Overprinting the pad aids the assembly process, and could balance wetting forces during reflow.

Pb-free materials differ from Pb-rich materials in aspects such as self-centering capability and wetting characteristics. These differences beg more questions regarding aperture design and assembly defects. This article addresses those, and examines what happens when the process window is intentionally violated.

Process window. The experiment focuses on violating the print process window and measuring the response. This was achieved by including x and y offsets to the solder paste deposits. To do so, we manually programmed the automatic stencil printer with offset values, which moves the alignment of stencil to board by the prescribed amount; this value can be adjusted in 1 µm steps. To understand the influence aperture designs and offsets have on Pb-free assembly, we have included a stencil which has many aperture geometries.

A stencil was fabricated from a Ni-formed blank. (Nickel was chosen due to benefits established from earlier work.) Apertures were cut via a YAG laser. The stencil had a full range of aperture designs fabricated into the artwork (Table 1).

Click here to see all tables and figures (1.6MB PDF file).

The purpose of varying the apertures within the stencil was to determine if certain apertures would influence the quality of SMT production. In addition, we could determine if aperture geometry would increase the process window for Pb-free assembly.

Solder paste printing for the experiment was conducted using a 125 µm-thick nickel laser-cut stencil. A thinner 100-µm stencil would provide better solder paste release for smaller devices, but would inherently reduce the solder paste volume available for larger component types. A 150 µm-thick stencil would have unacceptable solder paste transfer for smaller apertures.

The apertures were neither microetched nor surface finished. The metal mask was mounted in a center-justified configuration. One stencil was designed for the experiment; the passive component artwork on this stencil was broken into sub designs, which provided a total of 67 designs. Breakdowns of these designs are shown in Figure 1 and Table 2, with the board split into sections A and B. Aperture designs are labeled A to F and aperture sizes X to Z to help tabulate the results.

Test board. The test vehicle (Figure 3) was a double-sided panel measuring 140 x 204 mm. Attachment pad metallurgy was bare copper covered by Entek+ (OSP). Passive components included 0201, 0402, 0603, 0805 and 1206. Pad widths, lengths and spacing are documented in Table 3.

Print platform. The study was conducted using a DEK Europa platform that was calibrated mechanically. Using the manufacturer's defined procedure, Cp and Cpk values were verified to pass the minimum 1.6 values. To reduce statistical noise, the same machine, interface and transfer heads were used throughout the experiment. The same batch of substrates was also used for measurement purposes.

These printer process parameters were used throughout the experiment:

• Print speed = 80 mm/s.

• Print head = 300 mm cassette.

• Paste pressure = 1.8 bar.

• Print gap = 0 mm.

• Separation speed = 10 mm/sec.

Both x and y print offsets were included, to give insight to the robustness of a Pb-free print process. (Table 4 lists the level of offsets for all the runs.)

Solder paste. The solder paste sample used is shown in Table 5. The material was suspended in a no-clean flux medium.

Reflow. The boards were reflowed using a standard 10-zone oven. The board was profiled to ensure a correct preheat, soak and peak setup. Figure 4 shows the reflow profile used; additional information about the oven setup is below:

• Oven model: Vitronics-Soltec XPM2

• No. of heating zones: 8

• No. of cooling zones: 2

• Convection fan speed: 3500 rpm

• Oxygen level when nitrogen active: <50 ppm

• Conveyer speed: 53 cm/min.

• Peak temp.: 240°C

Nitrogen was used for a selected number of runs (Table 4).

Results

A recap of the factors included within this experiment:

• 67 individual aperture designs for each passive component.

• Offset shift in both x and y.

• Atmosphere of air and nitrogen.

The first setup focused on the effects of a standard process. During this setup no adjustments such as offsets were included, and the reflow atmosphere was nitrogen-enriched. The results from this batch of experiments showed that with a centered print process the assembly was extremely robust. The next batch of results shows the assembly yield when x and y offsets are included.

Figure 5 shows that a shift of +ve 140 µm in the x axis with nitrogen does not cause a large number of defects (four). The photographs show the top left corner of the 0201 chips, this includes aperture designs A at 1:1 (x), standard spec (y) and overprint (z). Figure 6 illustrates results observed during this setup.

When the same +ve 140 µm offset is added to the y axis, 576 defects were observed. The majority of these defects were tombstones. Figure 7 illustrates results from this setup.

The next step in the experiment was to maintain the 140 µm y offset without nitrogen and permit the reflow process to run in air. The result was remarkable; defects were noticeably reduced to 279. Figure 8 shows these results.

The final batch of tests increased the offsets to 280 µm. Again the y offset shift was more responsive and gave a total of 1,505 defects. Figure 9 illustrates the results observed during this setup. This set of photographs shows the reasonably good reflow results, which were obtained from the bottom row of 0201 components (the aperture design which corresponds to this row was AZ). The other two rows were the same shape but different aperture size; the poorer response can be seen clearly.

The 280 µm offset in the x axis was not as responsive, but this test did result in 300 defects. Figure 10 illustrates the results of this setup.

As seen in Figure 5, print processes in which the offsets do not deviate could be guaranteed, "no change" to stencil architecture is required. But we know that FR-4 has a dimensional stability of no greater than 0.08%, which would in the case of this test vehicle have potential excess of 200 µm offset. Therefore, tests conducted at 140 µm will be further analyzed because these results reflect real life.

Figure 11 shows a breakdown of the defects by aperture design. We focus only on the runs in which the 140 µm offsets were included. It is interesting to observe the effects of 280 µm offset (but when running with this amount of alignment offset Pb-free implementation would be the least of your worries). Figure 11 also shows that adding nitrogen has a detrimental effect on the process.

The results, which used air within the reflow process, show superior capability, but certain aperture designs still show issues when 140 µm offsets are used. Aperture design AZ shows good capability and is interdependent of the reflow atmosphere (AZ, a standard rectangle and overprint). This would be the preferred design for all passive components, as it can cope with the "normal" offset shifts that occur in SMT assembly.

Conclusions

During this experiment several points have been discovered: One of the most important is that a centered Pb-free print process should not cause concern for process engineers. Ensure that the repeatability and accuracy of the deposit is under control. As past investigations show, enclosed print heads and nickel/laser cut stencils enable deposition stability.

But, of course, a utopian process does not exist within an assembly facility, therefore, these statements are true only when manufacturing within a laboratory facility. Therefore, we need to fully appreciate the influence that stencil geometries can have on a real-life Pb-free process.

Figure 12 shows the combinations of aperture designs and size. Some designs were extremely detrimental to the process, but design AZ (standard rectangle with 50 µm added per side) showed significant benefit for all passive components that exhibit standard assembly offset creep. It would seem that this slight increase of solder paste volume aids the assembly process, particularly once the solder paste placement accuracy is breached. This increased capability could be explained by the fact that the overprint not only gives a "larger slop" area of placement for both "paste on pad" and "component to pad" alignment, but the additional solder paste material balances the wetting forces during reflow. A further benefit to this aperture design was that the OSP pad was tinned, which is more a cosmetic feature but one that is ever present in the discussion of Pb-free assemblies.

The addition of nitrogen clearly increased the number of defects. This could be due to increasing the wetting forces and reduced wetting time of the solder during reflow. Longer thermal reflow profiles could reduce the number of assembly defects observed with nitrogen.

One important observation is the different response between resistor and capacitor packages. The capacitor package accounted for approximately 80% of the defects recorded. This defect rate was more noticeable on 0201 and 0402 capacitor packages once the solder paste was shifted in the y axis. These packages both have a high height-to-width ratio, which suggests that the balance of total downward forces pulling down on the component is only slightly more than the surface tension of the end faces; therefore when paste is misaligned, the end face, which is not fully contacted with the solder paste, does not have sufficient surface tension to equal the pulling force of the opposite side. When this happens the chip component flips up due to imbalance of forces. This is not seen on larger chip components because the mass of these is greater than the end surface force.

Final conclusions:

• Apertures that were rectangular in profile, and include an additional 50 µm on the east and west edge (assuming the chip component is mounted vertically), display a greater assembly yield rate.

• Using nitrogen within the reflow process is detrimental to the assembly process window.

• Using the correct aperture designs can deliver a high yield rate even when "paste to pad" or "component to pad" offsets creep into an assembly process.

 

Clive Ashmore is manager, global applied process engineering group, DEK (dek.com); cashmore@dek.com.

• Prev
• Next