Why edge clamps provide more uniform paste transfer across the board.

For stencil printing, defects are typically caused by one or more of the following: poor alignment between substrate and stencil, incorrect paste chemistry, or variations in the amount of paste deposited. Variation in the amount of paste deposited in turn depends on several factors, some of which are paste chemistry, printer setup, squeegee blade type, stencil design, board support and board clamping.

Board support is an integral part of developing a robust printing process. Proper board support is essential to ensure consistent print results and higher yields. Without proper board support, the force applied to the board across the entire width of the PCB will vary, and proper gasketing between the stencil and board will not be achieved. Board support comes in two distinct forms: under board support and transport rail support. Here, we look at two common transport rail board support systems: “top clamp” and “edge clamp.” As components are placed closer to the edge, board clamping has become a significant factor in maintaining print quality.

The primary function of a clamping system is to hold the board tightly in place to provide optimum gasketing during printing process. Available various types of clamping systems include top clamp, snuggers, flippers and vacuum hold-down. Top clamp and snuggers, two primary clamping systems, operate slightly differently in providing the mechanism to hold the board. Top clamp, as the name implies, holds the board in place by applying a clamping system (a thin metal foil) on the top of the board, while edge clamp works by tightly snugging the board in the Y direction, without foil on top of the board. As the edge clamp holds the board without use of foil, it delivers optimal stencil-to-board gasketing, critical in providing higher paste release efficiency. 

Recently, a carefully designed experiment was conducted to understand the true effect of these two clamping systems on a specially designed stencil. The stencil was designed to baseline the clamping system using a simple design concept. The stencil consisted of 0.004˝ laser cut, stainless steel foil with an image size of 8˝ x 10˝. The aperture pattern started 3 mm from the edge of the board and increased by 0.5 mm every four pads (Figure 1).

An external SPI was used to characterize the paste deposit to determine the variation in paste volume and height. Figure 2 shows the box plot of transfer efficiency (TE) for the top half of the board. The top clamp shows a significantly higher TE close to the edge. Moving away from the edge, the TE stabilized to a more constant value. No such variation in TE is seen for the edge clamp system. The primary reason for the higher TE near the board edge is due to higher paste height (Figure 3). Why is paste height higher near the board edge for a top clamp system? As the squeegee blade moves over the stencil from the edge of the image to the center, the top clamp acts as a standoff height. This action prevents the stencil from providing the necessary gasketing for consistent paste transfer. As we move from the edge, the “standoff” effect reduces and proper gasketing occurs between the board and stencil. Since the edge clamp does not provide a standoff effect, paste transfer seems to be uniform across the board.

Results from this test and data from various other sources indicate edge clamp provides a more stable process as compared to top clamp. When dealing with a highly populated board or boards with components close to the edge, edge clamps have a distinct advantage. 

Rita Mohanty, Ph.D., is director advanced development at Speedline Technologies (speedlinetech.com); rmohanty@speedlinetech.com.

Print quality can be improved through better control of solder volume and consistent stencil wiping.

Print speed, squeegee print pressure and solder paste type typically are not changed during a working process. These parameters are not changed because doing so would cause the process to become unstable and unpredictable. Yet one facet of the solder printing process that constantly changes is print pressure. Print pressure is the amount of force the solder paste exerts during the filling of the aperture.

Our measurements indicate that a change in the solder paste volume on the stencil is one of the largest contributors to dynamic print pressure. (Paste volume can be stated as the diameter of the roll, or the weight of the total amount of paste on the stencil.)

In most processes, a certain number of boards are printed, and then either solder paste dispensed in a bead or an operator manually applies paste onto the stencil. Either way, the amount of the paste on the stencil and therefore, print pressure, can change dramatically through the course of a shift of production.

Dispensing small amounts of solder paste frequently in the center of the existing paste roll permits the print pressure to remain virtually stable. This, in turn, removes a significant variable in the print process.

It has long been suspected that solder print definition and volume are affected by factors not readily changeable during the print process. Two of these factors are blade angle and solder paste volume on the stencil.

Blade angle. Squeegee blade angle is the angle formed between the stencil and the squeegee at a zero pressure condition. This has generally been a function of the fixed angle of the squeegee holder, or the print head itself. Some earlier stencil printer designs had the ability, within a narrow range, to mechanically change this angle during setup. This limited the operator to a single blade angle between changeovers. As a result, and because the blade angle was difficult to change, it typically was set at one angle and never changed. Printer OEMs experimented with offering blades of different angles, but this never caught on in practice. With few exceptions, each printer OEM settled on a single blade angle for its entire range of printers.

Solder volume on the stencil. The solder paste volume on the stencil has an impact on overall solder paste print quality. For example, excessive paste will get caught between the blades and impede the natural rolling action of the paste, resulting in insufficient paste depositions. Insufficient paste causes the material to slide across the stencil, instead of rolling. Efforts to minimize the potential for excessive or insufficient paste have focused mainly on the frequency and amount of solder paste dispensed via an automatic dispenser. For purposes of time, dispense cycles typically are programmed to be rather infrequent. The net effect is only to prevent either extreme: too much or too little solder paste on the stencil.

Enclosed print chambers have largely minimized these effects. However, these print heads have not been adopted as industry standard due to considerations such as cost, maintenance, ease of cleaning, and substrate size variations. Additionally, some older solder paste formulations, as well as all water-soluble pastes, are not suited for this process.

Previous Experimentation

Previously, the effects of blade angle and squeegee pressure have been explored as they relate to the effects on fine feature printing. In a study performed last year, we determined that:

1. Greater solder paste transfer efficiency can be obtained for the same aperture size (area ratio) by reducing the blade contact angle.
2. Increasing print pressure may decrease the blade angle, but has a negative effect on transfer efficiency.
3. If blade angle is optimized, a lower area ratio can be used to print 01005s and µBGAs, allowing the opportunity to use a thicker stencil to print these fine feature devices.

Test Methodology

Testing was performed at Yamaha’s Hamamatsu, Japan, laboratories. The goal was to determine the solder paste fill pressure changes when modifying the following parameters:

1. Solder paste volume (100 – 500 g).
2. Blade angle (45° – 65°).
3. Squeegee pressure (50 – 100N).
4. Squeegee speed (30 – 90 mm/sec).
Testing was done on a Yamaha YGP Printer with 350 mm long stainless steel blades. Blade angle was modified via the use of a servo-controlled angle adjustment motor integrated into the machine’s print head. The material used was Alpha Metals OM-325MS SAC 305 solder paste.

For these experiments, a stainless steel stencil was used. A pressure transducer was mounted into the stencil to measure the downward paste pressure into the stencil. This was used to approximate the fill pressure of the paste into an aperture during normal printing operations. One parameter was changed at a time in order to determine the individual effects on pressure.

Experiment Results

Solder paste volume. The paste volume experiment used:
1. A blade angle of 55°.
2. A squeegee speed of 30 mm/sec.
3. A squeegee pressure of 50N.
4. Paste volume was varied between 100 – 500 g.

Pressure was measured throughout the print stroke and graphed. Figure 1 shows several things:

1. As solder volume increases, the transducer feels the pressure of the paste over a longer period of time. This is expected due to the large diameter of the paste roll: The paste is over the transducer for a longer time.
2. As solder initially contacts the paste sensor (Figure 2, Point A), there is a small spike in pressure. Because the solder is rolling forward, this is expected, due to the downward force of the paste at the leading edge of the roll.
3. After the small spike, there is a drop in pressure as the weight of the paste is transfered to the transducer (Figure 2, Point B).
4. Near the point of contact of the squeegee with the stencil, pressure is highest (Figure 2, Point C). This is due to the shearing action of the paste at the blade tip. As the blade begins to pass the aperture, there is no place for the paste to go but up the squeegee face. The filled stencil aperture provides the reaction force necessary to push the paste up the face of the blade.
5. It is important to note that the applied force on the squeegee blade is not the sole reason for the pressure increase. If this were the case, there would not be a gradual buildup of the pressure to the peak.

Rather, there would be a spike in pressure equal and opposite to the pressure drop at the end of the stroke. Further, if the pressure spike was a result of the squeegee force, then all the peak pressures would occur at the same value at the point of contact of the squeegee with the sensor.
These effects were common in all testing performed. In addition, significant improvement was observed in the paste pressure with a large diameter solder roll (Table 1).

In a majority of operations, operators scoop solder paste onto the stencil without precisely measuring the amount applied. A few companies specify the amount of solder paste to be placed on the stencil at the beginning of a production run (i.e., one full 350 g jar), but most do not. Yet even these former companies fall short of full control of the volume of solder paste on the stencil when setting up the paste dispenser. Most paste dispense processes are suited to a larger volume of paste (75 – 150 g), and dispensed at infrequent intervals (every 40 – 50 prints). As a result, the roll size can still change significantly during the course of a shift.

As seen from the above data, a 100 g change in solder roll size will cause approximately a 7% change in maximum paste filling pressure. This is a significant change. It is enough to dramatically alter the solder paste print quality, especially on small apertures.

Attack angle. The attack angle experiment used:

1. A 300 g roll of solder paste.
2. Squeegee speed of 30 mm/sec.
3. Squeegee pressure of 50 N.
4. Varied angle from 45 – 65°.

Pressure was measured throughout the print stroke and graphed in Figure 3. Similar characteristics are diplayed in this experiment. The lower squeegee angles flatten the solder paste roll. As a result, the same volume of solder paste contacts a larger area of the stencil (Figure 4). This causes the pressure to be felt over a longer period of the stencil stroke. 

Additionally, the lower angles result in a much larger maximum pressure. This makes sense, as the smaller volume at the tip of the blade should cause a larger force, as paste is pushed upward along the face of the squeegee blade. The same amount of solder needs to travel through this smaller volume in the same amount of time, resulting in a higher paste flow rate for a lower angle. Table 2 shows the actual values. 

Squeegee pressure. The squeegee pressure experiment used:

1. A 300 g roll of solder paste.
2. Squeegee speed of 30 mm/sec.
3. An angle of 55°.
4. Varied pressure between 50 and 100 N.

Again, paste pressure was measured throughout the print stroke and graphed (Figure 5). The results of the two pressure conditions in this experiment proved to be almost identical. Table 3 shows the actual values. This is an interesting result, as it previously has been thought a larger amount of squeegee force will bend the blade more, causing a higher filling force, as seen in Experiment 2. From this result, we can deduce that the angle changes that result from an increased print pressure are very small. If we were to calculate a hypothetical angle based on the pressure changes:

The value is 54.38°, a slight 0.62° change in angle for double the squeegee force.

It is standard practice to use only the minimum blade pressure necessary to wipe clean the surface of the stencil. This has been used in the past as an attempt to prolong the life of metal blades by minimizing their wear. Considering these data, the additional squeegee force has no appreciable effect on paste pressure. This eliminates any remaining reason to increase blade pressure beyond the minimum required to clear the stencil surface.

Squeegee speed. The squeegee speed experiment used:

1. A 300 g roll of solder paste.
2. Squeegee angle of 55°.
3. Pressure of 50 N.
4. Varied speed from 30 – 90 mm/sec.

Paste pressure was measured throughout the print stroke and graphed (Figure 6). The results are as expected:

As speed increases, the time over which the transducer sees the pressure is reduced. This is because the paste travels over the sensor much more quickly at faster speeds.

As speed increases, the maximum pressure increases, which is seen in the actual values (Table 4). 

The results make sense, as it takes more force on the stencil to push the paste up the face of the blade at a higher speed to maintain the rolling action of the solder paste. The dramatic peak that forms at higher print speeds better illustrates what happens during high-speed printing. Typically, higher speed printing requires the solder paste to be “worked” or printed repeatedly to thin the paste down to a viscosity that permits it to roll.
A thicker paste will need a higher force to get it moving back up the face of the blade from the tip. The higher the pressure, the more likely the tip of blade will be pushed up by the solder paste, causing a loss of contact with the stencil. The result is seen as the paste hydroplaning, or streaking behind the blade after it passes over it.

Efforts to eliminate this problem in the past have centered on increasing the squeegee pressure. This prevents the blade from hydroplaning (up to a point, until the blade is bent too far from excessive pressure). However, Experiment 3 shows this does very little to increase the paste filling pressure. The result is that, even with the higher paste pressure, there is too little time that the pressure is applied to the aperture, resulting in insufficient filling.

Discussion. As the data show, the parameters with the greatest effect on paste pressure are (in descending order):

1. Squeegee speed.
2. Blade angle.
3. Solder paste amount.
4. Blade pressure.

As a result, it would seem a faster squeegee speed would result in the best aperture filling during the print process. However, experience shows it has the opposite effect: The potential for insufficient aperture fill, especially for larger apertures increases with increasing speeds. It makes more sense to look at the aperture filling potential as the total area below the curve of pressure versus time, or

In a printing operation, it still takes a finite amount of time to fill each aperture. Aperture filling can be improved by increasing the time spent over it or increasing the pressure. In this case, the parameters that have the greatest effects on filling potential are:

1. Blade angle.
2. Solder paste amount.
3. Squeegee speed.
4. Blade pressure.

Practical Applications

In a printing process, concerns vary from the potential for insufficient filling of the apertures (caused by low pressure and a small amount of time over the aperture) to the potential for bridging due to excessive pressure or time over the aperture. How can aperture filling be optimized for common printing processes?

1. Optimize blade angle for squeegee speed. In a significant number of applications, print speed is dictated by a cycle time requirement. As a result, there needs to be a way to improve the time that the paste spends over the hole. Reducing the blade angle will flatten the roll, increasing the contact area of the paste. This not only increases the maximum filling force, it also increases the amount of time the paste is over the hole.

From our studies, it appears increasing squeegee force has little, if any, consequence on the effective squeegee angle. Specifically, when the applied force of the squeegee is doubled and resultant paste pressure measured, there was only a very slight increase in the overall maximum pressure reading obtained, correlating to a theoretical angle change of less than 1°.

To significantly adjust the blade angle with traditional solder paste printers, the options are:

A mechanical modification to the squeegee head.
Employing a different angle squeegee blade holder.
Neither of these options provides much process flexibility on the production floor. Further, they significantly increase the potential for setup errors. Unless the machine is equipped with a means to change squeegee angle as a process parameter, there is no way to effect this change in the machine program.

2. Increase the size of the solder paste roll for higher squeegee speeds. For printers that lack the ability to change the blade angle, a secondary way to improve the pressure and time over the hole in a high-speed printer is to optimize the amount of solder paste on the stencil. A larger solder paste roll will have a higher applied pressure, as well as a longer time that the pressure is applied. Too large a roll size for a given speed applies too much pressure to the aperture for too long a time, increasing the potential for bridging of fine pitch devices.

So, how could the paste roll size on the stencil be optimized? As mentioned, there has been a quasi-control of this by focusing on the frequency and amount of solder paste dispensed via an automatic dispenser. However, it takes a significant amount of time to perform this dispense. As a result, paste dispense is typically infrequent, with possibly too much material dispensed each time.

True solder paste volume control can be accomplished in one of two ways: via an enclosed print head, or to dispense more accurately and more often. For an enclosed print head, the paste roll size is constant: It is the size of the paste chamber. The problem with this method is there is no way to improve the time spent over an aperture without changing the print speed. In a short cycle time application, this might pose a problem. The ability to precisely dispense a small amount of solder paste after every print or after a relatively small number of prints will keep the solder paste roll volume/diameter constant. Most printers lack this option due to the long time required to dispense or the relative inaccuracy of the dispenser. (Note: The printer used for these experiments had the ability to dispense every print cycle without an impact on cycle time.)

During many recent print test experiments, there has been a focus on starting every trial with a fixed amount of solder paste on the stencil. In this way, the results of the testing are much more consistent than what is normally found in production. This same logic should be applied to the production floor.

As the data show, there is a significant difference in paste filling pressure when solder volume on the stencil is varied. Since we know the printing process consumes solder paste from the stencil at a fixed rate (for a given board / stencil combination), we can calculate the usage over time. For instance, if 2 g of paste were used per print stroke, the operator could add 150 – 200 g of solder paste every 75 boards, or the dispenser could put down 75 – 100 g every 40 boards. In both cases, the amount is approximate, and over the course of a shift, the errors in the volume added to the stencil can add up over time.

The errors in an operator adding solder paste by hand are readily evident. Different operators will place different amounts of solder paste on the stencil. Further, the same operator will not place the same amount down every time. The errors in paste dispensing are also readily evident. Determining the exact amount of solder paste to be dispensed is not typically done when a board file is programmed. Instead, a single dispense profile is created, and then the same settings are applied to all the other programs. In these cases, volume errors multiply, but now they do it at a fixed rate.

Paste dispensers themselves are not very accurate. Most are reciprocating bulk dispensers that apply air pressure to the back of the stopper in the cartridge. The time it takes to pressurize the cartridge varies significantly through its usage, depending on how much paste remains in the tube. As a result, the actual amount of solder paste dispensed varies significantly from the first to the last dispense cycle. Finally, when dispensing a large amount of solder paste with little paste remaining in the tube, the full amount required may not be dispensed, adding further error to the total amount of solder left on the stencil.

Better-designed paste dispensers place paste in the center of the roll and dispense paste after only a few boards. The ability to adjust the required pressurization time depending on the percentage of material left in the tube minimizes the difference in the amount dispensed from the first to the last dispense in the tube. Because the dispenser does not move across the length of the paste roll, there is no cycle time penalty to dispense as often as necessary. Dispensing in the center of the roll also minimizes the tendency for paste to spread beyond the blade ends. This further minimizes the variation of the amount of material actually printed.

The ability to cut the paste at the tube opening after dispense ensures that all of the paste that leaves the tube drops onto the paste roll. An automatic shutter covers the tube outlet, ensuring no additional paste drops out of the tube in between dispense cycles.

3. Minimize the paste sticking to the blades. One other factor in the effective solder paste roll size is the amount of paste that sticks to the blade. On a two-blade system, the more paste that sticks to the first blade, the less paste that is available for the return print. As a result, it is not uncommon to see adequate print quality on one squeegee direction, but not in the other. Much of this effect is due to the solder paste characteristics and the blade finish. Some solder pastes tend to stick more to the blades at colder temperatures. Additionally, some squeegee blade finishes are more prone to material sticking.

The ideal process has a single squeegee blade that flips over the solder paste roll in between prints, keeping the paste on only one side of the blade. This minimizes the effect of paste sticking to the blades because the paste is available on the same side of the blade for the return print. Using a squeegee with a durable, low-friction surface finish minimizes this effect further.

Conclusions

A robust screen-printing process requires control of as many sources of variation as possible. Not all variables are easy to control, but it is helpful to understand (and target) the largest sources of variation.

The testing performed revealed:
1. Blade angle affects the overall time the paste spends over the aperture, as well as the maximum filling pressure. Both characteristics have a significant effect on the resultant aperture fill.
2. The volume of solder paste on the stencil has a similar, but slightly smaller effect on aperture fill.
3. In operations where a change in blade angle is not possible, the volume should be tightly controlled to control the overall aperture fill. This is most easily done with more frequent, more precise dispense cycles.
4. Increasing speed has the most dramatic effect on print pressure. Since these effects are felt on the aperture for a much shorter amount of time, the net result is that the potential for insufficient aperture fill (especially for large apertures) increases with increasing speeds.
5. The minimum squeegee pressure necessary to completely wipe clean the surface of the stencil should be used all the time, as squeegee force has little, if any, effect on the overall filling force. 

Acknowledgments

I would like to thank Kouichi Sumioka of Yamaha Motor Corp. for the stencil design used in this application, as well as for setting up the experiment and collecting the data. The results of his work were instrumental in his design of the more precise, stationary solder paste dispenser described in this article.

References
1. George Babka, Scott Zerkle, Frank Andres, et. al., “A New Angle on Printing,” Global SMT & Packaging, February 2009.

George Babka is multinational account manager at Assembléon America Inc. (assembleon.com); george.babka@philips.com.

After years of study, tin whiskers continue to fascinate, perplex and astound.

As the compliance date for eliminating lead approached, component manufacturers began to implement Pb-free surface finishes for device leads. Numerous manufacturers selected pure tin.^1,2,3,4 This change sparked concern since pure-Sn plating has a well-known reliability problem: the potential to spontaneously grow tin whiskers.

Tin whiskers are electrically conductive crystalline structures that grow from pure tin-plated surfaces. Commonly growing as hair-like filaments, they may also take other forms such as “odd-shaped eruptions” and “nodules.” The main reliability concern associated with tin whiskers is their potential to cause transient or catastrophic electrical short circuits. Tin whiskers may also break free from their Sn-plate growth surface and find their way into, and interfere with, the operation of mechanical assemblies.^4,5

To better understand tin whisker growth, evaluate associated reliability risks, and develop mitigation techniques, Raytheon, along with industry, government, and academia colleagues, performed several experiments. This article shares observations and scanning electron microscope (SEM) images of tin whiskers captured during some of these investigations and experiments.

Tin whiskers remain a reliability concern, especially for space and military applications. This issue is especially a concern for equipment with long life requirements, since tin whisker initiation can occur soon after plating or can lie dormant for years before initiating.^2,4,5

There are numerous documented cases of tin whiskers causing equipment failure. The NASA Goddard Tin Whisker website lists many examples of satellites, military, medical, and industrial/power equipment that failed due to tin whiskers.⁵ At Raytheon, prior to implementing strict controls for pure Sn-plate usage, tin whiskers caused the failure of a rocket motor initiator by shorting across the device’s 0.010˝ spark gap.

Tin whiskers are single crystal structures whose growth mechanism(s) are not completely understood. They are reported to grow as long as 10 mm, but commonly grow to lengths less than 1 mm.⁵ Raytheon has documented multiple examples of whiskers growing greater than 1 mm in length, such as the 2.7 mm long whisker found growing on a connector shell shown in Figure 3.
 
Along their length, filament whiskers vary from being straight, kinked, curved or spiraled (Figures 1 to 7). Figure 4 also shows whiskers in the same general growth area can have widely varying filament thicknesses. Figure 5 provides an example of a kinked filament tin whisker having multiple bends along its length.

For one tin whisker study, Raytheon participated with an industry working group to evaluate the ability of different types of conformal coatings to mitigate tin whisker growth.6 For this investigation, bright Sn-plated brass substrate coupons (1 x 3˝ Sn-plated brass substrate strips) were selected due to their propensity to grow whiskers. As shown in Figures 1 and 8, this methodology was very successful in producing significant quantities of tin whiskers for experimentation and study. It was not uncommon for thousands of tin whiskers to grow on each Sn-plated coupon.

Figure 8 shows the results of the experiment in which silicone conformal coat was used to partially cover a Sn-plated brass coupon. The image shows that silicone conformal coating significantly reduced tin whisker growth in coated areas. This experiment, however, also concluded conformal coating did not prevent all whisker growth. Figure 9 shows a tin whisker penetrating through the conformal coat. Other whiskers were additionally observed to have punctured through the silicone conformal coat, especially in areas of thin coating. This experiment helped establish that conformal coating can be used to help mitigate tin whisker risk, but cannot be completely relied on to eliminate all risk.

Besides filament whiskers, tin whiskers can take on many unusual shapes. We label these as odd-shaped eruptions. Figures 10 to 12 provide SEM images of this category of tin whiskers.

During our experimentation, several tin whiskers were observed with unusual features that are difficult to describe and are best documented by providing images. The tin whisker featured in Figure 13 has a thick, straight base with multiple contorted appendages emanating from this base. Additionally, as highlighted in Figure 14, a portion of this unusual whisker splits off around a kink and then rejoins with the main whisker body. This splitting characteristic is rare, but not unique, as it was observed on other tin whiskers. Figure 14 also provides a good perspective on the relatively large size of tin whiskers as compared to the small size of the plating grains.

The unusual tin whisker featured in Figure 15 (growing among several straight filament whiskers) is equally difficult to describe. This whisker has several contorted appendages and a base comprised of two separate sections that merge together at the tip. Also making this tin whisker highly interesting is that it appears to have a straight filament whisker growing from its surface.

Figure 16 provides a SEM image from the perspective of looking down from the top of a tin whisker along its length to the Sn-plated surface. In this image, a hole in the Sn-plating surface is seen where the tin whisker emerges from the tin plate. During the months and years of our experimentation, as tin whisker growth progressed, we observed an increasing number of Sn-plating voids. These plating voids are areas of tin depletion that occur as the tin whiskers grow. In many instances, the voids were not located close to tin whiskers, implying some tin whisker growth occurs due to long range transport of tin. Figure 16 also provides excellent detail of striations that frequently run the length of filament tin whiskers.

Figure 17 is a high magnification SEM image of the tip of a tin whisker. This image, which is not typical of all tin whisker tips, provides significant detail of this tin whisker’s surface. In this image, the whisker tip appears to be comprised of multiple small strands. In contrast, Figure 18 shows a Focused Ion Beam (FIB) cross-section of a filament tin whisker with solid construction throughout.

From published literature, it is universally agreed tin whiskers grow from the addition of tin to their base and not from addition of tin to their tips.⁷ NASA has used time-lapse photography to document this growth.⁵ Interestingly, however, branching tin whiskers have been observed. Figures 19 and 20 show different magnification images of a tin whisker that has significant branching. It is not known how tin whiskers are able to branch with growth occurring from their base.

Another observation from experimentation was that there were noticeable differences in tin whiskers based on the substrate metal used. Figure 21 shows tin whiskers grown from a Sn-plated coupon with an Alloy 42 substrate. Typically, these whiskers were much shorter in length than those grown with brass substrates and grew from mound-shaped structures or “nodules.”

One tin whisker growth theory is that whiskers sprout through weak areas in the surface oxide layer. In Figure 22, tin whiskers are observed emerging from a cracked surface region. The whisker on the right side of the image has lifted off and still has a portion of the surface layer attached to its tip. In Figure 23, numerous tin whiskers sprout along surface scratches. Theories as to why tin whiskers are prone to grow along surface scratches include weaknesses in the surface oxide and the presence of increased stresses along the scratch.

FIB cross-sectioning was also used to investigate tin whiskers. We sectioned several tin whiskers and then cut rectangular trenches in the tin plating beneath them. We were surprised to find that after FIB trenches were cut, new tin whisker growths emerged horizontally (as shown in Figures 24 and 25) out of the FIB trench walls near the location where the original tin whisker had sprouted vertically from the surface.

The study of tin whiskers has been fascinating. During the months and years of investigating and experimenting with tin whiskers, we were continually surprised, perplexed and even sometimes astounded by their complexity, appearance and behavior. Each time a SEM examination was performed, we thought we had seen all that there was to observe about tin whiskers, only to find during our subsequent examinations, there was more to be discovered. 

Acknowledgments

The authors would like to recognize those who contributed to this research effort. Bill Rollins has been a driving force at Raytheon and in industry in the area of tin whisker research and whisker mitigation. Tom Woodrow (Boeing) provided the first Sn-plated brass and Alloy 42 test coupons in which initial tin whisker growth studies were conducted. Joe Colangelo (Raytheon) was the lead investigator on the rocket motor initiator failure analysis efforts. John Wolfgong, Ph.D., (Raytheon) was lead investigator and co-author of “Surface Oxidation as a Tin Whisker Growth Mechanism.” Bill Shieldes (Raytheon) provided support and funding to conduct tin whisker studies. Phuc Dinh Ngo (Microtech Analytical Labs, Inc.) provided FIB cross-sectioning services.

References
1. M. Warwick, “Implementing Lead Free Soldering – European Consortium Research,” Journal of SMT, vol. 12, no. 4, October 1999.
2. CALCE-EPSC Lead Free Forum, Global Transition to Pb-free/Green Electronics, 2004.
3. C. Xu, Y. Zhang, C. Fan and J. Abys, “Understanding Whisker Phenomenon: The Driving Force for Whisker Formation,”
CircuiTree, April 2002.
4. M. Mosterman, “Mitigation Strategies for Tin Whiskers,” CALCE-EPSC, August 2002.
5. NASA Goddard Tin Whisker Website, nepp.nasa.gov/whisker/.
6. T. Woodrow, B. Rollins, P. Nalley, B. Ogden, Tin Whisker Mitigation Study: Phase 1: Evaluation of Environments for Growing Tin Whiskers, draft report, released August 1, 2003, calce.umd.edu/lead-free/tin-whiskers/restricted/Phase1Draft2.pdf.
7. K.Subramanian, “Lead-free Electronic Solders: A Special Issue of the Journal of Materials,” page 359.

Robert R. Ogden is senior reliability engineer at Raytheon Space and Airborne Systems; r-ogden1@raytheon.com. Robert F. Champaign is principal failure analysis engineer at Raytheon Failure Analysis Lab, NCS Division.

Why operators skip important features – and what to do about it.

Automated vision inspection is among the top features of most advanced stencil printers. Other top features and process-related techniques include underside cleaning with vacuum, automatic placement of support pins and auto paste dispensing.

In some PCB assembly camps, vision inspection and other major stencil printer features, as well as setup process techniques, are not used. Reasons for not using them vary. For example, in some instances a stencil printer has vision inspection, but an operator simply ignores it. Other times, it’s not put to use due to poor operator training. A third reason assemblers avoid vision inspection as an option is its high cost.

Several problems can occur if a vision inspection isn’t deployed, especially on highly dense and complex PCBs. Among them: solder bridging, insufficient solder, missing solder, and print misregistration (Figure 1).

Underside cleaning. Underside cleaning is another key feature often overlooked. When a stencil printer machine is programmed to check stencil apertures, it also can be programmed to determine how often it performs that inspection by the number of cycles. Once that program is activated, it looks into the stencil apertures and sees from the initial program that it appears to be a clean aperture, meaning no clogging. But when the camera is activated, it sees stencil clogging. Depending on the percentage of the clogging programmed into the machine, the machine cleans the stencil once it sees the clogging (Figure 2).

An underside cleaning complemented by vacuum, if so equipped, further enhances the cleaning. A vacuum-equipped printer not only unclogs the apertures, but also removes smeared solder paste on the underside of the stencil screen while the automated cleaning operation is underway. Without a vacuum, the particles of some solder pastes that the wiper doesn’t completely clean will move around the apertures to other areas of the stencil underside. And when the next print is performed, the particles will be transferred to the PCB. This is particularly important on PCBs with gold fingers at the edges where solder is not wanted. Hence, those stencils are completely cleaned so that the next print is an extremely fine definition print.

If automatic stencil underside cleaning is not used, the operator is likely to continue printing boards, unaware the apertures are plugged and paste isn’t being deposited. This assembly issue is critical for boards with BGA devices. Once these components are installed, their joints are no longer visible to the human eye. Troubleshooting using x-ray must be used to determine the lack of paste on some locations of a BGA pad. Consequently, that BGA device or devices must be removed for rework. When they’re removed, the device, the PCB and other components on the board are subjected to another heat cycle. Each time boards are reheated, their lifetimes are reduced and reliability problems are induced.

Avoiding or eliminating solder bridges or misalignment also can be programmed into a stencil printer machine. During programming, the machine initially looks at the pads without the paste and then does a comparison after paste deposit. It looks at the differences or grey scale of the pads. The printer then determines if paste deposit is off or right on, and the machine will stop, especially when the more easily detected problem is the solder bridge or excessive solder.

The vision system recognizes the problem; the machine stops, and the operator is alerted. The operator has two options: manual cleaning or a program fine-tuning, doing the necessary adjustment to align the apertures with the pads. This way, boards are built using the right amount of solder, not too much and not too little. When a board lacks the correct amount of solder and is reflowed, it will be too late and costly rework required.

Figure 3 shows the placement of support pins underneath the circuit board is not a feature, but rather a critical aspect of the setup process. Support pins prevent board flexing when the squeegee runs over the top-side. Without the support pins, the result is uneven weight distribution, leaving uneven paste deposits at different PCB locations. This is particularly true when the PCB is populated with fine-pitch QFPs. During printing, the board flexes and the stencil flexes with it, thereby causing solder bridging and misregistration.

Newer machines have automatic solder pin placement, but are rarely used due to inadequate operator training. In some cases, operators resort to manually placing support pins. The problem here is the support pins are often not strategically placed at the proper locations, thus incurring likely printing issues. Figure 3b shows no support pins being used.

Stencil ordering can pose yet another setup process issue with a high probability of adversely affecting delivery time. In this instance, some assemblers order stencils in a foil with no frame as a way to cut costs (Figure 4). The issue here is the inordinate time required to mount the foil into an adapter, especially if an inexperienced operator is involved.

Time-consuming mistakes include mounting the foil upside down or in a reverse manner, as well as damaging it. Additionally, when stencil foils are not stretched properly when mounted on the adapter, it can adversely affect the quality of the paste deposit, causing bridging and misregistration. To avoid this problem, order stencil foils mounted in a frame (Figure 4). (This will incur a moderately higher cost). As a result, there is zero setup time. An operator takes the stencil, ensures it is the right one, inserts it into the machine, uploads the program, and is done.

Not putting into practice. Many assemblers fail to put into practice the available important stencil printer features and tools. Sometimes, due to pressures and demands in shipping schedules, theses features are bypassed, and the front-end process crew is not made aware of it. Then, rework is rushed at the back-end. The product is then subjected to unnecessary heat cycles or rework that can affect long-term reliability, and add delays and labor costs.

Assembly engineers or manufacturing managers can specify the need for a particular feature, fully knowing its strengths and results to reduce the amount of non-value rework into the product. However, in some cases, that feature is overlooked, ignored or improperly used during stencil printing.
Schedule changes may be the cause, triggering this non-use. An operator or supervisor may think that bypassing an already programmed feature can permit a quicker build. If that’s the case, the operator will disable the feature to meet shipment deadlines. More times than not, those PCBs incur failures at the end of the assembly line. Personnel at the beginning of the SMT line aren’t aware of what happens at the end of the line. Rework may be taking place at the end of the line, such as in the testing process, unbeknownst to personnel at the front-end. This is especially true if communication is also a problem on the manufacturing floor.

In the worst cases, by the time front-end SMT line personnel are made aware of the problem, they are already building a different product. Therefore, it becomes difficult to troubleshoot the problem and determine its root cause. They have already built the PCB and are rushing to ship it, but can’t because they are inundated with rework.

Getting full use. Personnel training is the first order of business to fully utilize stencil printer features. Well-organized and disciplined training, mandated from top EMS provider or CM management spearheads efficient use of an advanced stencil printer.

Written procedures or processes documentation augment this training. This document is called “manufacturing process instruction,” and it outlines the program name of a particular board and the stencil printer machine’s setup.

The QC inspector has a checklist of the parameters that are preset on the machine. He or she checks this list against the settings of the machine to make sure all those are turned on. For example, if there is a question or discrepancy there, then the quality person calls it to the attention of the process engineer to resolve the issue. If a deviation emerges, the process engineer has to approve it in writing with a time limit.

The manufacturing process instruction document also clearly spells out features that must be used. For example, procedures covering a stencil printer with a vision system specify the critical PCB locations the vision system is to inspect.

If boards are thin or large, support pin locations must be defined to ensure strategic placement. Those locations must be balanced to minimize board flexing, and if possible, no flexing at all.

The next step is to ensure all these features are used to do a first article of the process. This includes quality control (QC) personnel going through the line, comprehensively reading the documentation, and carefully inspecting the setup of the stencil printer, according to the documentation. All these steps ensure every aspect specified in the documentation is actually being applied. The QC person should not allow the line to run unless all the requirements per the written instructions are met per the formal document.

Once all these procedures are in place and the line is running, the last step is monitoring. Here, the QC technician and SMT line personnel monitor the printer to make sure no errors are being reported. If there are, notify engineering.

If the top features of the machine are fully deployed, personnel do not have to monitor it as frequently as when these features are not used. The stencil printer machine running with its full set of features will be the one to actually provide notification if something goes wrong.

By not fully exploiting these features, operators and SMT line personnel are forced to manually and visually inspect every single board coming out of the stencil printer machine. It is well established that repetitive tasks like this lead to operator fatigue and missed defects. An efficiently programmed stencil printer, complete with its full set of features and setup process techniques, doesn’t overlook these problems. It catches them, alerts the operator, and a problem with the potential of extending engineering costs is quickly resolved. 

Alex del Rosario is manufacturing manager at NexLogic Technologies Inc. (nexlogic.com); info@nexlogic.com.