ITRI
The Partitioning of Impurities in the Solder-Dross Recovery Process
Report No. TE 47418
Author:
Dr Paul; Harris, Dr Ariela Samuel-Lewis, Mr Kaldev Chaggar
1.
Introduction
One of the drawbacks of the wavesoldering technique is that substantial
quantities of dross are built up, and this must be removed at intervals
and the bath replenished with expensive virgin solder. Analysis of the
dross reveals that it consists of a mixture of oxides and metallic solder,
with normally more than 90% being the latter. The unoxidised metal is
effectively prevented from rejoining the solder in the bath by the presence
of the oxide with which it is intimately mixed.
EVS International
have recently developed a machine to mechanically separate the unreacted
metal so that it can be returned to the bath, and thus dramatically reduce
the quantities of virgin solder that must be purchased. The machine operates
by remelting the dross followed by mechanical compression against a perforated
wall, so as to squeeze the liquid metal out from the solid oxide component.
Potential customers for this machine have, however, expressed a concern
that impurities in the dross may be returned to the bath, and this might
necessitate more frequent bath changes.
The objective of
the present work is to conduct preliminary studies of the partitioning
of impurities in the solder-dross recovery process.
2.
Experimental
For the purposes of the study a solder recovery system (EVS) instrument
was brought to ITRI together with a batch of dross from a commercial bath.
Two batches of dross were processed and the recovered solder and residues
collected.
In order to minimise
errors due to inhomogeneity of the specimens, the materials were sampled
by drilling numerous cores randomly all over each specimen and combining
the swarf generated.
The fractions were
analysed using atomic absorption spectrometry, using a Varian Spectra
10 instrument. The quantity of oxide present in the dross was estimated
by remelting a known quantity under an excess of zinc chloride/ammonium
chloride flux, and stirring until all of the oxide had been dissolved,
before cleaning and reweighing.
Scanning electron
microscopy was used to study the dross and residues, using a Jeol 5400
instrument fitted with a Link Pentafet energy dispersive X-ray (EDX) microanalysis
system. X-ray photoelectron spectroscopy (XPS) measurements were made
using a Kratos XSAM 800 instrument.
3.
Results and Discussion
The dross contained approximately 5% oxide, with the remainder being metallic.
The Separation Process recovered approximately 50% (by weight) of the
material as solder, This would suggest that the residue would contain
ca. 10% oxide with remainder being metallic tin-lead. On this occasion
the dross was added cold. It is understood that higher recoveries can
be obtained by putting the dross in while still hot (presumably because
more complete remelting is achieved).
Preliminary examination
of the starting material detected low levels of copper and iron to be
present, and for the purposes of this study analyses for these two elements
were carried out on all of the specimens.
Table 1 shows the
results of the analyses. The quantity of copper present was not excessive,
only slightly above the maximum level that is specified by many solder
standards (e.g. BS 219 Grade KP Cu 0.08% max), although a number of manufacturers
market solders which contain much lower levels than this. Unfortunately
no specimen was available of the solder bath itself, and so it is not
possible to comment on how the purity of the dross relates to the solder
from which it was derived. It appears that little or no partitioning between
the various fractions had occurred.
XPS analysis of the
black powdery oxide material indicated that it was essentially tin oxide,
and there no evidence of any significant preferential oxidation of copper
and indeed, given that tin oxide is more thermodynamically stable than
those of lead or copper, none would be expected. At these concentrations
the copper would be in solid solution within the solder, with discrete
intermetallic crystals appearing only at higher concentrations. Hence
given that metallic solder is the major component of all three fractions
it is perhaps not too surprising that all three have similar compositions.
This situation might change if the copper concentration were to reach
0.3-0.4% at which point copper-tin intermetallic crystals could start
to appear (depending on bath temperature), and these might be preferentially
trapped in the residue.
Interestingly,
the iron impurities tended to report in the residues rather than in the
recovered solder. The reason for this discrepancy is not yet clear:
it may be due to the lower solubility of iron in solder (and hence that
it was present as discrete crystals of FeSn2 rather than in solution).
Alternatively it may have been present as some other form of particulate
material such as oxide (either due to oxidation in the bath or due to
contamination by rust particles). Earlier work at the Institute¹
indicated that reactive contaminants (e.g. zinc, aluminium etc) tended
to segregate preferentially into the dross. Given that they would be
present as solids (oxides) it seems likely that such materials would tend
to remain with the residues rather than the recovered solder. The
same would also be true of adventitious particulate contaminants, provided
they were insoluble in the bath. More noble metals on the other hand,
such as copper, gold or silver (which dissolve readily into the molten
solder), would tend to remain in solution and report in both fractions.
Under equilibrium
conditions the following relationship should hold:
Wa
= Wrj + Wrd
where Wa is the weight
or impurity added to the bath in unit time, Wrj is the weight of impurity
removed in the joints of product passing through the bath and Wrd the
weight removed in the dross. Wrj and Wrd will tend to increase with time
as the concentration of impurities builds up (assuming a constant volume
of solder is being removed in unit time). Wa will tend to decrease with
time, because as impurity concentrations build up the rate of dissolution
from the board will tend to decrease. Eventually, both sides of the above
equation will balance and equilibrium will be established (provided, of
course, that product throughput/dross removal rates etc. are kept constant).
The effect of the
solder recovery process would be to reduce Wrd and thus give rise to higher
equilibrium levels of contaminants. The key issue is whether equilibrium
will be reached at tolerable levels of impurity. If the answer to this
question is yes, then solder recovery offers an attractive way of minimising
costs. If the answer is to this no, then the next question is how long
can a bath be run before intervention is necessary, and will the reduction
in bath life nullify any cost advantage of the solder recovery system.
Unfortunately without information regarding the magnitude of the variables
under typical operating conditions, it is not possible to make predictions
on this matter.
References
1. |
A.M.
Stoneman et al, "Oxidation of Molten Solder Alloys Under Simulated
Wave-Soldering Conditions", ITRI Publication 547.
|
|
A.M.
Stoneman et al, "Oxidation and Drossing of Molten Solders: Effects
of Impurities", ITRI Publication 587. |
|
Copper
(weight %) |
Iron
(weight %) |
Dross |
0.10 |
0.012 |
Recovered
Solder - batch 1 (822g/53%) |
0.10 |
0.001 |
Residues
- batch 1 (727g/47%) |
0.10 |
0.051 |
Recovered
Solder - batch 2(1O78g/47%) |
0.09 |
0.008 |
Residues
- batch 2 (1225g/53%) |
0.10 |
0.183 |
|