June 1, 2006
Solder paste dispensing is not new. However, today's microelectronics require deposits smaller than 0.25 millimeter in diameter, and that has become a daunting challenge for suppliers of solder paste and the equipment to dispense it. Better paste formulations, more precise equipment and more tightly controlled processes are necessary to make such deposits reliably.
The challenges of microdot dispensing are easier to understand if put into perspective. A typical dispensing specification calls for deposits to be within ±10 percent (at 3s of the target diameter or mass. To get an idea of scale, consider that a 0.25-millimeter deposit is 40 percent the mass of just the 10 percent error value for a 0.5-millimeter deposit. That means a 0.5-millimeter deposit can vary by four times as much as a 0.25-millimeter deposit can and still meet the ±10 percent variation limit. Similarly, a 0.1-millimeter deposit is roughly 60 percent the mass of the 10 percent error value for a 0.25-millimeter deposit. That means control for a 0.1-millimeter deposit must be six times better than for a 0.25-millimeter deposit.
Another challenge involves the interaction between the solder paste and the auger valve. Each time the valve actuates, it applies force to the paste. This force causes a small amount of paste separation both within the auger channel and the dispense tip. As deposit size decreases, the number of force cycles that the paste is exposed to increases.
Among the dispense tips we have tested, the one that holds the least material, from the bottom of the auger to the end of the tip, contains roughly 1,600 solder paste deposits 0.5 millimeter in diameter. If the deposit size shrinks to 0.25 millimeter, the material within the tip must survive 6,400 cycles. Reduce the deposit size by just 0.05 millimeter, to 0.2 millimeter, and the number of cycles increases to 10,000. Go smaller still, to 0.1 millimeter in diameter, and the number of cycles inflates all the way to 40,000-and that is just for the material beyond the auger. It does not consider the material within the auger channel or the syringe.
Ordinarily, successful solder paste dispensing depends on controlling several variables: fluid pressure, auger pitch, auger rotation rate, tip design, and the gap between the tip and substrate. When the goal is deposits of 0.25 millimeter and smaller, some of these variables dominate the process, while others take on secondary roles. To assess these variables, we conducted a two-year study of microdot dispensing.
Study OverviewOur goal was to identify how much each variable influences the process and to determine how well each needs to be controlled to ensure a successful process.
In the first stage of the study, which was completed last year, we identified a solder paste formula that could survive the rigors of microdot dispensing. Using that solder paste as a vehicle, we then looked at each process variable to see which ones had the most impact on microdot dispensing. In the end, two variables mattered the most: temperature and the gap between tip and substrate (the Z gap).
In the second stage of the study, we evaluated these two variables individually. In this stage, several variables remained constant: valve type, auger design, tip design and solder paste. The valve elements and tip design were constants because they are functionally hard tooling with predictable variability. The paste was a constant since there were no other options; only one paste provided acceptable results for microdot dispensing.
The remaining variables were temperature; dot diameter vs. volume; valve control parameters; and the Z gap. In this report, we present the results of our tests for all these variables except valve control parameters.
When the Z gap was held constant, the standard deviation for deposit size was not a meaningful measure of success. That's because the range of deposits included both those that were larger and those that were smaller than optimal for the chosen Z gap. In all cases, data averages were used, since they accurately reflect material flow rate over a large population of deposits.
Temperature vs. Deposit SizeData was collected over a temperature range of 15 to 29 C in 2-degree increments. All other variables were kept equal. Test temperatures were maintained at ±0.1 C. For each temperature, we averaged a series of deposit diameter measurements and a series of deposit mass measurements.
With the exception of the deposit set made at 15 C, the first few deposits of each set had to be discarded. They were oversized as a side effect of paste flow under constant pressure. This was because the pressure setting was not adjusted down as paste viscosity dropped. In a production environment, this "oversized first dot" phenomenon would manifest itself as yield loss due to dot size variation. When using an auger valve to dispense solder paste, material viscosity must remain constant for a particular pressure setting to remain appropriate.
Our results showed that the average mass of the dots increased by 9.8 percent with each 1-degree increase in temperature. With the pressure constant, deposit size increased as temperature increased and paste viscosity decreased.
Diameter vs. VolumeIn our second experiment, we wanted to determine the relationship between the diameter of a deposit and its volume, with all other variables being equal. Data for dot diameter and mass were collected over a range of diameters with the Z gap held constant. (The mass of a dot directly correlates with its volume.) Temperature was maintained at ±0.1 C for each dot set.
Diameter and mass measurements were averaged over a range of deposit sizes. The relationship between deposit diameter and mass was the closest to linear with deposits ranging from 0.19 to 0.23 millimeter in diameter.
Looking at our results, the classic assumption that a deposit takes the form of a half sphere does not account for the true relationship between diameter and mass over the full range of diameters tested at the particular Z gap. The shape is closer to a spherical section with a small peak on top.
Z Gap vs. Process ControlFor our third experiment, the only variable we changed was the Z gap. We made two rows of dots at a particular diameter. The second row was done at a different Z gap than the first row, and each row had at least three dots. This experiment was done for five dot diameters, and a range of Z gap settings were tested.
The sensitivity of the process was tested at two levels. At one level, the threshold for process failure was defined as when a difference in Z gap caused the average diameter for all the paired dots in a set to differ by 5 percent. At the other level, the threshold was defined as when a difference in Z gap caused the average diameter to differ by 10 percent.
For each pair of rows at a given diameter, calculations were made to determine how large a difference in Z gap would cause the dots in the second row to vary by more than 5 percent or 10 percent from the dots in the first row. The Z gap was evaluated as both an absolute value and as a ratio relative to diameter.
Although the relationship between deposit diameter and volume is nonlinear, we found the relationship between Z gap and diameter to be remarkably linear at both the 5 percent and 10 percent levels.
Without a system for absolute Z gap control, such as hard stops and dedicated tooling, the challenge of substrate and tip location becomes much more difficult when producing deposits between 0.1 and 0.25 millimeter in diameter.
To put this in perspective, consider this: Based on the 5 percent linear relationship formula and a 0.25 millimeter deposit, the maximum allowable limit for Z gap variation is 0.0133 millimeter. Expressed as a tolerance, this is ±0.00667 millimeter or 6.67 microns. At a diameter of 0.2 millimeter, this tolerance drops to ±5 microns. At 0.1 millimeter, the tolerance is 1.71 microns.
At the 10 percent tolerance level, the maximum allowable limit for Z gap variation does not change much. For 0.25-, 0.2- and 0.1-millimeter deposits, the Z gap tolerance levels are ±8 microns, ±5.95 microns and ±1.74 microns, respectively.
RecommendationsBased on our experiments, we draw three significant conclusions.
First, if the goal is to maintain deposit size to within 10 percent of the target size at ±3s a variation in temperature as small as ±0.25 C can throw a seemingly well-controlled microdot dispensing process out of tolerance. A 1-degree difference in temperature will cause deposit mass to vary by 9 percent to 10 percent. That's with no variation in Z gap and does not take into account inherent variation in the process due to equipment issues.
Within the data sets generated at each temperature, deposit diameter variation at 3s was between 7.1 percent and 9.9 percent, as long as the Z gap was appropriate for the dot size. Our data show that a repeatable process can be achieved at a particular temperature with a particular set of equipment and materials. Of course, not all assemblers are currently able to control temperature as well we did in this study.
Second, 2D visual inspection cannot capture the true variation in a dispensing process, due to the nonlinear relationship between deposit mass and deposit diameter. If the quality of a solder joint depends on the volume of alloy in the joint, one of two things is required. Either assemblers must characterize their dispensing processes the same way that we did in this study, or they must use a 3D inspection system with sufficient resolution to detect the difference between acceptable and unacceptable deposits.
Finally, to make consistent solder paste deposits 0.1 millimeter in diameter, the height of the dispense tip above the board must be controlled to within 1 to 2 microns of the ideal setting. New positioning technology will be required to achieve this goal.
For more information on microdot dispensing, call 800-338-4353, visit www.efdsolder.com or Reply 20.