Reducing
solder microballs in innert wave soldering
Articles / Newsletter
Reducing solder microballs in innert wave soldering.
Investigation of small solder balls produced during wave
soldering suggest a tightening of process controls and the
correct solder mask.
MCT
Process
During the past few years, research and development activities
have been focused on finding processes and materials that
eliminate the need to depend on chlorofluorocarbon (cfc) solvents
for cleaning. No-clean soldering has emerged as a viable process
option. However, before widespread acceptance was possible,
several major challenges had to be overcome. These included:
The reliability of a product with flux residue.
The impact of flux residue on testability.
A culture that traditionally equated a visually clean
product with a quality product.
These issues were addressed and satisfactory solutions were
found.
Then, the focus turned to the final challenge - solder microballs.
Solder microballs are common to all wave-soldering processes.
Close observation of a traditional RMA process with CFC cleaning
reveals that microballs exist prior to the cleaning process.
Cleaning typically removes most microballs, but the lack of
awareness during inspection misses the remainder.
Figure
1a.
Although originally
thought to be a random occurrence, the location of the
solder balls is very predictable.
When an assembly is exiting the wave area, solder begins
to pull away from the board.
However, solder will adhere to the metalized surfaces
of the board (Fig. 1a)
until the force of gravity overcomes the surface tension
of the solder (Fig. 1b).
Figure
1b.
Figure
1c.
When this occurs, a
snapping action takes place downward into the solder pot,
forcing a sphere, or microball, of solder to be propelled
toward the board (Fig. 1c and
1d.)
Figure
1d.
Therefore, if this
theory is correct, high-risk areas for solder microballs
should be in close proximity to metalized areas such as
through-hole leads and surface mount pads. Once a solder
ball has been thrust toward the board surface, it may,
under certain conditions, adhere to the surface of the
solder mask (Fig. 1e).
Figure
1e.
Nitrogen inerted wave soldering is becoming increasingly
popular since it provides a wider process window and reduces
dross. However, a nitrogen environment can give rise to an
increase in solder balls. In an atmosphere with under 50 ppm
of oxygen, the solder wave does not have even a thin layer
of dross and is, therefore, even more prone to tile "milk
drop" phenomenon. Additionally, due to the lack of dross to
act as a protective layer, solder flowing down the baffle
creates a splashing action that can send small particles of
molten solder everywhere. Fortunately, most manufacturers
have addressed this issue by shielding various areas in the
solder pot or improving wave flow.
Many theories have been put forth to describe why a microball
would adhere to the solder mask. One common conclusion of
most studies is that the surface texture of the solder mask
and the effect of various processes on the texture is a key
element. Other factors, such as mask cure, mask hardness and
chemical composition, have also been blamed. Even with these
factors identified, process experimentation is needed to fully
understand and eliminate solder microballs.
Patterning
the investigation
A multivariate experiment was designed to test the relationship
between various process parameters and solder ball occurrence.
Variables included were soldering mask type, wave configuration,
solder temperature, preheat temperature and flux quantity.
An inert wave soldering machine was used, which allowed both
the preheat and soldering sections to be maintained below
5 ppm oxygen. A flux with approximately 2 percent solids content
was used throughout the entire investigation. A tightly controlled
quantity of flux was applied with a spray fluxer.
A matrix
was constructed that would allow the variables and their
interrelationships to be observed for each mask type (Table
1). Wave configuration refers to the use of either
a single, laminar flow wave or a turbulent wave in con-junction
with the laminar wave.
Table 1.
Preheat temperature was measured on
the bottom side of the module by a pyrometer prior to entering
the solder wave.
Working closely with a printed circuit board supplier, a standard
test board was chosen. For each solder mask that was to be tested,
the supplier produced 40 boards, allowing five boards to be
used for each trial. A total of six different solder masks were
used in the evaluation: five were liquid photoimageable and
the sixth was a liquid photoimageable/dry film.
Data
and results
A multivariate regression analysis was performed on the data.
This is a technique similar to the Taguchi interpretation
method. Basically, the goal is to establish a correlation
between the variation designed into the experiment and the
variability found in the data.
In agreement with previously referenced experiments,
solder mask type did play a significant role in the
formation of solder balls (Fig.
2).
In fact, mask 6 demon-strated this dramatically.
In all trials, excessive amounts of solder webbing formed
on the soldered surface of every PCB coated with mask
6.
Due to the inability
to quantify the number of solder balls, mask 6 was dropped
from the experiment.
Figure 2.
Variation of solder ball formation based on the type
of solder mask.
Microballs were formed, as predicted by theory, in
relatively high quantities around DIP and connector
leads but were almost non-existent in open areas of
the board.
Figure 3.
An example of the concentration of solder balls in close
proximity to metalized areas.
Furthermore, each
mask type reacted differently, throughout the course of
testing. Some masks were highly sensitive to the controlled
variations in the experiment while others were fairly
insensitive (Fig. 4).
Figure
4. Some masks were highly sensitive to process variation;
others were insensitive.
The explanation of the observed
differences between solder masks has a great deal to do
with surface topography. If the adhesion force between
the solder particle and the PCB is greater than the gravitational
force, the microball will stick to the mask surface. The
adhesion force is related to the contact area between
the solder ball and the mask, and the gravitational force
is dependent upon the size of the solder ball.
One method of reducing the probability of a microball
attaching to the PCB surface is to roughen the surface
of the mask through chemical formulation, mechanical abrasion
or chemical etching. This, in effect, decreases the contact
area available between the microball and solder mask (Fig.
5).
Figure
5. Solder ball formation to rough (left) and smooth solder
masks.
Although
mask type does influence the number of solder balls formed,
the variable that had the strongest correlation was the
quantity of flux applied to the module (Fig.
6).
Figure 6.
While mask surface texture influences the formation
of solder balls, quantity of flux has the strongest
correlation.
The explanation for this is
straightforward. As discussed previously, microballs are formed
during the exit portion of the wave where the "snapping" action
of solder pulling away from metalized surfaces causes spheres
of solder to be propelled from the solder pot toward the board.
One of the inherent chemical properties of flux is to reduce
the surface tension of solder. By applying higher quantities
of flux, more will be available during the exit of the module
from the solder pot. Lower solder surface tension equates decreased
"snapping" forces and therefore, fewer solder balls.
Preheat and solder temperatures were also found to
be significant
(Fig. 7).
Since time and heat are involved with the volatilisation
of no-clean fluxes, preheat and solder temperatures
correlate with the amount of flux remaining on the PCB
during wave exit and therefore, the number of solder
balls created.
Figure 7.
Effect of preheat and solder temperature on solder ball
formation.
Conclusion
Based on the results of this investigation, several conclusions
can be drawn:
Given a specific set of process conditions, mask 1 will
produce the least amount of solder balls and mask 5 will
allow the widest process window of any of the masks tested.
None of the masks tested produced zero solder balls.
Therefore, a specification needs to be established based
upon quantity and size of the microballs.
Tight control of flux deposition is absolutely necessary
to minimise solder balls. Currently, spray fluxing is the
only known method of obtaining this level of control.
Despite the fact that more flux tends to reduce the quantity
of microballs, limits need to be defined that allow for
the minimum number of solder balls, acceptable solder yields
and high reliabilit
