Click Here to View Paper

Analyzing Lead-Free
Soldering Defects in
Wave Soldering
Using
Taguchi Methods

Presented at Published by IPC

Apex 2000 2215 Sanders Road
March 14-16, 2000 Northbrook, IL 60062-6135
Long Beach Convention Center Tel 847 790.5362
Long Beach, California Fax 847 509.9798
© IPC 2000 URL www.ipc.org
 Analyzing Lead-Free Soldering Defects in Wave Soldering Using Taguchi Methods

Gerjan Diepstraten
Vitronics Soltec
Oosterhout, NL

Introduction
Production soldering with lead-free alloys, driven by environmental, health, and legislative concerns, is in the electronics
industry’s near future. Process development is certainly ongoing. The European “IDEALS” consortium and others have
conducted investigative studies on Pb-free wave soldering, and have concluded that Pb-free soldering is technologically
possible, but a number of implementation issues must be addressed. For example, the introduction of VOC-free flux-
technology will significantly impact the Pb-free soldering process, and the soldering process in general will have to be
modified to accommodate the requisite higher soldering temperatures. Taguchi Methods and Statistic Process Control are an
effective way to approach and evaluate Pb-free process development in wave soldering, to determine the most essential
control parameters in order to achieve the optimum process setup for an individual application.

Taguchi Method Summarized In our experiment, three C1 factors were selected:

The recognized analysis approach known as the B = contact time;
Taguchi method integrates an innovative approach to C = preheat temperature; and
quality with traditional methods for the design of D = flux amount.
experiments, in which a series of interrelated techniques The solder temperature is a C2 factor due to the time it
are developed that in concert help minimize unwanted takes to increase/decrease the temperature.
variability, reduce production waste, and provide
greater customer satisfaction. 2. Noise factors:
Noise factors are the product of process variables that
The Taguchi approach for reducing variation in affect variation, but are impossible or cost-ineffective to
production is a two-step process: control. Examples of noise are changes in environment
1. Manufacture the product in the best manner most temperature, humidity, dust, etc. during the
of the time (less deviation from the target); and production/experiment.
2. Produce all products as identically as possible (less
variation between the products). For practical reasons, there was no “noise” element
factored into the experiment. The main goal of this
The Taguchi method employs experiments using a experiment was to estimate the contribution of the
specially constructed “orthogonal array” table to affect individual quality-influencing factors. Other
the design process, such that quality is built into a experiments need to be conducted to quantify their
product during its design stage. An orthorgonal array is response to noise in the process. Finally, a selection
an experimental design constructed to allow a had to be made as to which output characteristics we
mathematically independent asessment of the effect of wanted to measure. There were two criteria selected:
the different factors affecting the experiment. 1. The number of pins without bridges.
2. Qualification of through-hole fill.
Preparing the Taguchi Experiment
The preparation for a Taguchi experiment begins with a Experiment Layout And Design
“brainstorming” session, in which a team of participants In contrast to other methodologies, which use mostly
from different disciplines establish clear statements of one-factor-at-a-time studies to identify controllable
problems, objectives, desired output characteristics and parameters, this experiment uses an L9 orthogonal
method of measurement used to design the appropriate array. In only 9 experimental runs, 4 factors at 3 levels
experiment. Afterwards, all the process parameters are are investigated, as shown in Table 1. The layout of the
identified. Relevant factors affecting the outcome are array is shown in Tables 4 and 5.
defined thus: Table 1-Process Factors
Factors Level 1 Level 2 Level 3
1. Controllable factors: A Solder 250 260 275
temperature [°C]
C1 = factors that contribute significantly to the process B Contact time [sec] 1.8 3.0 4.2
and can be directly controlled; and C Preheat 90 110 130
C2 = factors for which we need to stop the process if we temperature
change them. Topside PCB [°C]
D Wet flux amount 355 474 639
[mg/dm²]

Session P-SM4 / 1- 1
Output Characteristics Materials Used in the Experiment
A proper experiment setup is required in order to obtain The Pb-free Alloy
the most reliable data. For example, the range of control The most commonly used Pb-free alloys for wave
parameters must be as extreme as practical in order to soldering are SnCu and SnAgCu. SnCu is one of the
make problems (in this case bridging and bad hole least expensive Pb-free alloys available. This alloy has,
penetration) apparent. To quantify the effect of however, a high melting point (227 ºC), and its
bridging, the number of soldered pins without bridges mechanical properties are poor compared to other Pb-
were counted. free alloys.

Each board had 200 pins, so the maximum score could SnAgCu is an improvement over the basic SnAg.
only be 200. SnAg3.8Cu0.7 forms joints of higher reliability and the
solderability is better than SnAg and SnCu. The
addition of antimony, Sb (0.25-0.5%) provides added
thermal resistance. The strength of the alloy increases
by the intermetallic structures of antimony with silver
and copper. However, there is some concern regarding
the toxicity of Sb, although toxic Sb-oxides are
generated at temperatures above 600 ºC.

SnAg3.8Cu0.7Sb0.25 (SACS), with a melting point of

217 °C, is used in our experiment (see Table 2).

Table 2 - Alloy used in the experiment

Property Unit “SACS”
Composition Mass-% SnAg3.8Cu0.8Sb0.25
Figure 1 - Bridging Melting point °C 217
The score for the component in Figure 1 was ‘6’, Density Kgm -3 7300 (solid/25 °C)
because 6 pins do not exhibit bridging, whereas the 7000 (liquid/250 °C)
others do. Coefficient of Ppm°C-1 (20-100 °C) 21
thermal expansion
Viscosity MPas (250 °C) 2.0
For the effects of through-hole penetration, each hole Surface tension Nm-1 (250 °C) 0.46 (N2)
filled with solder was marked as shown in Figure 2: Creep strength at High
room temperature
Price per mass ± 4 x SnPb

Board finish
The Organic Solderability Preservative (OSP), Entek™
Plus Cu-106A, a high performance copper protective
coating, that selectively protects and maintains
solderability of through-holes, was selected for this test
board. Earlier studies demonstrated that SnAgCu is
compatible with OSP finishes.

OSP is a replacement for hot air levelling (HASL) and

other metallic printed circuit board finishes. With
higher preheat settings, the thin organic coating
(thickness 0.2-0.5 microns) loses activity. The
compatibility of OSP with water-soluble fluxes is
excellent. Acids and solvents contained within the flux
Figure 2 - Hole completely filled: 2 Points (Left); quickly dissolve the thin OSP coating, which becomes
Sides of holes wetted with solder: 1 Point part of the flux, and volatilizes when the molten solder
(Right); Sides of holes NOT wetted with solder: 0 contacts the board.
points.
The Flux
A total of 4662 points were the maximum achievable The flux selected for this test was 396-RX synthetic
for each board. VOC-free with less then 2% solids (physical properties
shown in Table 3). This halide free, low residue flux
was selected due to its good solderability on copper
surfaces. 396-RX is also very effective in preventing
bridging.

Session P-SM4 / 1- 2
Preheat for this flux, measured in topside temperatures, Table 4 shows the data related to bridging obtained
may range from 100 °C to 112 °C, depending on the from the experiment.
configuration of the assembly to be soldered.
Table 4-Experimental Data: Bridging
Table 3-Physical properties flux Contol Factors Number good
soldered pins
Product type 396-RX
Group 1 A B C D Run 1 Run 2
Density at 20 °C 1.007 (± 0.5%) 1 1 1 1 1 198 196
Acid number (mg KOH/g) 15.5 (± 5.0%) 2 1 2 2 2 188 198
Color Colorless 3 1 3 3 3 171 161
Solid content < 2.0% w/w 4 2 1 2 3 200 200
5 2 2 3 1 169 192
Flash point None 6 2 3 1 2 192 196
DIN 8527 corrosion test Passes 7 3 1 3 2 196 192
Halides 0% 8 3 2 1 3 200 200
PH (5%) 5 9 3 3 2 1 178 172

Flux Application
Of the flux application technologies available, a nozzle-
spray fluxer is the best type to apply an appropriate
layer to the board. Flux droplet size can be controlled
with adjustment to air atomizing pressure control. With
VOC-free fluxes, it is important to achieve the finest
(smallest) droplets possible in order to achieve good Figure 3 – Response to Bridging
through-hole penetration and successful volatilization
of the water film. Therefore, a water-based flux should Figure 3 illustrates the incidence of bridging relative to
have carefully formulated surface properties, yielding a process factors. As mentioned earlier, the higher the
smooth interface with both the metallic and non- number, the better (200 = no bridging).
metallic surfaces.
Figure 3 shows that, with respect to bridging, contact
The nozzle fluxer allows the user to control the amount time and preheater temperature affected output data the
of flux very precisely, from approximately 300 mg/dm² most; i.e., changing one of these settings will have the
to 750 mg/dm² (wet flux). The maximum is 750 most dramatic effect on the number of bridges.
mg/dm², because beyond this point excess flux begins
dripping off the board. Optimal settings for bridging based on experiment data
were determined to be A2, B1, C1, and D2. Although
Test Board Design and Material the difference between A2 and A3 was very small, we
Test board dimensions were 160x100x1.6 (mm). selected A2 for two reasons:
Material was FR4 with copper double side plated 1. Solder temperature of 260 °C is preferred due to
through-holes. lower energy consumption requirements;
2. Solder temperature of 260 °C causes less thermal
The connectors have 10 pins, double row with a 0.2-µm shock to components and board material.
AuNi finish.

Test Results
For the test, we ran 18 boards (9 boards with one
repetition) to obtain the data needed for Orthogonal
Array analysis to achieve the following objectives:
♦ To estimate the contribution of individual quality
influencing factors.
♦ To gain the best, optimum condition for the Pb-free
process.
♦ To approximate the response of the control
parameters under optimum conditions.

ANalysis Of VAriance, a statistical treatment, is used

to analyse the results of the Orthogonal Array
experiment and to determine how much each quality Figure 4: Pie Chart Bridging
influencing factor has contributed.
Figure 4 shows the impact in % of each control
parameter with respect to bridging.
Session P-SM4 / 1- 3
Table 5 lists the experimental results related to through-
hole wetting.

Table 5- Experimental Data: Through-hole

penetration
Control Factors Number for quality
through-hole penetration
Group 1 A B C D Run 1 Run 2
1 1 1 1 1 4426 4464
2 1 2 2 2 4526 4539
3 1 3 3 3 3776 3792
4 2 1 2 3 4652 4651
5 2 2 3 1 3464 3061
6 2 3 1 2 4530 4538
7 3 1 3 2 4316 4505
8 3 2 1 3 4516 4593
9 3 3 2 1 4586 4578
Figure 7- Cross section of the bridge shown in
Figure 1.

Figure 5 – Response to through-hole penetration

In Figure 5, once again, the higher the count, the better

(4662 = a 100% good soldered board). Preheat
temperature (130 °C) affects the process the most, while
the impact of the other factors is about equal. The
experiment repetition error was found to be small for
through-hole penetration.

Optimal settings established for the best through-hole

penetration of solder based on experiment data were
A3, B1, C2, and D2. Figure 8- Cross section of a pin to illustrate that no
fillet lifting was found in the experiment.

Figure 6– Pie Chart Through-hole Penetration

Figure 6 shows the impact in % of each control

parameter related to through-hole penetration.
Figure 9- Intermetallic Ni-Sn layer between the pin
Microscopic Pictures of joints and the Pb-free solder.
Cross sections were made of the ‘SACS’ solder joints
of testboards from the experiment.

Session P-SM4 / 1- 4
Discussion: Results of the Experiment chemicals and materials into the process. This review
The impact of solder temperature is minor relative to its can be facilitated, and process development expedited,
affect on bridging. For through-hole penetration, a by conducting a Taguchi analysis and appropriately-
higher solder temperature is better. However, this designed experiment. This enables the process engineer
option may be limited due to potential damage to to reach a practical understanding of what will be
components, flux activators and board material. required in his own specific application with only a
small number of test runs. Although new Pb-free wave
Shorter contact times yielded better results for this soldering processes will have a smaller process window
experiment, possibly because the flux activator system due to higher temperatures and other materials, SPC can
was compromised at the higher preheat and solderpot be a valuable tool to help the engineer remain within
temperature settings. This is typical for this type of flux correct specifications and to achieve the minimum
under these process conditions. Other tests revealed that variation in the newly developed process.
longer contact times could be of benefit if the flux
activations system is strong enough to withstand the This test board used in this experiment was run on a
higher temperatures. Otherwise, contact times of 2.5 to standard (Vitronics Soltec) Delta 6622 wave soldering
4 seconds are recommended. machine equipped with a spray-nozzle fluxer, a three-
stage preheater, and nitrogen obtaining repeatable
In terms of preheater temperature, a setting of 110 ºC is results in this Pb-free process at a conveyor speed of 2
the champ for this process. At a higher preheater m/min without problems. This illustrates that special or
temperature (130 ºC) the process window narrows non-standard equipment or options will in most cases
remarkably; the OSP coating and the flux may lose not be required to convert to a Pb-free process. The
activity. Pre-heating water-based VOC-free flux spray-fluxer used is capable of delivering adequate flux
requires special consideration. Once the flux is applied, to the through-hole barrels. This is particularly
it is essential to improve the chemical bond between the encouraging, since foam fluxers in general will not
flux and the board surface. This can be achieved by work properly with most water-based VOC-free flux;
heating the flux. Consequently, at the end of the first special newly-formulated fluxes are required.
process zone (600 mm), the temperature of the top side
of the board should be approx. 70-80 ºC. For this test, References
medium-wave Calrod IR-generating elements were 1. P. Langeveld, Dr. D. Schwarzbach and E. de
selected. The elements provide the appropriate volume Kluizenaar, Lead-Free Wave Soldering. Feasibility
and wavelength of IR energy to initiate activity without Study, Philips Electronic Packaging & Joining,
forcing the water to boil out of the material at the start. CTR594-98-0051, 1998.
Forced convection heating is used in the 2nd and 3rd 2. P. Langeveld, Dr. D. Schwarzbach and E. de
zones to eliminate excess water before entering the Kluizenaar, Lead-Free Wave Soldering. á-
solder waves. verification, Philips Electronic Packaging &
Joining, CTR594-99-0012, 1999.
It is essential to have a continuous, uniform spray 3. B.P. Richards, C.L Levoguer & C.P. Hunt, K.
pattern of flux all across the board. The finest possible Nimmo, S. Peters and P. Cusack, An Analysis of
droplets applied with the lowest possible air pressure the Current Status of Lead-Free Soldering, NPL
setting will deliver the best result. Higher settings may and ITRI, 1998.
cause a droplet ‘bouncing’ effect, rather than improved 4. Marconi Materials Technology, Improved Design
wetting of the board surface. The D2 setting was the Life and Environmentally Aware Manufacturing of
paper champ in our experiment. electronics Assemblies by Lead-Free
Soldering:”IDEALS”, BRPR-CT96-0140, 1999.
The overall paper champ for this experiment is listed as 5. D. Suraski, A Study if Antimony in Solder, AIM.
follows: A3, B1, C2, and D2. 6. K. Hollevoet, Organic surface protection of copper
renders superior to HAL, Back to the soldering
A. Solder temperature = 275 ºC; to avoid thermal basics, Interflux Belgium, EPP 1999.
damage, between 265-270 ºC. 7. K. Wengenroth, OSPs: Guidelines for Successful
B. Contact time = 1.8 sec. Soldering, Enthone-OMI, Inc., West Haven, CT.
C. Preheat temperature (topside) PCB = 110 ºC. 8. http://kernow.curtin.edu.au/Taguch1
D. Wet flux volume = 474 mg/dm² 9. E. Westerlaken, H. Trip, Process Considerations
for the VOC-Free Concept, Cobar Europe BV,
Conclusion 1999.
For Pb-free wave soldering in production volumes to be 10. Dr. M. Warwick, Implementing Lead Free
implemented successfully, the entire process must be Soldering Consortium Research, Multicore Solders
reviewed, for it is not simply a matter of dropping new Limited, Hemel Hempstead, UK, SMTA 1999.

Session P-SM4 / 1- 5