Five Steps To Successful lead-Free Soldering: Step 2.

The first article in this series discussed how materials and machine configurations should be selected in preparation for the conversion to a lead-free soldering process. This article will examine the methods of determining what parameters have the most and least impacts on lead-free soldering. The intent is to establish a lead-free process whose quality and repeatability can be controlled.

Developing an Efficient Method

Now that lead-free soldering in the production line is imminent, an effective method must be developed to determine the correct process settings. Lead-free soldering is not merely replacing one alloy with another; no "drop-in" replacements exist. The introduction of a new material affects the entire process; therefore, all machine settings must be reviewed.

In reflow soldering, the goal is to meet or reproduce the correct profile for the solder paste and to stay within the specification for the components and board materials. The challenges are to accomplish this goal with machines that are currently in production and without decreasing throughput.

To realize this goal, the machines should have good heat transfer characteristics and uniformity (small temperature differences on the board). Most of today's hot air/nitrogen convection ovens are capable of soldering lead-free alloys. However, infrared-lamp ovens will most likely need to be replaced due to poor heating uniformity performance and high temperature differences on the board.

For the wave soldering process, switching to lead-free will also affect most machine parameters. For this process, the goal is to implement lead-free alloys in combination with volatile organic compound (VOC)-free water-based fluxes (while eliminating halogenated flame retardants) without decreasing productivity or thoughput.

A proper experiment must be designed to determine if the intended production machines can accommodate the goal of converting to lead-free soldering. Design of experiment (DOE), especially the Taguchi approach, offers a very effective method of ascertaining the equipment's capability. By learning and applying this technique, the time required for experimental investigations can be significantly reduced.

Designing an Effective Experiment

Taguchi experiments optimize the product/process design to make the performance minimally sensitive to the different causes of variation, without actually eliminating these causes, in the most economical way. The costs of research and development, manufacturing and operations are included. Taguchi experiments are based on orthogonal arrays, which reduce the number of experimental runs.

The design of a Taguchi experiment is very important because the quality of the results depends on a proper setup. This setup requires careful planning, prudent layout of the experiment and expert analysis of the output data. The experiment begins with a brainstorming session, which employees from different disciplines (design, operations, quality and manufacturing) are invited to attend. All individuals should have a firsthand knowledge of soldering. Every member has one vote in all selections that have to be made by this team. Therefore, an uneven number of team members is preferable.

The team's job is to formulate the problem. The goal is to realize lead-free soldering with the highest possible quality and best performance obtainable by determining the optimum combination of design factors.

The first step is to list the control factors or those parameters that will have a dominant effect on the solder quality or input that can be controlled. Examples of control factors for wave soldering include the amount of flux, preheat settings, conveyor speed and solder temperature. In reflow soldering, the control factors may include the use of nitrogen, conveyor speed and temperature settings of soak and peak-zone. Flux type and board finishes are examples of controlled input factors.

If interactions are present between some of these factors, they also should be listed. Each team member separately ranks the factors in order of importance toward affecting output quality.

Noise factors are those process or product factors that affect variation, but are either impossible or not cost-effective to control. Examples are board quality, ambient temperature and humidity. These factors can be integrated into an experiment, in the so-called outer array, when the robustness of a design must be quantified.

Now, the experiment layout must be selected. The Taguchi method uses orthogonal arrays, which are stringently defined matrixes that can be filled with simultaneously varying factors. Each level of each factor is tested an equal number of times against every level of each factor's levels. The orthogonal array and number of repetitions that will be selected for the experiment depend upon costs, time and materials available. A number of matrixes are available; for example, L4 (23), which stands for 4 (number of runs), 2 (number of levels) and 3 (number of factors); L8 (27), L9 (34), L12 (211), and several L18 variations.

The number of runs (determined by the orthogonal array) and the number of repetitions are now selected, and the levels of the varying factors must be defined. The team should be bold at this step, because a major goal in this type of experiment is to see variations. If quality differences are not seen, the team was not bold enough or, even worse, the control factors selected were not capable of affecting quality.

Quality Characteristics And Approach

The output characteristics (response data) allow the results of the experimental run to be quantified. These characteristics will indicate whether or not the product is soldered in conformance with the quality specifications or if the joint quality is poor. A Pareto analysis of the solder failures from the process can provide good input in the selection of the output characteristics. However, be aware that lead-free soldering will have some specific quality issues like fillet lifting, voids and solderballs. Because the lead-free solder temperature is closer to the melting point than it was with tin/lead, hole filling and reliability must also be quantified.

A Lead-Free Soldering Experiment

An actual experiment was run to demonstrate how a Taguchi analysis can be applied. For this experiment, the brainstorm team decided to do the soldering on a basic wave solder machine using an L8 array with three repetition runs (Figure 1). In the outer array, two fluxes were tested. This experiment resulted in a total of 48 runs: eight runs for the L8, three repetitions and two times for the outer array (Table 1).

A tin/silver/copper (SnAgCu) alloy was selected to solder. Two VOC-free water-based fluxes were used, and the preheat temperature was determined by the specifications of the fluxes. The wave soldering machine was equipped with a main wave with a "smart" wave. The smart wave has a hexagon shaft turning in the wave, thereby creating turbulence on the wave. The result is higher vertical forces of the solder to provide better hole filling.

An FR-4 board with a thickness of 1.6 mm was used. In total, 14 pin connectors were assembled (280 pins, which equals 280 potential bridges).

Output Characteristics

In this experiment, through-hole penetration and the bridging between the pins were studied. Because hole filling is more difficult with lead-free alloys (Figure 2), those variables that can help solder flow to the top of the holes should be quantified. Variables that may help this response factor are contact time, nitrogen, flux, board finish and solder temperature. The solder temperature was limited to 265 degrees C to prevent board warpage.

Earlier tests showed that, in the relationship among bridging, flux and preheat settings, preheat played a dominant role. Preheat settings that are too high can disturb the flux activators, thereby causing an absence of flux at the wave exit and allowing oxides to create bridges. To avoid this condition, the maximum temperatures on the topside of the boards, as specified by the flux suppliers, were not exceeded.

Analyze the Data

The number of poorly filled holes is listed in Table 2. The seventh run with flux A gave the best results; only four holes of more than 4,000 were poorly filled with SnAgCu solder.

The number of bridges is listed in Table 3. Several runs had no bridges, whereas run 4 with flux B showed 278 pins with bridges. The hole filling was evaluated with Anova software (Table 4); the rho column shows the impact in percent of the factors toward hole filling, with rho being the source contribution ratio (rho equals the source pure variation, S', divided by the standard deviation).

The flux amount had the most impact on hole filling. This result makes sense, because if no flux was in the hole, the solder did not flow into it. Another interesting result was that the impact of the smart wave was almost 20 percent. The solder temperature had no impact, probably due to the small difference (not bold enough) of 10 degrees C.

The Anova software was also used to evaluate the bridges (Table 5). The scores in the rho column show that all factors contributed to bridging more or less equally. As expected, nitrogen and the flux were the major factors. Bridges will appear once oxides are on the solder surface if no strong flux is present to prevent the bridging.

Figure 3 depicts the analysis of the through-hole penetrations; the smaller the number, the better. The best settings to get good filled soldered holes are:

- A1/A2-Solder temperature: no preference, select the most economical value

- B1-Nitrogen on

- C2-Contact time: 4.3 sec

- D1-Smart wave on

- E1/E2-Preheat temperature: no preference, select the most economical value

- F2-Flux amount: more flux

- G1-Board finish: organic solderability preservative (OSP)

- N1-Flux type A

Figure 4 depicts the bridging analysis; again, the smaller, the better. The best settings for no bridges occurring are:

- A2-Solder temperature: 265 degrees C

- B1-Nitrogen on

- C1-Contact time: 2.3 sec

- D1-Smartwave on

- E1-Preheat temperature: low

- F2-Flux amount: more flux

- G2-Board finish: nickel/gold (NiAu)

- N1-Flux type A

Conclusions

The experiment results revealed that 265 degrees C was the preferred solder temperature. The use of nitrogen for SnAgCu made sense because it reduced dross formation and also solder failures. The best results were found with longer contact times. Then, the hole filling was better and, unless not enough flux was on the board, no bridges occurred.

This experiment also proved that a smart wave contributed to better solder quality. Preheat temperature was of minor importance as long as the specification was followed. This result was an advantage because a larger temperature difference over the board did not significantly impact hole filling and bridging. A smaller amount of flux was also used, but the test showed that this approach did not work. The board finish can be discussed; from a cost viewpoint, OSP was best.

This data cannot be translated exactly to every wave soldering process. However, the data confirm theories regarding the use of nitrogen, smart wave and contact times, as well as other process-related issues.

A confirmation run can also be performed. This additional run uses the best parameter settings and will reveal the quality of the end result. The confirmation run should be compared with the software prediction to determine if the experiment was set up correctly. If the confirmation run does not meet the prediction, the causes of the disparities should be considered; the causes may be interactions or factors that are not part of the experiment.

Gerjan Diepstraten is a senior process engineer with Vitronics Soltec BV in The Netherlands; e-mail:[e-mail].

http://www.circuitsassembly.com

Copyright [copyright] 2001 Miller Freeman LLC