ark, N. J C A. MACKAY, formerly with Alpha
Metals, is currently Director — Research and due to its large mass, served to stabilize
and unify the alloy temperature. The
Development for Allied Semi-Alloys, Mt. Versilver strip samples were mounted in split
non, N. Y.
Table 1—Compositions (%) and Melting Points for Range of Solders Examined
Sn
Pb
60.0
62.0
62.5
96.5
98.5
95.0
10.0
40.0
36.0
36.1
—
1.0
50.0
49.0
—
-
88.0
97.5
97.5
40.0
50.0
86-s I M A R C H 1985
Ag
Sb
ASTM
GRADE
Cu
60A
—
2.0
1.4
3.5
—
96.5TS
1.5
95TA
5.0
2.0
2.5
1.5
-
1.0
2.5S
1.5S
10.0
Melting range
°C
°F
183-185
179-183
179
221
227-280
232-240
268-302
304
309
120-167
179-210
(361-365)
(354-361)
(354)
(430)
(441-536)
(450-464)
(514-575)
(580)
(588)
(248-333)
(354-410)
120
IMMERSION TIME
Fig. 1-Effect of silver additions to near eutectic alloy showing silver
dissolution inhibition
rubber holders and suspended in the
molten alloys for times between 30 seconds (s) and 2 minutes (min).
The dipped preweighed masked coupons were then stripped of the solder
coating by immersion in an aqueous solution of 50 g/l NaOH with 35 g/l onitrophenol at 70°C (158°F). Weight loss
was determined using a Sartorius 2462
precision balance capable of measuring
to 0.1 mg.
In the first stage of the investigation, a
single weight loss vs. time curve was
established. Holding temperatures of
liquidus + 50°C (90°F) for non-eutectics
were used for this stage. Subsequently,
the most common alloys used either in
normal soldering or in microcircuitry
were examined at three temperatures,
melting point + 30°C, + 5 0 ° C and
-F80°C ( + 54°F, 4-90°F, -F144°F) from
180
IM1ERS10N TINE (SECS.)
(SECS.)
Fig. 2 —Effect of saturation of tin or tin/lead with other elements on
silver dissolution rate
which Arrhenius plots were constructed.
Arrhenius plots could be obtained. These
are shown in Fig. 5.
Results
The effect of adding silver to a tin/lead
alloy was demonstrated; the results,
which are for both 60 Sn/40 Pb and
62Sn/36Pb/2Ag alloys, are shown in Fig.
1. Dissolution rates of various tin and
tin/lead solders with tin/silver solder are
compared in Fig. 2, which shows the
weight losses in 95Sn/5Sb, 98.5Sn/
1.5Cu, 60Sn/40Pb, 50Sn/40Pb/10Bi and
96.5Sn/3.5Ag.
The results for the higher tin content
silver-containing alloys are shown in Fig.
3, while those for the high lead alloys are
reported in Fig. 4. Those alloys showing
the lowest dissolution rates were then
further examined at superheats of +30°C
and +80°C (4-54°F and + 144°F) so that
120
180
IJMESSION TIME <SECS.)
Discussion
Figure 1 shows the decrease in solubility of silver produced by saturation of the
liquid alloy by silver at a melt superheat of
50°C (90°F). The rate of dissolution has
been reduced from 0.20 mg/s to 0.07
mg/s, demonstrating the dissolution inhibition effect claimed.
Saturation of the tin constituent of
solders — generally considered the active
wetting component and also the reactive
compound forming phase —with other
elements such as copper or antimony
does not appear to inhibit silver dissolution—Fig. 2. In addition, the rates of
dissolution appear to increase as a function of the alloy melting point, thereby
120
180
IMMERSION TIME
(SECS.)
Fig. 3 — Effect of changing tin and lead levels of silver-containing solders Fig. 4 - Comparison of silver dissolution rates for high lead alloys with
on silver dissolution rate
eutectic tin/silver alloy
WELDING RESEARCH SUPPLEMENT 187-s
-
0,0
s
a
O
>
ui
a
X
o
cc
<c
Ul
v>
Ul
rr
- -1.0
- -2,0
-3,0
60 S N / 4 0 P B
Z
ui
£
a.
- -4,0
O
-
SN/36 PB/2 AG
\97,5
-6.0
X
o
c£
Ul
10 SN/88 PB/2 AG
(/>
Ul
z
ui
2
a.
O
_i
ui
>
Uf
a
x
u
CE
J
t
0,15
I
I
I
J
L
0,20
1/TOK (xlOO)
Fig. 5 - Arrhenuis plots for 60Sn/40Pb, 62Sn/36Pb/2Ag, 97.5Pb/2.5Ag and WSn/88Pb/2Ag
0,10
<
Ul
H
Z
v////.
a.
O
lSn/97,5
Pb/l,5Ag
300
X
o
oc
<
ui
(A
Ul
OC
h-
z
ui
2
a.
O
_i
ui
>
Ul
Q
X
o
OC
<
Ul
X///A
W7A
97,5 Pb/
2,5Ag
•350
10 SrV88
PLV2Ag
200
49Sn/50
Pb/lAg
/ / / /
• I SOSnAiOPb
52SP/3S
Pb/2Ag
+
96,5Sn/
3.5Ag
+
250
suggesting no effect on tin dilution by the
alloying additions.
When the results for silver-containing
solders were examined, however, the
same function of alloy melting point was
not apparent in that some of the higher
melting point solders had significantly
lower dissolution rates. As an example,
the 1Sn/97.5Pb/1.5Ag alloy shown in Fig.
3 with melting point 309°C (588°F), has a
lower silver dissolution rate than does
96.5Sn/3.5Ag with a melting point of
221 °C (430°F). Nor was there any obvious indication that a particular tin/silver
or tin/lead or even lead/silver ratio was
superior.
In absolute terms the lowest dissolution rate was exhibited by the 10Sn/
88Pb/2Ag alloy both from the point of
view of reaction rate (slope of curve) and
weight loss measurements, by a considerable amount.
According to their measured reaction
rates, both the 62.5Sn/36.1Pb/1.4Ag and
62Sn/36Pb/2Ag solders dissolve the silver at the same rates. However, owing to
the nonlinear rate of dissolution at short
times with the first of these solders, the
weight loss curves were not identical.
These differences are probably not significant, showing that eutectic and near
eutectic alloys behave similarly. Both the
lead rich alloys tested also fell in this
region of reaction rate.
The same considerations that govern
selection of a soldering alloy — in the case
of say, PCB assembly and fabrication —
come into play when selecting an alloy
that will not dissolve away a silver coating
or conductor track. When fabricating a
hybrid component, for example, once
the highest melting point alloy consistent
with continued electronic performance
of the device to be fixed and minimal
dimensional distortion, has been used,
subsequent assembly or fabrication
stages must use lower melting point
alloys to avoid destruction or disruption
of the existing joints. Thus, although the
10Sn/88Pb/2Ag would appear to be the
best alloy from a silver leaching standpoint, it could be necessary to use either
1Sn/97.5Pb/1.5Ag or 97.5Pb/2.5Ag on
account of the melting points and so that
96.5Sn/3.5Ag alloy may be used at a later
assembly stage.
Figure 6 shows a selection diagram
indicating the melting points of the vari•150
100
ous alloys, their suggested usable temperature ranges, and their observed silver
dissolution rates. The best solder choice
would be that showing the lowest solution rates which lie in the temperature
block corresponding to the use, i.e.,
0,1
0,2
0,3
0.4
those in the cross-hatched region. Judging from this chart, the preferred solders
for the whole range of possible applicaSILVER DISSOLUTION RATE (mg/s)
Fig. 6 —Selection diagram for low silver dissolution rate alloys in order of melting range. Hatched tions are 62Sn/36Pb/2Ag, 10Sn/88Pb/
2Ag and lSn/97.5Pb/1.5Ag.
region is preferred selection region
88-sl M A R C H 1985
50Sn/40Pb/
10 Bi
-
as:
1,0
Ln (rate) = — + A
KI
95 SN/5 SB
0.0
S.5SN/1.5CU
-1.0
96.5SN/3.5AG
49 SN/50 PB/1 AG
97.5 PB/2.5 AG
1SN/97.5PB/1.5AG
4
+
>50 SN/40 PB/
\
IAG
t
+
62 SN/36 PB/2 AG
10 SN/88 PB/2 AG
\
t
J
0,10
4
V 52.5 Sw/36.1 PB/
--3.0
-4'0
5 0 S N / 4 0 PB
\
X
L
0,15
0.20
(1)
an activation energy value of 17.6 (Real
units) was obtained. When the corresponding value for 96.5Sn/3.5Ag alloy
was plotted, this was below the line
shown.
Our assumption is that alloying with
antimony, copper or bismuth did not
affect the silver dissolution rate. If this is
correct, then both the activation energy
value for the binary tin-lead solder in Fig.
5 and the value for the range of nonsilver bearing solders in Fig. 7 should be
similar. In fact, the agreement between
the t w o is quite good.
Plotting the results for the silver-containing alloys in Fig. 7 showed that all fall
below the line to differing extents. If we
can assume that all alloys tested will make
the same intercept with the y (1n (rate))
axis (i.e., that they have the same frequency factor), then the straightline
through the point for each silver bearing
alloy would have a steeper slope, thus
indicating an increased activation energy.
From Fig. 5 the intercepts for each alloy
except 10Sn/88Pb/2Ag lay in the range
(—13-*-—24) cm-s" 1 , and from Fig. 7 it
was —16 cm-s - 1 . Considering the closeness of the temperature range from
which they were determined, this was
probably an acceptable agreement.
_L
From equation (1) it was clear that:
T (Ln (rate) — A) = constant.
Fig. 7—Arrhenius plot for all alloys tested. Line links alloy of same activation energy. The further For the point on the straight line relaaway from line the alloy point then the higher the activation energy and the lower the silver
tionship in Fig. 7 this was sensibly the
dissolution rate
same and equal to the slope of the line.
For the silver-containing alloys it was
reduced by an amount indicative of the
the melting point + 50°C (90°F) temperFigure 5 shows the Arrhenius plots for
distance of the point for that alloy from
ature (the actual experimental temperathese solders. From this, the dissolution
the straightline. The ratio of any t w o
ture).
rate for each alloy for any temperature
constants, therefore, indicates the relacan be found and the weight loss curve
The implication of such a curve is that
tive order of merit for silver dissolution
reconstructed. From each line, the activathe activation energies of dissolution of
inhibition for that alloy.
tion energy for solution of silver from a
silver into each molten alloy were identisolid surface into the melt can be calcucal. In other words, neither antimony,
Using this hypothesis, the differences
lated, and these are listed in Table 2.
copper, lead nor lead + bismuth has any
between the constant for points off the
effect on the energy barrier which the
line with those defining the line represent
As mentioned previously, the results
silver atom must surmount in order to
some measure of the effectiveness of
for non-silver containing solders appear
dissolve into the melt. From the slope of
that alloy inhibiting silver dissolution.
to increase as a function of temperature.
this line and the Arrhenius equation given
Table 3 indicates the relative order of
If the natural log of the dissolution rates
for these solders was plotted against the
inverse of the melting points in K, then a
near straightline plot is obtained. In fact,
Table 3—A Relative Order of Merit for Eleven Alloys
Fig. 7 shows such a plot, although it is for
np+5o °K (x 100)
Alloy
Table 2—Activation Energies for Five Alloys
Alloy
Activation
energy,
kcal
60/40
62/36/2
1/97.5/1.5
-/97.5/2.5
10/88/2
15.04
23.81
29.24
34.14
45.62
60/40
98.5/1.5Cu
95/5Sb
50/40/10Bi
96.5/3.5Ag
49/50/1Ag
62.5/36/ 1.4Ag
62/36/2 Ag
-/97.5/2.5Ag
1/97.5/1.5 Ag
10/88/2
Constant
Ratio
-8951
-8822
-8849
-8824
-9339
-9475
-9474
-9527
-11451
-11677
-12280
1.009
0.9942
0.9973
0.9945
1.0535
1.068
1.066
1.074
1.291
1.316
1.384
Inhibition
_
5%
7%
7%
7%
29%
32%
38%
WELDING RESEARCH SUPPLEMENT 189-s
merit for the alloys tested, the % inhibition figures being the difference of these
same ratios expressed as a percentage
(constant
determined
from
Fig.
7 = -8873).
On the basis of these results some
comments may be made about the
mechanism of silver dissolution inhibition.
Using a different dissolution measuring
technique, both Bader (Ref. 3) and Berg
and Hall (Ref. 2) showed a similar near
linear relationship between the quantity
of dissolved metal and immersion to
those reported in Figs. 1-4. O n the basis
of the work of Lommel and Chalmers
(Ref. 7), they showed that the dissolution
mechanism must be either governed by a
surface reaction or limited by extreme
thermal agitation or stirring producing
continuous limitation of the interface
boundary layer.
In support of this theory, Berg and Hall
recalculated Bader's results and obtained
an activation energy for silver dissolving
into 60Sn/40Pb alloy of 11.5 kcal, which
agrees with the results presented here.
This model, however, does not explain
the marked increase in activation energy
for the silver dissolution inhibiting alloy
found by Berg and Hall or as reported in
this research.
It has been well documented that the
rate of transfer of atoms for a high
concentration region was dependent
upon the concentration gradient because
of the concentration dependence of the
diffusion constant. Thus, for example,
annealing times for low levels of alloying
addition in the iron/tin system were
longer than for higher levels (Refs. 8,9). It
could be argued, therefore, that the
reduction in concentration gradient
brought about by small alloying additions
to the molten alloy could modify the
dissolution rates. However, this does not
appear to be the case, since results
reported here do not indicate a monotonic decrease in dissolution rate with
increasing level of tin addition.
Another explanation could be that the
rate limiting function was the formation
of intermetallic compound from saturated solid solution so that dissolving
material was always attempting to super
saturate the molten alloy. In such an
instance, it might be expected either that
the more silver saturated alloys would
show greater inhibition. This, however,
was not the case. Compare, for example,
62Sn/36Pb/2Ag with 62.5Sn/36.1Pb/
1.4Ag and 1Sn/97.5Pb/1.5Ag or consider
that since all alloys tested were at or
exceeded their eutectic silver content
then all would behave equally, which
again was not the case.
The effect appears to be due to a
change in the activation energy of dissolution so that an atom of silver in the solid
had to be excited to a higher energy level
to dissolve into a molten silver bearing
solder than for molten silver free alloys.
Conclusion
These experiments have shown that
silver dissolved in a solder does have an
inhibiting effect on the scavenging of
silver from a silver substrate. The effectiveness of the various alloys commonly
used for this purpose does not appear to
be readily predictable.
Recommendations of most suitable
alloys for particular temperature intervals
should be based on experimental results.
Such recommendations have been
offered in this paper.
The effect of dissolved silver in the
molten alloy appears to be to increase
the activation energy for dissolution; consequently, a silver atom in the solid must
be raised to a higher energy level in order
to dissolve into a silver containing solder
than was necessary with silver free alloys.
Accordingly, for a given temperature,
correspondingly fewer atoms pass into
solution in a fixed time interval, and a
silver substrate does not dissolve away.
A ckno wledgments
The authors thank the management of
Alpha Metals for permission to publish
this paper and Paul Lotosky for his help
with the experimental work.
References
1. Alpha Technical Bulletin, 6a-36. Newark,
New Jersey: Alpha Metals.
2. Berg, H., and Hall, G. L. 1973. Proceedings 11th Annual Reliability Phys. Symposium,
pp. 10-20.
3. Bader, W. G. 1969. Dissolution of Au,
Ag, Pd, Pt, Cu, and Ni in a molten tin-lead
solder. Welding Journal 48(n):551-s to 557-s.
4. Bader, W. G. 1975. Lead alloys for high
temperature soldering of magnet wire. Welding Journal 54(10): 370-s to 376-s.
5. Allen, B. M. 1972 (Feb.). Wireless
World.
6. ASTM B-32.
7. Lommel, J. M., and Chalmers, B. 1959.
Trans. AIME 215: 499.
8. Speight, E. A. 1972. Metal Science Journal
6:57-60.
9. MacKay, C. A. 1971. Journal of Applied
Crystallography 4 ptl: 77-78.
WRC Bulletin 296
July 1984
Fitness-for-Service Criteria for Pipeline Girth-Weld Quality
By R. P. Reed, M. B. Kasen, H. I. McHenry, C. M. Fortunko and D. T. Read
In this report, criteria have been developed for applying fitness-for-service analyses t o flaws in the
girth welds of the Alaska Natural Gas Transmission System pipeline. A critical crack-openingdisplacement elastic-plastic fracture mechanics model was developed and experimentally modified.
Procedures for constructing curves based on this model are provided. A significantly improved ultrasonic
m e t h o d for detecting and dimensioning significant weld flaws was also developed.
Publication of this report was co-sponsored by the National Bureau of Standards and the Weldability
Committee of the Welding Research Council. The price of WRC Bulletin 296 is $18.00 per copy, plus
$5.00 for postage and handling. Orders should be sent w i t h payment t o the Welding Research Council,
Room 1 3 0 1 , 345 E. 47 St., New York, NY 10017.
90-s | M A R C H 1985
A m e r i c a n W e l d i n g S o c i e t y 550 N.W. Lejeune Road, P.O. BOX 351040, Miami, Florida 33135
Telex: AMWELD SOC No: 51-9245 (305)443-9353
AN INVITATION TO AUTHORS
to present Brazing Papers at the
17th AWS International Brazing and Soldering
Conference
Atlanta, Georgia, April 15-17, 1986
The American Welding Society's C3 Committee on Brazing and Soldering invites you to present your
outstanding and unpublished work in the field of brazing development, research or application at the
17th International AWS Brazing and Soldering Conference. This event will be held in conjunction with the
Society's 67th Annual Convention at the Georgia World Congress Center, Atlanta, Georgia.
Please submit your abstract(s) by August 15, 1985, to be screened by the C3 Papers Selection
Committee for the 1986 conference. Authors will be notified sometime in November, 1985, regarding
acceptance of their papers.
Each abstract should be sufficiently descriptive to give a clear idea of the content of the proposed
paper. In any case, it must contain not less than 500—but preferably not more than 1000—words.
Manuscripts of approximately 5,000 words must be submitted before final approval is given. Repeated
references to a company and/or the use of advertisement, trade names, trade marks (or expressions
considered as such by the industry) are not permitted. Suitable generic terms must be used, in
accordance with those standardized by the American Welding Society, where applicable.
Papers may be considered for publication in the Welding Journal regardless of acceptance for
presentation at the conference. Topics of particular interest in brazing are:
Applied technologies of (1) aerospace structures, (2) machine tools, (3) nuclear assemblies, (4)
automotive assemblies, (5) electronic equipment, (6) food processing equipment and (7) pressure
vessels. Of special interest is the application of brazing to titanium, aluminum and other base metals
including brazement strength data.
New research and development on (1) brazing filler metals, (2) brazing filler metal/base metal
interaction, (3) nuclear properties of brazements, (4) electronic properties of brazements, (5) corrosion
of brazements and (6) strength of brazed joints.
In addition, papers dealing with educational and informative aspects of brazing production, engineering, research and metallurgy are welcomed if the subject falls within the scope of the session.
Please fill out the Author's Application Form (reverse side), attach abstract thereto and return to AWS.
To assure your paper's consideration for the 1986 conference, your abstract must be postmarked no
later than August 15, 1985.
fo2^ LO • /<c~^^^y
Paul W. Ramsey
Executive Director
IMPORTANT: ABSTRACTS MUST BE AT LEAST 500 WORDS AND BE POSTMARKED NO LATER THAN AUGUST 15, 1985, TO
ASSURE CONSIDERATION.
°E*DCL'N,E6
AUTHOR APPLICATION FORM
AUGUST l o ,
1985
FOR
BRAZING PAPERS
Complete in Full and Return to American Welding Society, Inc., 5 5 0 N.W. LeJeune Rd., P. 0 . Box 3 5 1 0 4 0 , Miami, Florida
33135
Date Mailed
Author's Name
Check how addressed: Mr. •
Title or Position
Ms. D
Dr. •
Other
Company or Operation
Mailing Address
City
State
Zip
Telephone (including Area Code)
If there are to be joint
Authorships, give name(s)
of other author(s)
Name
Address
Name
Address
Proposed title (10 words or less):
ABSTRACT:
•
•
•
•
Type, double spaced, an abstract of not less than 500—but preferably not more than 1000—words on
separate sheets and attach to this form.
Be sure to give sufficient information to enable the Papers Selection Committee to obtain a clear idea of
content of the proposed paper; confine background to about 100 words. Also be sure to emphasize Results
and Conclusions since this material, together with information on what is NEW, will have a very important
bearing on the final decision of the AWS C3 Papers Selection Committee.
If complete manuscript is available, in addition to abstract, please attach two copies to this form.
Application Form and Abstract must be postmarked not later than August 15, 1985, to assure
consideration.
MANUSCRIPT DEADLINES:
•
•
•
All manuscripts must be in the hands of the Papers Selection Committee no later than March 15, 1986. If
received by that date, every effort will be made to publish them in a special Brazing and Soldering issue.
It is expected that the Committee's selections will be announced sometime in November, 1985.
If your paper is made a part of the program, which of the following manuscript deadlines will you be able to
meet?
December 15, 1985 D
January 15, 1986 •
February 15, 1986 •
PRESENTATION AND PUBLICATION OF PAPERS:
•
•
•
Has material in this paper been previously presented in a meeting or published?
Yes •
No D
When?
Where?
Following presentation at the Conference, would you accept invitations to present this paper before AWS
Sections? Yes •
No D
Papers accepted for presentation become the property of the Society with original publication rights
assigned to the Welding Journal.
RETURN TO AWS HEADQUARTERS. MUST BE POSTMARKED NOT LATER THAN AUGUST 15, 1985, TO
ENSURE CONSIDER A TION.
Author's
Signature